diff --git a/cts/cli/regression.error_codes.exp b/cts/cli/regression.error_codes.exp
index 0b9eba43b7..0e0244df9f 100644
--- a/cts/cli/regression.error_codes.exp
+++ b/cts/cli/regression.error_codes.exp
@@ -1,564 +1,568 @@
=#=#=#= Begin test: Get legacy return code =#=#=#=
Error
=#=#=#= End test: Get legacy return code - OK (0) =#=#=#=
* Passed: crm_error - Get legacy return code
=#=#=#= Begin test: Get legacy return code (XML) =#=#=#=
=#=#=#= End test: Get legacy return code (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get legacy return code (XML)
=#=#=#= Begin test: Get legacy return code (with name) =#=#=#=
pcmk_err_generic - Error
=#=#=#= End test: Get legacy return code (with name) - OK (0) =#=#=#=
* Passed: crm_error - Get legacy return code (with name)
=#=#=#= Begin test: Get legacy return code (with name) (XML) =#=#=#=
=#=#=#= End test: Get legacy return code (with name) (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get legacy return code (with name) (XML)
=#=#=#= Begin test: Get multiple legacy return codes =#=#=#=
Error
Operation requires quorum
=#=#=#= End test: Get multiple legacy return codes - OK (0) =#=#=#=
* Passed: crm_error - Get multiple legacy return codes
=#=#=#= Begin test: Get multiple legacy return codes (XML) =#=#=#=
=#=#=#= End test: Get multiple legacy return codes (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get multiple legacy return codes (XML)
=#=#=#= Begin test: Get multiple legacy return codes (with names) =#=#=#=
pcmk_err_generic - Error
pcmk_err_no_quorum - Operation requires quorum
=#=#=#= End test: Get multiple legacy return codes (with names) - OK (0) =#=#=#=
* Passed: crm_error - Get multiple legacy return codes (with names)
=#=#=#= Begin test: Get multiple legacy return codes (with names) (XML) =#=#=#=
=#=#=#= End test: Get multiple legacy return codes (with names) (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get multiple legacy return codes (with names) (XML)
=#=#=#= Begin test: List legacy return codes (spot check) =#=#=#=
201: Error
202: Operation requires quorum
203: Update does not conform to the configured schema
204: Schema transform failed
205: Update was older than existing configuration
206: Application of update diff failed
207: Application of update diff failed, requesting full refresh
208: On-disk configuration was manually modified
209: Could not archive previous configuration
=#=#=#= End test: List legacy return codes (spot check) - OK (0) =#=#=#=
* Passed: crm_error - List legacy return codes (spot check)
=#=#=#= Begin test: List legacy return codes (spot check) (XML) =#=#=#=
=#=#=#= End test: List legacy return codes (spot check) (XML) - OK (0) =#=#=#=
* Passed: crm_error - List legacy return codes (spot check) (XML)
=#=#=#= Begin test: List legacy return codes (spot check) (with names) =#=#=#=
201: pcmk_err_generic Error
202: pcmk_err_no_quorum Operation requires quorum
203: pcmk_err_schema_validation Update does not conform to the configured schema
204: pcmk_err_transform_failed Schema transform failed
205: pcmk_err_old_data Update was older than existing configuration
206: pcmk_err_diff_failed Application of update diff failed
207: pcmk_err_diff_resync Application of update diff failed, requesting full refresh
208: pcmk_err_cib_modified On-disk configuration was manually modified
209: pcmk_err_cib_backup Could not archive previous configuration
=#=#=#= End test: List legacy return codes (spot check) (with names) - OK (0) =#=#=#=
* Passed: crm_error - List legacy return codes (spot check) (with names)
=#=#=#= Begin test: List legacy return codes (spot check) (with names) (XML) =#=#=#=
=#=#=#= End test: List legacy return codes (spot check) (with names) (XML) - OK (0) =#=#=#=
* Passed: crm_error - List legacy return codes (spot check) (with names) (XML)
=#=#=#= Begin test: Get unknown Pacemaker return code =#=#=#=
Error
=#=#=#= End test: Get unknown Pacemaker return code - OK (0) =#=#=#=
* Passed: crm_error - Get unknown Pacemaker return code
=#=#=#= Begin test: Get unknown Pacemaker return code (XML) =#=#=#=
=#=#=#= End test: Get unknown Pacemaker return code (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get unknown Pacemaker return code (XML)
=#=#=#= Begin test: Get unknown Pacemaker return code (with name) =#=#=#=
Unknown - Error
=#=#=#= End test: Get unknown Pacemaker return code (with name) - OK (0) =#=#=#=
* Passed: crm_error - Get unknown Pacemaker return code (with name)
=#=#=#= Begin test: Get unknown Pacemaker return code (with name) (XML) =#=#=#=
=#=#=#= End test: Get unknown Pacemaker return code (with name) (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get unknown Pacemaker return code (with name) (XML)
=#=#=#= Begin test: Get negative Pacemaker return code =#=#=#=
Node not found
=#=#=#= End test: Get negative Pacemaker return code - OK (0) =#=#=#=
* Passed: crm_error - Get negative Pacemaker return code
=#=#=#= Begin test: Get negative Pacemaker return code (XML) =#=#=#=
=#=#=#= End test: Get negative Pacemaker return code (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get negative Pacemaker return code (XML)
=#=#=#= Begin test: Get negative Pacemaker return code (with name) =#=#=#=
pcmk_rc_node_unknown - Node not found
=#=#=#= End test: Get negative Pacemaker return code (with name) - OK (0) =#=#=#=
* Passed: crm_error - Get negative Pacemaker return code (with name)
=#=#=#= Begin test: Get negative Pacemaker return code (with name) (XML) =#=#=#=
=#=#=#= End test: Get negative Pacemaker return code (with name) (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get negative Pacemaker return code (with name) (XML)
=#=#=#= Begin test: List Pacemaker return codes (non-positive) =#=#=#=
+-1041: More IPC message fragments to send
-1040: DC is not yet elected
-1039: Compression/decompression error
-1038: Nameserver resolution error
-1037: No active transaction found
-1036: Bad XML patch format
-1035: Bad input value provided
-1034: Disabled
-1033: Two or more XML elements have the same ID
-1032: Unable to parse CIB XML
-1031: Cluster simulation produced invalid transition
-1030: Error writing graph file
-1029: Error writing dot(1) file
-1028: Value too small to be stored in data type
-1027: Input file not available
-1026: Output message produced no output
-1025: Result occurs after given range
-1024: Result occurs within given range
-1023: Result occurs before given range
-1022: Result undetermined
-1021: Not applicable under current conditions
-1020: IPC server process is active but not accepting connections
-1019: IPC server is unresponsive
-1018: IPC server is blocked by unauthorized process
-1017: Operation requires quorum
-1016: Update does not conform to the configured schema
-1015: Schema is already the latest available
-1014: Schema transform failed
-1013: Update was older than existing configuration
-1012: Application of update diff failed
-1011: Application of update diff failed, requesting full refresh
-1010: On-disk configuration was manually modified
-1009: Could not archive previous configuration
-1008: Could not save new configuration to disk
-1007: Could not parse on-disk configuration
-1006: Resource active on multiple nodes
-1005: Node not found
-1004: Already in requested state
-1003: Bad name/value pair given
-1002: Unknown output format
-1001: Error
0: OK
=#=#=#= End test: List Pacemaker return codes (non-positive) - OK (0) =#=#=#=
* Passed: crm_error - List Pacemaker return codes (non-positive)
=#=#=#= Begin test: List Pacemaker return codes (non-positive) (XML) =#=#=#=
+
=#=#=#= End test: List Pacemaker return codes (non-positive) (XML) - OK (0) =#=#=#=
* Passed: crm_error - List Pacemaker return codes (non-positive) (XML)
=#=#=#= Begin test: List Pacemaker return codes (non-positive) (with names) =#=#=#=
+-1041: pcmk_rc_ipc_more More IPC message fragments to send
-1040: pcmk_rc_no_dc DC is not yet elected
-1039: pcmk_rc_compression Compression/decompression error
-1038: pcmk_rc_ns_resolution Nameserver resolution error
-1037: pcmk_rc_no_transaction No active transaction found
-1036: pcmk_rc_bad_xml_patch Bad XML patch format
-1035: pcmk_rc_bad_input Bad input value provided
-1034: pcmk_rc_disabled Disabled
-1033: pcmk_rc_duplicate_id Two or more XML elements have the same ID
-1032: pcmk_rc_unpack_error Unable to parse CIB XML
-1031: pcmk_rc_invalid_transition Cluster simulation produced invalid transition
-1030: pcmk_rc_graph_error Error writing graph file
-1029: pcmk_rc_dot_error Error writing dot(1) file
-1028: pcmk_rc_underflow Value too small to be stored in data type
-1027: pcmk_rc_no_input Input file not available
-1026: pcmk_rc_no_output Output message produced no output
-1025: pcmk_rc_after_range Result occurs after given range
-1024: pcmk_rc_within_range Result occurs within given range
-1023: pcmk_rc_before_range Result occurs before given range
-1022: pcmk_rc_undetermined Result undetermined
-1021: pcmk_rc_op_unsatisfied Not applicable under current conditions
-1020: pcmk_rc_ipc_pid_only IPC server process is active but not accepting connections
-1019: pcmk_rc_ipc_unresponsive IPC server is unresponsive
-1018: pcmk_rc_ipc_unauthorized IPC server is blocked by unauthorized process
-1017: pcmk_rc_no_quorum Operation requires quorum
-1016: pcmk_rc_schema_validation Update does not conform to the configured schema
-1015: pcmk_rc_schema_unchanged Schema is already the latest available
-1014: pcmk_rc_transform_failed Schema transform failed
-1013: pcmk_rc_old_data Update was older than existing configuration
-1012: pcmk_rc_diff_failed Application of update diff failed
-1011: pcmk_rc_diff_resync Application of update diff failed, requesting full refresh
-1010: pcmk_rc_cib_modified On-disk configuration was manually modified
-1009: pcmk_rc_cib_backup Could not archive previous configuration
-1008: pcmk_rc_cib_save Could not save new configuration to disk
-1007: pcmk_rc_cib_corrupt Could not parse on-disk configuration
-1006: pcmk_rc_multiple Resource active on multiple nodes
-1005: pcmk_rc_node_unknown Node not found
-1004: pcmk_rc_already Already in requested state
-1003: pcmk_rc_bad_nvpair Bad name/value pair given
-1002: pcmk_rc_unknown_format Unknown output format
-1001: pcmk_rc_error Error
0: pcmk_rc_ok OK
=#=#=#= End test: List Pacemaker return codes (non-positive) (with names) - OK (0) =#=#=#=
* Passed: crm_error - List Pacemaker return codes (non-positive) (with names)
=#=#=#= Begin test: List Pacemaker return codes (non-positive) (with names) (XML) =#=#=#=
+
=#=#=#= End test: List Pacemaker return codes (non-positive) (with names) (XML) - OK (0) =#=#=#=
* Passed: crm_error - List Pacemaker return codes (non-positive) (with names) (XML)
=#=#=#= Begin test: Get unknown crm_exit_t exit code =#=#=#=
Unknown exit status
=#=#=#= End test: Get unknown crm_exit_t exit code - OK (0) =#=#=#=
* Passed: crm_error - Get unknown crm_exit_t exit code
=#=#=#= Begin test: Get unknown crm_exit_t exit code (XML) =#=#=#=
=#=#=#= End test: Get unknown crm_exit_t exit code (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get unknown crm_exit_t exit code (XML)
=#=#=#= Begin test: Get unknown crm_exit_t exit code (with name) =#=#=#=
CRM_EX_UNKNOWN - Unknown exit status
=#=#=#= End test: Get unknown crm_exit_t exit code (with name) - OK (0) =#=#=#=
* Passed: crm_error - Get unknown crm_exit_t exit code (with name)
=#=#=#= Begin test: Get unknown crm_exit_t exit code (with name) (XML) =#=#=#=
=#=#=#= End test: Get unknown crm_exit_t exit code (with name) (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get unknown crm_exit_t exit code (with name) (XML)
=#=#=#= Begin test: Get crm_exit_t exit code =#=#=#=
Error occurred
=#=#=#= End test: Get crm_exit_t exit code - OK (0) =#=#=#=
* Passed: crm_error - Get crm_exit_t exit code
=#=#=#= Begin test: Get crm_exit_t exit code (XML) =#=#=#=
=#=#=#= End test: Get crm_exit_t exit code (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get crm_exit_t exit code (XML)
=#=#=#= Begin test: Get crm_exit_t exit code (with name) =#=#=#=
CRM_EX_ERROR - Error occurred
=#=#=#= End test: Get crm_exit_t exit code (with name) - OK (0) =#=#=#=
* Passed: crm_error - Get crm_exit_t exit code (with name)
=#=#=#= Begin test: Get crm_exit_t exit code (with name) (XML) =#=#=#=
=#=#=#= End test: Get crm_exit_t exit code (with name) (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get crm_exit_t exit code (with name) (XML)
=#=#=#= Begin test: Get all crm_exit_t exit codes =#=#=#=
0: OK
1: Error occurred
2: Invalid parameter
3: Unimplemented
4: Insufficient privileges
5: Not installed
6: Not configured
7: Not running
8: Promoted
9: Failed in promoted role
64: Incorrect usage
65: Invalid data given
66: Input file not available
67: User does not exist
68: Host does not exist
69: Necessary service unavailable
70: Internal software bug
71: Operating system error occurred
72: System file not available
73: Cannot create output file
74: I/O error occurred
75: Temporary failure, try again
76: Protocol violated
77: Insufficient privileges
78: Invalid configuration
100: Fatal error occurred, will not respawn
101: System panic required
102: Not connected
103: Update was older than existing configuration
104: Digest mismatch
105: No such object
106: Quorum required
107: Operation not safe
108: Requested item already exists
109: Multiple items match request
110: Requested item has expired
111: Requested item is not yet in effect
112: Could not determine status
113: Not applicable under current conditions
114: DC is not yet elected
124: Timeout occurred
190: Service is active but might fail soon
191: Service is promoted but might fail soon
193: No exit status available
=#=#=#= End test: Get all crm_exit_t exit codes - OK (0) =#=#=#=
* Passed: crm_error - Get all crm_exit_t exit codes
=#=#=#= Begin test: Get all crm_exit_t exit codes (XML) =#=#=#=
=#=#=#= End test: Get all crm_exit_t exit codes (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get all crm_exit_t exit codes (XML)
=#=#=#= Begin test: Get all crm_exit_t exit codes (with name) =#=#=#=
0: CRM_EX_OK OK
1: CRM_EX_ERROR Error occurred
2: CRM_EX_INVALID_PARAM Invalid parameter
3: CRM_EX_UNIMPLEMENT_FEATURE Unimplemented
4: CRM_EX_INSUFFICIENT_PRIV Insufficient privileges
5: CRM_EX_NOT_INSTALLED Not installed
6: CRM_EX_NOT_CONFIGURED Not configured
7: CRM_EX_NOT_RUNNING Not running
8: CRM_EX_PROMOTED Promoted
9: CRM_EX_FAILED_PROMOTED Failed in promoted role
64: CRM_EX_USAGE Incorrect usage
65: CRM_EX_DATAERR Invalid data given
66: CRM_EX_NOINPUT Input file not available
67: CRM_EX_NOUSER User does not exist
68: CRM_EX_NOHOST Host does not exist
69: CRM_EX_UNAVAILABLE Necessary service unavailable
70: CRM_EX_SOFTWARE Internal software bug
71: CRM_EX_OSERR Operating system error occurred
72: CRM_EX_OSFILE System file not available
73: CRM_EX_CANTCREAT Cannot create output file
74: CRM_EX_IOERR I/O error occurred
75: CRM_EX_TEMPFAIL Temporary failure, try again
76: CRM_EX_PROTOCOL Protocol violated
77: CRM_EX_NOPERM Insufficient privileges
78: CRM_EX_CONFIG Invalid configuration
100: CRM_EX_FATAL Fatal error occurred, will not respawn
101: CRM_EX_PANIC System panic required
102: CRM_EX_DISCONNECT Not connected
103: CRM_EX_OLD Update was older than existing configuration
104: CRM_EX_DIGEST Digest mismatch
105: CRM_EX_NOSUCH No such object
106: CRM_EX_QUORUM Quorum required
107: CRM_EX_UNSAFE Operation not safe
108: CRM_EX_EXISTS Requested item already exists
109: CRM_EX_MULTIPLE Multiple items match request
110: CRM_EX_EXPIRED Requested item has expired
111: CRM_EX_NOT_YET_IN_EFFECT Requested item is not yet in effect
112: CRM_EX_INDETERMINATE Could not determine status
113: CRM_EX_UNSATISFIED Not applicable under current conditions
114: CRM_EX_NO_DC DC is not yet elected
124: CRM_EX_TIMEOUT Timeout occurred
190: CRM_EX_DEGRADED Service is active but might fail soon
191: CRM_EX_DEGRADED_PROMOTED Service is promoted but might fail soon
193: CRM_EX_NONE No exit status available
=#=#=#= End test: Get all crm_exit_t exit codes (with name) - OK (0) =#=#=#=
* Passed: crm_error - Get all crm_exit_t exit codes (with name)
=#=#=#= Begin test: Get all crm_exit_t exit codes (with name) (XML) =#=#=#=
=#=#=#= End test: Get all crm_exit_t exit codes (with name) (XML) - OK (0) =#=#=#=
* Passed: crm_error - Get all crm_exit_t exit codes (with name) (XML)
diff --git a/daemons/based/based_notify.c b/daemons/based/based_notify.c
index 243c1f7db7..154bb68b0c 100644
--- a/daemons/based/based_notify.c
+++ b/daemons/based/based_notify.c
@@ -1,198 +1,220 @@
/*
* Copyright 2004-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU General Public License version 2
* or later (GPLv2+) WITHOUT ANY WARRANTY.
*/
#include
#include
#include
#include
#include
#include // PRIx64
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
struct cib_notification_s {
const xmlNode *msg;
struct iovec *iov;
int32_t iov_size;
};
static void
cib_notify_send_one(gpointer key, gpointer value, gpointer user_data)
{
const char *type = NULL;
gboolean do_send = FALSE;
int rc = pcmk_rc_ok;
pcmk__client_t *client = value;
struct cib_notification_s *update = user_data;
if (client->ipcs == NULL && client->remote == NULL) {
crm_warn("Skipping client with NULL channel");
return;
}
type = crm_element_value(update->msg, PCMK__XA_SUBT);
CRM_LOG_ASSERT(type != NULL);
if (pcmk_is_set(client->flags, cib_notify_diff)
&& pcmk__str_eq(type, PCMK__VALUE_CIB_DIFF_NOTIFY, pcmk__str_none)) {
do_send = TRUE;
} else if (pcmk_is_set(client->flags, cib_notify_confirm)
&& pcmk__str_eq(type, PCMK__VALUE_CIB_UPDATE_CONFIRMATION,
pcmk__str_none)) {
do_send = TRUE;
} else if (pcmk_is_set(client->flags, cib_notify_pre)
&& pcmk__str_eq(type, PCMK__VALUE_CIB_PRE_NOTIFY,
pcmk__str_none)) {
do_send = TRUE;
} else if (pcmk_is_set(client->flags, cib_notify_post)
&& pcmk__str_eq(type, PCMK__VALUE_CIB_POST_NOTIFY,
pcmk__str_none)) {
do_send = TRUE;
}
if (do_send) {
switch (PCMK__CLIENT_TYPE(client)) {
case pcmk__client_ipc:
rc = pcmk__ipc_send_iov(client, update->iov,
crm_ipc_server_event);
- if (rc != pcmk_rc_ok) {
+
+ /* EAGAIN isn't an error for server events. Sending did fail
+ * with EAGAIN, but the iov was added to the send queue and we
+ * will attempt to send it again the next time pcmk__ipc_send_iov
+ * is called, or when crm_ipcs_flush_events_cb happens.
+ */
+ if ((rc != EAGAIN) && (rc != pcmk_rc_ok)) {
crm_warn("Could not notify client %s: %s " QB_XS " id=%s",
pcmk__client_name(client), pcmk_rc_str(rc),
client->id);
}
break;
case pcmk__client_tls:
case pcmk__client_tcp:
crm_debug("Sent %s notification to client %s (id %s)",
type, pcmk__client_name(client), client->id);
pcmk__remote_send_xml(client->remote, update->msg);
break;
default:
crm_err("Unknown transport for client %s "
QB_XS " flags=%#016" PRIx64,
pcmk__client_name(client), client->flags);
}
}
}
static void
cib_notify_send(const xmlNode *xml)
{
struct iovec *iov;
struct cib_notification_s update;
-
+ GString *iov_buffer = NULL;
ssize_t bytes = 0;
- int rc = pcmk__ipc_prepare_iov(0, xml, &iov, &bytes);
+ int rc = pcmk_rc_ok;
+ uint16_t index = 0;
+
+ iov_buffer = g_string_sized_new(1024);
+ pcmk__xml_string(xml, 0, iov_buffer, 0);
+
+ do {
+ rc = pcmk__ipc_prepare_iov(0, iov_buffer, index, &iov, &bytes);
+
+ if ((rc != pcmk_rc_ok) && (rc != pcmk_rc_ipc_more)) {
+ crm_notice("Could not notify clients: %s " QB_XS " rc=%d",
+ pcmk_rc_str(rc), rc);
+ break;
+ }
- if (rc == pcmk_rc_ok) {
update.msg = xml;
update.iov = iov;
update.iov_size = bytes;
pcmk__foreach_ipc_client(cib_notify_send_one, &update);
pcmk_free_ipc_event(iov);
- } else {
- crm_notice("Could not notify clients: %s " QB_XS " rc=%d",
- pcmk_rc_str(rc), rc);
- }
+ if (rc == pcmk_rc_ok) {
+ break;
+ }
+
+ index++;
+ } while (true);
+
+ g_string_free(iov_buffer, TRUE);
}
void
cib_diff_notify(const char *op, int result, const char *call_id,
const char *client_id, const char *client_name,
const char *origin, xmlNode *update, xmlNode *diff)
{
int add_updates = 0;
int add_epoch = 0;
int add_admin_epoch = 0;
int del_updates = 0;
int del_epoch = 0;
int del_admin_epoch = 0;
uint8_t log_level = LOG_TRACE;
xmlNode *update_msg = NULL;
xmlNode *wrapper = NULL;
if (diff == NULL) {
return;
}
if (result != pcmk_ok) {
log_level = LOG_WARNING;
}
cib_diff_version_details(diff, &add_admin_epoch, &add_epoch, &add_updates,
&del_admin_epoch, &del_epoch, &del_updates);
if ((add_admin_epoch != del_admin_epoch)
|| (add_epoch != del_epoch)
|| (add_updates != del_updates)) {
do_crm_log(log_level,
"Updated CIB generation %d.%d.%d to %d.%d.%d from client "
"%s%s%s (%s) (%s)",
del_admin_epoch, del_epoch, del_updates,
add_admin_epoch, add_epoch, add_updates,
client_name,
((call_id != NULL)? " call " : ""), pcmk__s(call_id, ""),
pcmk__s(origin, "unspecified peer"), pcmk_strerror(result));
} else if ((add_admin_epoch != 0)
|| (add_epoch != 0)
|| (add_updates != 0)) {
do_crm_log(log_level,
"Local-only change to CIB generation %d.%d.%d from client "
"%s%s%s (%s) (%s)",
add_admin_epoch, add_epoch, add_updates,
client_name,
((call_id != NULL)? " call " : ""), pcmk__s(call_id, ""),
pcmk__s(origin, "unspecified peer"), pcmk_strerror(result));
}
update_msg = pcmk__xe_create(NULL, PCMK__XE_NOTIFY);
crm_xml_add(update_msg, PCMK__XA_T, PCMK__VALUE_CIB_NOTIFY);
crm_xml_add(update_msg, PCMK__XA_SUBT, PCMK__VALUE_CIB_DIFF_NOTIFY);
crm_xml_add(update_msg, PCMK__XA_CIB_OP, op);
crm_xml_add(update_msg, PCMK__XA_CIB_CLIENTID, client_id);
crm_xml_add(update_msg, PCMK__XA_CIB_CLIENTNAME, client_name);
crm_xml_add(update_msg, PCMK__XA_CIB_CALLID, call_id);
crm_xml_add(update_msg, PCMK__XA_SRC, origin);
crm_xml_add_int(update_msg, PCMK__XA_CIB_RC, result);
wrapper = pcmk__xe_create(update_msg, PCMK__XE_CIB_UPDATE_RESULT);
pcmk__xml_copy(wrapper, diff);
crm_log_xml_trace(update_msg, "diff-notify");
cib_notify_send(update_msg);
pcmk__xml_free(update_msg);
}
diff --git a/include/crm/common/ipc.h b/include/crm/common/ipc.h
index 3890919d81..df84d15013 100644
--- a/include/crm/common/ipc.h
+++ b/include/crm/common/ipc.h
@@ -1,216 +1,220 @@
/*
* Copyright 2004-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#ifndef PCMK__CRM_COMMON_IPC__H
#define PCMK__CRM_COMMON_IPC__H
#include
#include
#include
#include
#ifdef __cplusplus
extern "C" {
#endif
/**
* \file
* \brief IPC interface to Pacemaker daemons
*
* \ingroup core
*/
/*
* The library supports two methods of creating IPC connections. The older code
* allows connecting to any arbitrary IPC name. The newer code only allows
* connecting to one of the Pacemaker daemons.
*
* As daemons are converted to use the new model, the old functions should be
* considered deprecated for use with those daemons. Once all daemons are
* converted, the old functions should be officially deprecated as public API
* and eventually made internal API.
*/
/*
* Pacemaker daemon IPC
*/
/* @COMPAT This is also used internally for cluster message types, but it's not
* worth the hassle of redefining this public API just to change the name.
*/
//! Available IPC interfaces
enum pcmk_ipc_server {
pcmk_ipc_unknown, //!< Unknown or invalid
pcmk_ipc_attrd, //!< Attribute manager
pcmk_ipc_based, //!< CIB manager
pcmk_ipc_controld, //!< Controller
pcmk_ipc_execd, //!< Executor
pcmk_ipc_fenced, //!< Fencer
pcmk_ipc_pacemakerd, //!< Launcher
pcmk_ipc_schedulerd, //!< Scheduler
};
// NOTE: sbd (as of at least 1.5.2) uses this enum
//! Possible event types that an IPC event callback can be called for
enum pcmk_ipc_event {
pcmk_ipc_event_connect, //!< Result of asynchronous connection attempt
// NOTE: sbd (as of at least 1.5.2) uses this value
pcmk_ipc_event_disconnect, //!< Termination of IPC connection
// NOTE: sbd (as of at least 1.5.2) uses this value
pcmk_ipc_event_reply, //!< Daemon's reply to client IPC request
pcmk_ipc_event_notify, //!< Notification from daemon
};
//! How IPC replies should be dispatched
enum pcmk_ipc_dispatch {
pcmk_ipc_dispatch_main, //!< Attach IPC to GMainLoop for dispatch
pcmk_ipc_dispatch_poll, //!< Caller will poll and dispatch IPC
pcmk_ipc_dispatch_sync, //!< Sending a command will wait for any reply
};
// NOTE: sbd (as of at least 1.5.2) uses this
//! Client connection to Pacemaker IPC
typedef struct pcmk_ipc_api_s pcmk_ipc_api_t;
/*!
* \brief Callback function type for Pacemaker daemon IPC APIs
*
* \param[in,out] api IPC API connection
* \param[in] event_type The type of event that occurred
* \param[in] status Event status
* \param[in,out] event_data Event-specific data
* \param[in,out] user_data Caller data provided when callback was registered
*
* \note For connection and disconnection events, event_data may be NULL (for
* local IPC) or the name of the connected node (for remote IPC, for
* daemons that support that). For reply and notify events, event_data is
* defined by the specific daemon API.
*/
typedef void (*pcmk_ipc_callback_t)(pcmk_ipc_api_t *api,
enum pcmk_ipc_event event_type,
crm_exit_t status,
void *event_data, void *user_data);
// NOTE: sbd (as of at least 1.5.2) uses this
int pcmk_new_ipc_api(pcmk_ipc_api_t **api, enum pcmk_ipc_server server);
// NOTE: sbd (as of at least 1.5.2) uses this
void pcmk_free_ipc_api(pcmk_ipc_api_t *api);
// NOTE: sbd (as of at least 1.5.2) uses this
int pcmk_connect_ipc(pcmk_ipc_api_t *api, enum pcmk_ipc_dispatch dispatch_type);
void pcmk_disconnect_ipc(pcmk_ipc_api_t *api);
int pcmk_poll_ipc(const pcmk_ipc_api_t *api, int timeout_ms);
void pcmk_dispatch_ipc(pcmk_ipc_api_t *api);
// NOTE: sbd (as of at least 1.5.2) uses this
void pcmk_register_ipc_callback(pcmk_ipc_api_t *api, pcmk_ipc_callback_t cb,
void *user_data);
const char *pcmk_ipc_name(const pcmk_ipc_api_t *api, bool for_log);
bool pcmk_ipc_is_connected(pcmk_ipc_api_t *api);
int pcmk_ipc_purge_node(pcmk_ipc_api_t *api, const char *node_name,
uint32_t nodeid);
/*
* Generic IPC API (to eventually be deprecated as public API and made internal)
*/
enum crm_ipc_flags
{
crm_ipc_flags_none = UINT32_C(0),
//! \deprecated Since 3.0.1
crm_ipc_compressed = (UINT32_C(1) << 0),
//! _ALL_ replies to proxied connections need to be sent as events
crm_ipc_proxied = (UINT32_C(1) << 8),
//! A response is expected in reply
crm_ipc_client_response = (UINT32_C(1) << 9),
// These are options for Pacemaker's internal use only (pcmk__ipc_send_*())
//! Send an Event instead of a Response
crm_ipc_server_event = (UINT32_C(1) << 16),
//! Free the iovec after sending
crm_ipc_server_free = (UINT32_C(1) << 17),
//! All replies to proxied connections are sent as events. This flag
//! preserves whether the events should be treated as an Event or a Response
crm_ipc_proxied_relay_response = (UINT32_C(1) << 18),
+ //! This is a multi-part IPC message
+ crm_ipc_multipart = (UINT32_C(1) << 19),
+ //! This is the end of a multi-part IPC message
+ crm_ipc_multipart_end = (UINT32_C(1) << 20),
};
typedef struct crm_ipc_s crm_ipc_t;
crm_ipc_t *crm_ipc_new(const char *name, size_t max_size);
void crm_ipc_close(crm_ipc_t * client);
void crm_ipc_destroy(crm_ipc_t * client);
void pcmk_free_ipc_event(struct iovec *event);
int crm_ipc_send(crm_ipc_t *client, const xmlNode *message,
enum crm_ipc_flags flags, int32_t ms_timeout, xmlNode **reply);
int crm_ipc_get_fd(crm_ipc_t * client);
bool crm_ipc_connected(crm_ipc_t * client);
int crm_ipc_ready(crm_ipc_t * client);
long crm_ipc_read(crm_ipc_t * client);
const char *crm_ipc_buffer(crm_ipc_t * client);
uint32_t crm_ipc_buffer_flags(crm_ipc_t * client);
const char *crm_ipc_name(crm_ipc_t * client);
unsigned int crm_ipc_default_buffer_size(void);
/*!
* \brief Check the authenticity of the IPC socket peer process (legacy)
*
* If everything goes well, peer's authenticity is verified by the means
* of comparing against provided referential UID and GID (either satisfies),
* and the result of this check can be deduced from the return value.
* As an exception, detected UID of 0 ("root") satisfies arbitrary
* provided referential daemon's credentials.
*
* \param[in] sock IPC related, connected Unix socket to check peer of
* \param[in] refuid referential UID to check against
* \param[in] refgid referential GID to check against
* \param[out] gotpid to optionally store obtained PID of the peer
* (not available on FreeBSD, special value of 1
* used instead, and the caller is required to
* special case this value respectively)
* \param[out] gotuid to optionally store obtained UID of the peer
* \param[out] gotgid to optionally store obtained GID of the peer
*
* \return 0 if IPC related socket's peer is not authentic given the
* referential credentials (see above), 1 if it is,
* negative value on error (generally expressing -errno unless
* it was zero even on nonhappy path, -pcmk_err_generic is
* returned then; no message is directly emitted)
*
* \note While this function is tolerant on what constitutes authorized
* IPC daemon process (its effective user matches UID=0 or \p refuid,
* or at least its group matches \p refgid), either or both (in case
* of UID=0) mismatches on the expected credentials of such peer
* process \e shall be investigated at the caller when value of 1
* gets returned there, since higher-than-expected privileges in
* respect to the expected/intended credentials possibly violate
* the least privilege principle and may pose an additional risk
* (i.e. such accidental inconsistency shall be eventually fixed).
*/
int crm_ipc_is_authentic_process(int sock, uid_t refuid, gid_t refgid,
pid_t *gotpid, uid_t *gotuid, gid_t *gotgid);
#ifdef __cplusplus
}
#endif
#endif
diff --git a/include/crm/common/ipc_internal.h b/include/crm/common/ipc_internal.h
index 72b6f7f189..108210cb78 100644
--- a/include/crm/common/ipc_internal.h
+++ b/include/crm/common/ipc_internal.h
@@ -1,286 +1,288 @@
/*
* Copyright 2013-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#ifndef PCMK__CRM_COMMON_IPC_INTERNAL__H
#define PCMK__CRM_COMMON_IPC_INTERNAL__H
#include // bool
#include // uint32_t, uint64_t, UINT64_C()
#include // struct iovec
#include // uid_t, gid_t, pid_t, size_t
#include // gnutls_session_t
#include // guint, gpointer, GQueue, ...
#include // xmlNode
#include // qb_ipcs_connection_t, ...
#include // HAVE_GETPEEREID
#include // crm_system_name
#include
#include // pcmk_controld_api_reply
#include // pcmk_pacemakerd_{api_reply,state}
#include // mainloop_io_t
#ifdef __cplusplus
extern "C" {
#endif
/* denotes "non yieldable PID" on FreeBSD, or actual PID1 in scenarios that
require a delicate handling anyway (socket-based activation with systemd);
we can be reasonably sure that this PID is never possessed by the actual
child daemon, as it gets taken either by the proper init, or by pacemakerd
itself (i.e. this precludes anything else); note that value of zero
is meant to carry "unset" meaning, and better not to bet on/conditionalize
over signedness of pid_t */
#define PCMK__SPECIAL_PID 1
// Timeout (in seconds) to use for IPC client sends, reply waits, etc.
#define PCMK__IPC_TIMEOUT 120
#if defined(HAVE_GETPEEREID)
/* on FreeBSD, we don't want to expose "non-yieldable PID" (leading to
"IPC liveness check only") as its nominal representation, which could
cause confusion -- this is unambiguous as long as there's no
socket-based activation like with systemd (very improbable) */
#define PCMK__SPECIAL_PID_AS_0(p) (((p) == PCMK__SPECIAL_PID) ? 0 : (p))
#else
#define PCMK__SPECIAL_PID_AS_0(p) (p)
#endif
/*!
* \internal
* \brief Check the authenticity and liveness of the process via IPC end-point
*
* When IPC daemon under given IPC end-point (name) detected, its authenticity
* is verified by the means of comparing against provided referential UID and
* GID, and the result of this check can be deduced from the return value.
* As an exception, referential UID of 0 (~ root) satisfies arbitrary
* detected daemon's credentials.
*
* \param[in] name IPC name to base the search on
* \param[in] refuid referential UID to check against
* \param[in] refgid referential GID to check against
* \param[out] gotpid to optionally store obtained PID of the found process
* upon returning 1 or -2
* (not available on FreeBSD, special value of 1,
* see PCMK__SPECIAL_PID, used instead, and the caller
* is required to special case this value respectively)
*
* \return Standard Pacemaker return code
*
* \note Return codes of particular interest include pcmk_rc_ipc_unresponsive
* indicating that no trace of IPC liveness was detected, and
* pcmk_rc_ipc_unauthorized indicating that the IPC endpoint is blocked by
* an unauthorized process.
* \note This function emits a log message for return codes other than
* pcmk_rc_ok and pcmk_rc_ipc_unresponsive, and when there isn't a perfect
* match in respect to \p reguid and/or \p refgid, for a possible
* least privilege principle violation.
*
* \see crm_ipc_is_authentic_process
*/
int pcmk__ipc_is_authentic_process_active(const char *name, uid_t refuid,
gid_t refgid, pid_t *gotpid);
int pcmk__connect_generic_ipc(crm_ipc_t *ipc);
int pcmk__ipc_fd(crm_ipc_t *ipc, int *fd);
int pcmk__connect_ipc(pcmk_ipc_api_t *api, enum pcmk_ipc_dispatch dispatch_type,
int attempts);
/*
* Server-related
*/
typedef struct pcmk__client_s pcmk__client_t;
struct pcmk__remote_s {
/* Shared */
char *buffer;
size_t buffer_size;
size_t buffer_offset;
int auth_timeout;
int tcp_socket;
mainloop_io_t *source;
time_t uptime;
char *start_state;
/* CIB-only */
char *token;
/* TLS only */
// Must be created by pcmk__new_tls_session()
gnutls_session_t tls_session;
};
enum pcmk__client_flags {
// Lower 32 bits are reserved for server (not library) use
// Next 8 bits are reserved for client type (sort of a cheap enum)
//! Client uses plain IPC
pcmk__client_ipc = (UINT64_C(1) << 32),
//! Client uses TCP connection
pcmk__client_tcp = (UINT64_C(1) << 33),
//! Client uses TCP with TLS
pcmk__client_tls = (UINT64_C(1) << 34),
// The rest are client attributes
//! Client IPC is proxied
pcmk__client_proxied = (UINT64_C(1) << 40),
//! Client is run by root or cluster user
pcmk__client_privileged = (UINT64_C(1) << 41),
//! Local client to be proxied
pcmk__client_to_proxy = (UINT64_C(1) << 42),
/*!
* \brief Client IPC connection accepted
*
* Used only for remote CIB connections via \c PCMK_XA_REMOTE_TLS_PORT.
*/
pcmk__client_authenticated = (UINT64_C(1) << 43),
//! Client TLS handshake is complete
pcmk__client_tls_handshake_complete = (UINT64_C(1) << 44),
};
#define PCMK__CLIENT_TYPE(client) ((client)->flags & UINT64_C(0xff00000000))
struct pcmk__client_s {
unsigned int pid;
char *id;
char *name;
char *user;
uint64_t flags; // Group of pcmk__client_flags
int request_id;
void *userdata;
int event_timer;
GQueue *event_queue;
/* Depending on the client type, only some of the following will be
* populated/valid. @TODO Maybe convert to a union.
*/
qb_ipcs_connection_t *ipcs; /* IPC */
struct pcmk__remote_s *remote; /* TCP/TLS */
unsigned int queue_backlog; /* IPC queue length after last flush */
unsigned int queue_max; /* Evict client whose queue grows this big */
};
#define pcmk__set_client_flags(client, flags_to_set) do { \
(client)->flags = pcmk__set_flags_as(__func__, __LINE__, \
LOG_TRACE, \
"Client", pcmk__client_name(client), \
(client)->flags, (flags_to_set), #flags_to_set); \
} while (0)
#define pcmk__clear_client_flags(client, flags_to_clear) do { \
(client)->flags = pcmk__clear_flags_as(__func__, __LINE__, \
LOG_TRACE, \
"Client", pcmk__client_name(client), \
(client)->flags, (flags_to_clear), #flags_to_clear); \
} while (0)
#define pcmk__set_ipc_flags(ipc_flags, ipc_name, flags_to_set) do { \
ipc_flags = pcmk__set_flags_as(__func__, __LINE__, LOG_TRACE, \
"IPC", (ipc_name), \
(ipc_flags), (flags_to_set), \
#flags_to_set); \
} while (0)
#define pcmk__clear_ipc_flags(ipc_flags, ipc_name, flags_to_clear) do { \
ipc_flags = pcmk__clear_flags_as(__func__, __LINE__, LOG_TRACE, \
"IPC", (ipc_name), \
(ipc_flags), (flags_to_clear), \
#flags_to_clear); \
} while (0)
guint pcmk__ipc_client_count(void);
void pcmk__foreach_ipc_client(GHFunc func, gpointer user_data);
void pcmk__client_cleanup(void);
pcmk__client_t *pcmk__find_client(const qb_ipcs_connection_t *c);
pcmk__client_t *pcmk__find_client_by_id(const char *id);
const char *pcmk__client_name(const pcmk__client_t *c);
const char *pcmk__client_type_str(uint64_t client_type);
pcmk__client_t *pcmk__new_unauth_client(void *key);
pcmk__client_t *pcmk__new_client(qb_ipcs_connection_t *c, uid_t uid, gid_t gid);
void pcmk__free_client(pcmk__client_t *c);
void pcmk__drop_all_clients(qb_ipcs_service_t *s);
void pcmk__set_client_queue_max(pcmk__client_t *client, const char *qmax);
xmlNode *pcmk__ipc_create_ack_as(const char *function, int line, uint32_t flags,
const char *tag, const char *ver, crm_exit_t status);
#define pcmk__ipc_create_ack(flags, tag, ver, st) \
pcmk__ipc_create_ack_as(__func__, __LINE__, (flags), (tag), (ver), (st))
int pcmk__ipc_send_ack_as(const char *function, int line, pcmk__client_t *c,
uint32_t request, uint32_t flags, const char *tag,
const char *ver, crm_exit_t status);
#define pcmk__ipc_send_ack(c, req, flags, tag, ver, st) \
pcmk__ipc_send_ack_as(__func__, __LINE__, (c), (req), (flags), (tag), (ver), (st))
-int pcmk__ipc_prepare_iov(uint32_t request, const xmlNode *message,
- struct iovec **result, ssize_t *bytes);
+int pcmk__ipc_prepare_iov(uint32_t request, const GString *message,
+ uint16_t index, struct iovec **result, ssize_t *bytes);
int pcmk__ipc_send_xml(pcmk__client_t *c, uint32_t request,
const xmlNode *message, uint32_t flags);
int pcmk__ipc_send_iov(pcmk__client_t *c, struct iovec *iov, uint32_t flags);
+void pcmk__ipc_free_client_buffer(crm_ipc_t *client);
+int pcmk__ipc_msg_append(GByteArray **buffer, guint8 *data);
xmlNode *pcmk__client_data2xml(pcmk__client_t *c, void *data,
uint32_t *id, uint32_t *flags);
int pcmk__client_pid(qb_ipcs_connection_t *c);
void pcmk__serve_attrd_ipc(qb_ipcs_service_t **ipcs,
struct qb_ipcs_service_handlers *cb);
void pcmk__serve_fenced_ipc(qb_ipcs_service_t **ipcs,
struct qb_ipcs_service_handlers *cb);
void pcmk__serve_pacemakerd_ipc(qb_ipcs_service_t **ipcs,
struct qb_ipcs_service_handlers *cb);
qb_ipcs_service_t *pcmk__serve_schedulerd_ipc(struct qb_ipcs_service_handlers *cb);
qb_ipcs_service_t *pcmk__serve_controld_ipc(struct qb_ipcs_service_handlers *cb);
void pcmk__serve_based_ipc(qb_ipcs_service_t **ipcs_ro,
qb_ipcs_service_t **ipcs_rw,
qb_ipcs_service_t **ipcs_shm,
struct qb_ipcs_service_handlers *ro_cb,
struct qb_ipcs_service_handlers *rw_cb);
void pcmk__stop_based_ipc(qb_ipcs_service_t *ipcs_ro,
qb_ipcs_service_t *ipcs_rw,
qb_ipcs_service_t *ipcs_shm);
static inline const char *
pcmk__ipc_sys_name(const char *ipc_name, const char *fallback)
{
return ipc_name ? ipc_name : ((crm_system_name ? crm_system_name : fallback));
}
const char *pcmk__pcmkd_state_enum2friendly(enum pcmk_pacemakerd_state state);
const char *pcmk__controld_api_reply2str(enum pcmk_controld_api_reply reply);
const char *pcmk__pcmkd_api_reply2str(enum pcmk_pacemakerd_api_reply reply);
#ifdef __cplusplus
}
#endif
#endif // PCMK__CRM_COMMON_IPC_INTERNAL__H
diff --git a/include/crm/common/results.h b/include/crm/common/results.h
index 7525282c3f..9222325ac5 100644
--- a/include/crm/common/results.h
+++ b/include/crm/common/results.h
@@ -1,403 +1,404 @@
/*
* Copyright 2012-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#ifndef PCMK__CRM_COMMON_RESULTS__H
#define PCMK__CRM_COMMON_RESULTS__H
#include // gboolean
#ifdef __cplusplus
extern "C" {
#endif
/*!
* \file
* \brief Function and executable result codes
* \ingroup core
*/
// Lifted from config.h
/* The _Noreturn keyword of C11. */
#ifndef _Noreturn
# if (defined __cplusplus \
&& ((201103 <= __cplusplus && !(__GNUC__ == 4 && __GNUC_MINOR__ == 7)) \
|| (defined _MSC_VER && 1900 <= _MSC_VER)))
# define _Noreturn [[noreturn]]
# elif ((!defined __cplusplus || defined __clang__) \
&& (201112 <= (defined __STDC_VERSION__ ? __STDC_VERSION__ : 0) \
|| 4 < __GNUC__ + (7 <= __GNUC_MINOR__)))
/* _Noreturn works as-is. */
# elif 2 < __GNUC__ + (8 <= __GNUC_MINOR__) || 0x5110 <= __SUNPRO_C
# define _Noreturn __attribute__ ((__noreturn__))
# elif 1200 <= (defined _MSC_VER ? _MSC_VER : 0)
# define _Noreturn __declspec (noreturn)
# else
# define _Noreturn
# endif
#endif
/*
* Function return codes
*
* Most Pacemaker API functions return an integer return code. There are two
* alternative interpretations. The legacy interpration is that the absolute
* value of the return code is either a system error number or a custom
* pcmk_err_* number. This is less than ideal because system error numbers are
* constrained only to the positive int range, so there's the possibility that
* system errors and custom errors could collide (which did in fact happen
* already on one architecture). The new intepretation is that negative values
* are from the pcmk_rc_e enum, and positive values are system error numbers.
* Both use 0 for success.
*
* For system error codes, see:
* - /usr/include/asm-generic/errno.h
* - /usr/include/asm-generic/errno-base.h
*/
// Legacy custom return codes for Pacemaker API functions (deprecated)
// NOTE: sbd (as of at least 1.5.2) uses this
#define pcmk_ok 0
#define PCMK_ERROR_OFFSET 190 /* Replacements on non-linux systems, see include/portability.h */
#define PCMK_CUSTOM_OFFSET 200 /* Purely custom codes */
#define pcmk_err_generic 201
#define pcmk_err_no_quorum 202
#define pcmk_err_schema_validation 203
#define pcmk_err_transform_failed 204
#define pcmk_err_old_data 205
// NOTE: sbd (as of at least 1.5.2) uses this
#define pcmk_err_diff_failed 206
// NOTE: sbd (as of at least 1.5.2) uses this
#define pcmk_err_diff_resync 207
#define pcmk_err_cib_modified 208
#define pcmk_err_cib_backup 209
#define pcmk_err_cib_save 210
#define pcmk_err_schema_unchanged 211
#define pcmk_err_cib_corrupt 212
#define pcmk_err_multiple 213
#define pcmk_err_node_unknown 214
#define pcmk_err_already 215
/* On HPPA 215 is ENOSYM (Unknown error 215), which hopefully never happens. */
#ifdef __hppa__
#define pcmk_err_bad_nvpair 250 /* 216 is ENOTSOCK */
#define pcmk_err_unknown_format 252 /* 217 is EDESTADDRREQ */
#else
#define pcmk_err_bad_nvpair 216
#define pcmk_err_unknown_format 217
#endif
/*!
* \brief Return codes for Pacemaker API functions
*
* Any Pacemaker API function documented as returning a "standard Pacemaker
* return code" will return pcmk_rc_ok (0) on success, and one of this
* enumeration's other (negative) values or a (positive) system error number
* otherwise. The custom codes are at -1001 and lower, so that the caller may
* use -1 through -1000 for their own custom values if desired. While generally
* referred to as "errors", nonzero values simply indicate a result, which might
* or might not be an error depending on the calling context.
*/
enum pcmk_rc_e {
/* When adding new values, use consecutively lower numbers, update the array
* in lib/common/results.c, and test with crm_error.
*/
+ pcmk_rc_ipc_more = -1041,
pcmk_rc_no_dc = -1040,
pcmk_rc_compression = -1039,
pcmk_rc_ns_resolution = -1038,
pcmk_rc_no_transaction = -1037,
pcmk_rc_bad_xml_patch = -1036,
pcmk_rc_bad_input = -1035,
pcmk_rc_disabled = -1034,
pcmk_rc_duplicate_id = -1033,
pcmk_rc_unpack_error = -1032,
pcmk_rc_invalid_transition = -1031,
pcmk_rc_graph_error = -1030,
pcmk_rc_dot_error = -1029,
pcmk_rc_underflow = -1028,
pcmk_rc_no_input = -1027,
pcmk_rc_no_output = -1026,
pcmk_rc_after_range = -1025,
pcmk_rc_within_range = -1024,
pcmk_rc_before_range = -1023,
pcmk_rc_undetermined = -1022,
pcmk_rc_op_unsatisfied = -1021,
pcmk_rc_ipc_pid_only = -1020,
pcmk_rc_ipc_unresponsive = -1019,
pcmk_rc_ipc_unauthorized = -1018,
pcmk_rc_no_quorum = -1017,
pcmk_rc_schema_validation = -1016,
pcmk_rc_schema_unchanged = -1015,
pcmk_rc_transform_failed = -1014,
pcmk_rc_old_data = -1013,
pcmk_rc_diff_failed = -1012,
pcmk_rc_diff_resync = -1011,
pcmk_rc_cib_modified = -1010,
pcmk_rc_cib_backup = -1009,
pcmk_rc_cib_save = -1008,
pcmk_rc_cib_corrupt = -1007,
pcmk_rc_multiple = -1006,
pcmk_rc_node_unknown = -1005,
pcmk_rc_already = -1004,
pcmk_rc_bad_nvpair = -1003,
pcmk_rc_unknown_format = -1002,
// Developers: Use a more specific code than pcmk_rc_error whenever possible
pcmk_rc_error = -1001,
// Values -1 through -1000 reserved for caller use
// NOTE: sbd (as of at least 1.5.2) uses this
pcmk_rc_ok = 0
// Positive values reserved for system error numbers
};
/*!
* \brief Exit status codes for resource agents
*
* The OCF Resource Agent API standard enumerates the possible exit status codes
* that agents should return. Besides being used with OCF agents, these values
* are also used by the executor as a universal status for all agent standards;
* actual results are mapped to these before returning them to clients.
*/
enum ocf_exitcode {
PCMK_OCF_OK = 0, //!< Success
// NOTE: booth (as of at least 1.1) uses this value
PCMK_OCF_UNKNOWN_ERROR = 1, //!< Unspecified error
PCMK_OCF_INVALID_PARAM = 2, //!< Parameter invalid (in local context)
PCMK_OCF_UNIMPLEMENT_FEATURE = 3, //!< Requested action not implemented
PCMK_OCF_INSUFFICIENT_PRIV = 4, //!< Insufficient privileges
PCMK_OCF_NOT_INSTALLED = 5, //!< Dependencies not available locally
PCMK_OCF_NOT_CONFIGURED = 6, //!< Parameter invalid (inherently)
// NOTE: booth (as of at least 1.1) uses this value
PCMK_OCF_NOT_RUNNING = 7, //!< Service safely stopped
PCMK_OCF_RUNNING_PROMOTED = 8, //!< Service active and promoted
PCMK_OCF_FAILED_PROMOTED = 9, //!< Service failed and possibly in promoted role
PCMK_OCF_DEGRADED = 190, //!< Service active but more likely to fail soon
PCMK_OCF_DEGRADED_PROMOTED = 191, //!< Service promoted but more likely to fail soon
/* These two are Pacemaker extensions, not in the OCF standard. The
* controller records PCMK_OCF_UNKNOWN for pending actions.
* PCMK_OCF_CONNECTION_DIED is used only with older DCs that don't support
* PCMK_EXEC_NOT_CONNECTED.
*/
PCMK_OCF_CONNECTION_DIED = 189, //!< \deprecated See PCMK_EXEC_NOT_CONNECTED
PCMK_OCF_UNKNOWN = 193, //!< Action is pending
};
// NOTE: sbd (as of at least 1.5.2) uses this
/*!
* \brief Exit status codes for tools and daemons
*
* We want well-specified (i.e. OS-invariant) exit status codes for our daemons
* and applications so they can be relied on by callers. (Function return codes
* and errno's do not make good exit statuses.)
*
* The only hard rule is that exit statuses must be between 0 and 255; all else
* is convention. Universally, 0 is success, and 1 is generic error (excluding
* OSes we don't support -- for example, OpenVMS considers 1 success!).
*
* For init scripts, the LSB gives meaning to 0-7, and sets aside 150-199 for
* application use. OCF adds 8-9 and 190-191.
*
* sysexits.h was an attempt to give additional meanings, but never really
* caught on. It uses 0 and 64-78.
*
* Bash reserves 2 ("incorrect builtin usage") and 126-255 (126 is "command
* found but not executable", 127 is "command not found", 128 + n is
* "interrupted by signal n").
*
* tldp.org recommends 64-113 for application use.
*
* We try to overlap with the above conventions when practical.
*
* NOTE: When new exit codes are added here, remember to also update
* python/pacemaker/exitstatus.py.
*/
typedef enum crm_exit_e {
// Common convention
CRM_EX_OK = 0, //!< Success
CRM_EX_ERROR = 1, //!< Unspecified error
// LSB + OCF
CRM_EX_INVALID_PARAM = 2, //!< Parameter invalid (in local context)
CRM_EX_UNIMPLEMENT_FEATURE = 3, //!< Requested action not implemented
CRM_EX_INSUFFICIENT_PRIV = 4, //!< Insufficient privileges
CRM_EX_NOT_INSTALLED = 5, //!< Dependencies not available locally
CRM_EX_NOT_CONFIGURED = 6, //!< Parameter invalid (inherently)
CRM_EX_NOT_RUNNING = 7, //!< Service safely stopped
CRM_EX_PROMOTED = 8, //!< Service active and promoted
CRM_EX_FAILED_PROMOTED = 9, //!< Service failed and possibly promoted
// sysexits.h
CRM_EX_USAGE = 64, //!< Command line usage error
CRM_EX_DATAERR = 65, //!< User-supplied data incorrect
CRM_EX_NOINPUT = 66, //!< Input file not available
CRM_EX_NOUSER = 67, //!< User does not exist
CRM_EX_NOHOST = 68, //!< Host unknown
CRM_EX_UNAVAILABLE = 69, //!< Needed service unavailable
CRM_EX_SOFTWARE = 70, //!< Internal software bug
CRM_EX_OSERR = 71, //!< External (OS/environmental) problem
CRM_EX_OSFILE = 72, //!< System file not usable
CRM_EX_CANTCREAT = 73, //!< File couldn't be created
CRM_EX_IOERR = 74, //!< File I/O error
CRM_EX_TEMPFAIL = 75, //!< Try again
CRM_EX_PROTOCOL = 76, //!< Protocol violated
CRM_EX_NOPERM = 77, //!< Non-file permission issue
CRM_EX_CONFIG = 78, //!< Misconfiguration
// Custom
CRM_EX_FATAL = 100, //!< Do not respawn
CRM_EX_PANIC = 101, //!< Panic the local host
CRM_EX_DISCONNECT = 102, //!< Lost connection to something
CRM_EX_OLD = 103, //!< Update older than existing config
CRM_EX_DIGEST = 104, //!< Digest comparison failed
CRM_EX_NOSUCH = 105, //!< Requested item does not exist
CRM_EX_QUORUM = 106, //!< Local partition does not have quorum
CRM_EX_UNSAFE = 107, //!< Requires --force or new conditions
CRM_EX_EXISTS = 108, //!< Requested item already exists
CRM_EX_MULTIPLE = 109, //!< Requested item has multiple matches
CRM_EX_EXPIRED = 110, //!< Requested item has expired
CRM_EX_NOT_YET_IN_EFFECT = 111, //!< Requested item is not in effect
CRM_EX_INDETERMINATE = 112, //!< Could not determine status
CRM_EX_UNSATISFIED = 113, //!< Requested item does not satisfy constraints
CRM_EX_NO_DC = 114, //!< DC is not yet elected, e.g. right after cluster restart
// Other
CRM_EX_TIMEOUT = 124, //!< Convention from timeout(1)
/* Anything above 128 overlaps with some shells' use of these values for
* "interrupted by signal N", and so may be unreliable when detected by
* shell scripts.
*/
// OCF Resource Agent API 1.1
CRM_EX_DEGRADED = 190, //!< Service active but more likely to fail soon
CRM_EX_DEGRADED_PROMOTED = 191, //!< Service promoted but more likely to fail soon
/* Custom
*
* This can be used to initialize exit status variables or to indicate that
* a command is pending (which is what the controller uses it for).
*/
CRM_EX_NONE = 193, //!< No exit status available
CRM_EX_MAX = 255, //!< Ensure crm_exit_t can hold this
} crm_exit_t;
/*!
* \brief Execution status
*
* These codes are used to specify the result of the attempt to execute an
* agent, rather than the agent's result itself.
*/
enum pcmk_exec_status {
PCMK_EXEC_UNKNOWN = -2, //!< Used only to initialize variables
PCMK_EXEC_PENDING = -1, //!< Action is in progress
PCMK_EXEC_DONE, //!< Action completed, result is known
PCMK_EXEC_CANCELLED, //!< Action was cancelled
PCMK_EXEC_TIMEOUT, //!< Action did not complete in time
PCMK_EXEC_NOT_SUPPORTED, //!< Agent does not implement requested action
PCMK_EXEC_ERROR, //!< Execution failed, may be retried
PCMK_EXEC_ERROR_HARD, //!< Execution failed, do not retry on node
PCMK_EXEC_ERROR_FATAL, //!< Execution failed, do not retry anywhere
PCMK_EXEC_NOT_INSTALLED, //!< Agent or dependency not available locally
PCMK_EXEC_NOT_CONNECTED, //!< No connection to executor
PCMK_EXEC_INVALID, //!< Action cannot be attempted (e.g. shutdown)
PCMK_EXEC_NO_FENCE_DEVICE, //!< No fence device is configured for target
PCMK_EXEC_NO_SECRETS, //!< Necessary CIB secrets are unavailable
// Add new values above here then update this one below
PCMK_EXEC_MAX = PCMK_EXEC_NO_SECRETS, //!< Maximum value for this enum
};
/*!
* \brief Types of Pacemaker result codes
*
* A particular integer can have different meanings within different Pacemaker
* result code families. It may be interpretable within zero, one, or multiple
* families.
*
* These values are useful for specifying how an integer result code should be
* interpreted in situations involving a generic integer value. For example, a
* function that can process multiple types of result codes might accept an
* arbitrary integer argument along with a \p pcmk_result_type argument that
* specifies how to interpret the integer.
*/
enum pcmk_result_type {
pcmk_result_legacy = 0, //!< Legacy API function return code
pcmk_result_rc = 1, //!< Standard Pacemaker return code
pcmk_result_exitcode = 2, //!< Exit status code
pcmk_result_exec_status = 3, //!< Execution status
};
int pcmk_result_get_strings(int code, enum pcmk_result_type type,
const char **name, const char **desc);
const char *pcmk_rc_name(int rc);
// NOTE: sbd (as of at least 1.5.2) uses this
const char *pcmk_rc_str(int rc);
crm_exit_t pcmk_rc2exitc(int rc);
enum ocf_exitcode pcmk_rc2ocf(int rc);
int pcmk_rc2legacy(int rc);
int pcmk_legacy2rc(int legacy_rc);
// NOTE: sbd (as of at least 1.5.2) uses this
const char *pcmk_strerror(int rc);
const char *pcmk_errorname(int rc);
const char *crm_exit_name(crm_exit_t exit_code);
// NOTE: sbd (as of at least 1.5.2) uses this
const char *crm_exit_str(crm_exit_t exit_code);
_Noreturn crm_exit_t crm_exit(crm_exit_t rc);
/* coverity[+kill] */
void crm_abort(const char *file, const char *function, int line,
const char *condition, gboolean do_core, gboolean do_fork);
static inline const char *
pcmk_exec_status_str(enum pcmk_exec_status status)
{
switch (status) {
case PCMK_EXEC_PENDING: return "Pending";
case PCMK_EXEC_DONE: return "Done";
case PCMK_EXEC_CANCELLED: return "Cancelled";
case PCMK_EXEC_TIMEOUT: return "Timed out";
case PCMK_EXEC_NOT_SUPPORTED: return "Unsupported";
case PCMK_EXEC_ERROR: return "Error";
case PCMK_EXEC_ERROR_HARD: return "Hard error";
case PCMK_EXEC_ERROR_FATAL: return "Fatal error";
case PCMK_EXEC_NOT_INSTALLED: return "Not installed";
case PCMK_EXEC_NOT_CONNECTED: return "Internal communication failure";
case PCMK_EXEC_INVALID: return "Cannot execute now";
case PCMK_EXEC_NO_FENCE_DEVICE: return "No fence device";
case PCMK_EXEC_NO_SECRETS: return "CIB secrets unavailable";
default: return "Unrecognized status (bug?)";
}
}
#ifdef __cplusplus
}
#endif
#if !defined(PCMK_ALLOW_DEPRECATED) || (PCMK_ALLOW_DEPRECATED == 1)
#include
#endif
#endif
diff --git a/lib/common/crmcommon_private.h b/lib/common/crmcommon_private.h
index 2a45e41c41..a2c04f9cfc 100644
--- a/lib/common/crmcommon_private.h
+++ b/lib/common/crmcommon_private.h
@@ -1,478 +1,479 @@
/*
* Copyright 2018-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#ifndef PCMK__COMMON_CRMCOMMON_PRIVATE__H
#define PCMK__COMMON_CRMCOMMON_PRIVATE__H
/* This header is for the sole use of libcrmcommon, so that functions can be
* declared with G_GNUC_INTERNAL for efficiency.
*/
#include // uint8_t, uint32_t
#include // bool
#include // size_t
#include // G_GNUC_INTERNAL, G_GNUC_PRINTF, gchar, etc.
#include // xmlNode, xmlAttr
#include // xmlChar
#include // struct qb_ipc_response_header
#include // pcmk_ipc_api_t, crm_ipc_t, etc.
#include // crm_time_t
#include // LOG_NEVER
#include // mainloop_io_t
#include // pcmk__output_t
#include // crm_exit_t
#include // pcmk_rule_input_t
#include // enum pcmk__xml_flags
#ifdef __cplusplus
extern "C" {
#endif
// Decent chunk size for processing large amounts of data
#define PCMK__BUFFER_SIZE 4096
#if defined(PCMK__UNIT_TESTING)
#undef G_GNUC_INTERNAL
#define G_GNUC_INTERNAL
#endif
/*!
* \internal
* \brief Information about an XML node that was deleted
*
* When change tracking is enabled and we delete an XML node using
* \c pcmk__xml_free(), we free it and add its path and position to a list in
* its document's private data. This allows us to display changes, generate
* patchsets, etc.
*
* Note that this does not happen when deleting an XML attribute using
* \c pcmk__xa_remove(). In that case:
* * If \c force is \c true, we remove the attribute without any tracking.
* * If \c force is \c false, we mark the attribute as deleted but leave it in
* place until we commit changes.
*/
typedef struct pcmk__deleted_xml_s {
gchar *path; //!< XPath expression identifying the deleted node
int position; //!< Position of the deleted node among its siblings
} pcmk__deleted_xml_t;
/*!
* \internal
* \brief Private data for an XML node
*/
typedef struct xml_node_private_s {
uint32_t check; //!< Magic number for checking integrity
uint32_t flags; //!< Group of enum pcmk__xml_flags
xmlNode *match; //!< Pointer to matching node (defined by caller)
} xml_node_private_t;
/*!
* \internal
* \brief Private data for an XML document
*/
typedef struct xml_doc_private_s {
uint32_t check; //!< Magic number for checking integrity
uint32_t flags; //!< Group of enum pcmk__xml_flags
char *acl_user; //!< User affected by \c acls (for logging)
//! ACLs to check requested changes against (list of \c xml_acl_t)
GList *acls;
//! XML nodes marked as deleted (list of \c pcmk__deleted_xml_t)
GList *deleted_objs;
} xml_doc_private_t;
// XML private data magic numbers
#define PCMK__XML_DOC_PRIVATE_MAGIC 0x81726354UL
#define PCMK__XML_NODE_PRIVATE_MAGIC 0x54637281UL
// XML entity references
#define PCMK__XML_ENTITY_AMP "&"
#define PCMK__XML_ENTITY_GT ">"
#define PCMK__XML_ENTITY_LT "<"
#define PCMK__XML_ENTITY_QUOT """
#define pcmk__set_xml_flags(xml_priv, flags_to_set) do { \
(xml_priv)->flags = pcmk__set_flags_as(__func__, __LINE__, \
LOG_NEVER, "XML", "XML node", (xml_priv)->flags, \
(flags_to_set), #flags_to_set); \
} while (0)
#define pcmk__clear_xml_flags(xml_priv, flags_to_clear) do { \
(xml_priv)->flags = pcmk__clear_flags_as(__func__, __LINE__, \
LOG_NEVER, "XML", "XML node", (xml_priv)->flags, \
(flags_to_clear), #flags_to_clear); \
} while (0)
G_GNUC_INTERNAL
const char *pcmk__xml_element_type_text(xmlElementType type);
G_GNUC_INTERNAL
bool pcmk__xml_reset_node_flags(xmlNode *xml, void *user_data);
G_GNUC_INTERNAL
void pcmk__xml_set_parent_flags(xmlNode *xml, uint64_t flags);
G_GNUC_INTERNAL
void pcmk__xml_new_private_data(xmlNode *xml);
G_GNUC_INTERNAL
void pcmk__xml_free_private_data(xmlNode *xml);
G_GNUC_INTERNAL
void pcmk__xml_free_node(xmlNode *xml);
G_GNUC_INTERNAL
xmlDoc *pcmk__xml_new_doc(void);
G_GNUC_INTERNAL
int pcmk__xml_position(const xmlNode *xml, enum pcmk__xml_flags ignore_if_set);
G_GNUC_INTERNAL
bool pcmk__xc_matches(const xmlNode *comment1, const xmlNode *comment2);
G_GNUC_INTERNAL
void pcmk__xc_update(xmlNode *parent, xmlNode *target, xmlNode *update);
G_GNUC_INTERNAL
void pcmk__free_acls(GList *acls);
G_GNUC_INTERNAL
void pcmk__unpack_acl(xmlNode *source, xmlNode *target, const char *user);
G_GNUC_INTERNAL
bool pcmk__is_user_in_group(const char *user, const char *group);
G_GNUC_INTERNAL
void pcmk__apply_acl(xmlNode *xml);
G_GNUC_INTERNAL
void pcmk__apply_creation_acl(xmlNode *xml, bool check_top);
G_GNUC_INTERNAL
int pcmk__xa_remove(xmlAttr *attr, bool force);
G_GNUC_INTERNAL
void pcmk__mark_xml_attr_dirty(xmlAttr *a);
G_GNUC_INTERNAL
bool pcmk__xa_filterable(const char *name);
G_GNUC_INTERNAL
void pcmk__log_xmllib_err(void *ctx, const char *fmt, ...)
G_GNUC_PRINTF(2, 3);
G_GNUC_INTERNAL
void pcmk__mark_xml_node_dirty(xmlNode *xml);
G_GNUC_INTERNAL
bool pcmk__marked_as_deleted(xmlAttrPtr a, void *user_data);
G_GNUC_INTERNAL
void pcmk__dump_xml_attr(const xmlAttr *attr, GString *buffer);
G_GNUC_INTERNAL
int pcmk__xe_set_score(xmlNode *target, const char *name, const char *value);
G_GNUC_INTERNAL
bool pcmk__xml_is_name_start_char(const char *utf8, int *len);
G_GNUC_INTERNAL
bool pcmk__xml_is_name_char(const char *utf8, int *len);
/*
* Date/times
*/
// For use with pcmk__add_time_from_xml()
enum pcmk__time_component {
pcmk__time_unknown,
pcmk__time_years,
pcmk__time_months,
pcmk__time_weeks,
pcmk__time_days,
pcmk__time_hours,
pcmk__time_minutes,
pcmk__time_seconds,
};
G_GNUC_INTERNAL
const char *pcmk__time_component_attr(enum pcmk__time_component component);
G_GNUC_INTERNAL
int pcmk__add_time_from_xml(crm_time_t *t, enum pcmk__time_component component,
const xmlNode *xml);
G_GNUC_INTERNAL
void pcmk__set_time_if_earlier(crm_time_t *target, const crm_time_t *source);
/*
* IPC
*/
#define PCMK__IPC_VERSION 1
#define PCMK__CONTROLD_API_MAJOR "1"
#define PCMK__CONTROLD_API_MINOR "0"
// IPC behavior that varies by daemon
typedef struct pcmk__ipc_methods_s {
/*!
* \internal
* \brief Allocate any private data needed by daemon IPC
*
* \param[in,out] api IPC API connection
*
* \return Standard Pacemaker return code
*/
int (*new_data)(pcmk_ipc_api_t *api);
/*!
* \internal
* \brief Free any private data used by daemon IPC
*
* \param[in,out] api_data Data allocated by new_data() method
*/
void (*free_data)(void *api_data);
/*!
* \internal
* \brief Perform daemon-specific handling after successful connection
*
* Some daemons require clients to register before sending any other
* commands. The controller requires a CRM_OP_HELLO (with no reply), and
* the CIB manager, executor, and fencer require a CRM_OP_REGISTER (with a
* reply). Ideally this would be consistent across all daemons, but for now
* this allows each to do its own authorization.
*
* \param[in,out] api IPC API connection
*
* \return Standard Pacemaker return code
*/
int (*post_connect)(pcmk_ipc_api_t *api);
/*!
* \internal
* \brief Check whether an IPC request results in a reply
*
* \param[in,out] api IPC API connection
* \param[in] request IPC request XML
*
* \return true if request would result in an IPC reply, false otherwise
*/
bool (*reply_expected)(pcmk_ipc_api_t *api, const xmlNode *request);
/*!
* \internal
* \brief Perform daemon-specific handling of an IPC message
*
* \param[in,out] api IPC API connection
* \param[in,out] msg Message read from IPC connection
*
* \return true if more IPC reply messages should be expected
*/
bool (*dispatch)(pcmk_ipc_api_t *api, xmlNode *msg);
/*!
* \internal
* \brief Perform daemon-specific handling of an IPC disconnect
*
* \param[in,out] api IPC API connection
*/
void (*post_disconnect)(pcmk_ipc_api_t *api);
} pcmk__ipc_methods_t;
// Implementation of pcmk_ipc_api_t
struct pcmk_ipc_api_s {
enum pcmk_ipc_server server; // Daemon this IPC API instance is for
enum pcmk_ipc_dispatch dispatch_type; // How replies should be dispatched
crm_ipc_t *ipc; // IPC connection
mainloop_io_t *mainloop_io; // If using mainloop, I/O source for IPC
bool free_on_disconnect; // Whether disconnect should free object
pcmk_ipc_callback_t cb; // Caller-registered callback (if any)
void *user_data; // Caller-registered data (if any)
void *api_data; // For daemon-specific use
pcmk__ipc_methods_t *cmds; // Behavior that varies by daemon
};
typedef struct pcmk__ipc_header_s {
struct qb_ipc_response_header qb;
uint32_t size;
uint32_t flags;
uint8_t version;
+ uint16_t part_id; // If this is a multipart message, which part is this?
} pcmk__ipc_header_t;
G_GNUC_INTERNAL
int pcmk__send_ipc_request(pcmk_ipc_api_t *api, const xmlNode *request);
G_GNUC_INTERNAL
void pcmk__call_ipc_callback(pcmk_ipc_api_t *api,
enum pcmk_ipc_event event_type,
crm_exit_t status, void *event_data);
G_GNUC_INTERNAL
bool pcmk__valid_ipc_header(const pcmk__ipc_header_t *header);
G_GNUC_INTERNAL
pcmk__ipc_methods_t *pcmk__attrd_api_methods(void);
G_GNUC_INTERNAL
pcmk__ipc_methods_t *pcmk__controld_api_methods(void);
G_GNUC_INTERNAL
pcmk__ipc_methods_t *pcmk__pacemakerd_api_methods(void);
G_GNUC_INTERNAL
pcmk__ipc_methods_t *pcmk__schedulerd_api_methods(void);
/*
* Logging
*/
//! XML is newly created
#define PCMK__XML_PREFIX_CREATED "++"
//! XML has been deleted
#define PCMK__XML_PREFIX_DELETED "--"
//! XML has been modified
#define PCMK__XML_PREFIX_MODIFIED "+ "
//! XML has been moved
#define PCMK__XML_PREFIX_MOVED "+~"
/*
* Output
*/
G_GNUC_INTERNAL
int pcmk__bare_output_new(pcmk__output_t **out, const char *fmt_name,
const char *filename, char **argv);
G_GNUC_INTERNAL
void pcmk__register_option_messages(pcmk__output_t *out);
G_GNUC_INTERNAL
void pcmk__register_patchset_messages(pcmk__output_t *out);
G_GNUC_INTERNAL
bool pcmk__output_text_get_fancy(pcmk__output_t *out);
/*
* Rules
*/
// How node attribute values may be compared in rules
enum pcmk__comparison {
pcmk__comparison_unknown,
pcmk__comparison_defined,
pcmk__comparison_undefined,
pcmk__comparison_eq,
pcmk__comparison_ne,
pcmk__comparison_lt,
pcmk__comparison_lte,
pcmk__comparison_gt,
pcmk__comparison_gte,
};
// How node attribute values may be parsed in rules
enum pcmk__type {
pcmk__type_unknown,
pcmk__type_string,
pcmk__type_integer,
pcmk__type_number,
pcmk__type_version,
};
// Where to obtain reference value for a node attribute comparison
enum pcmk__reference_source {
pcmk__source_unknown,
pcmk__source_literal,
pcmk__source_instance_attrs,
pcmk__source_meta_attrs,
};
G_GNUC_INTERNAL
enum pcmk__comparison pcmk__parse_comparison(const char *op);
G_GNUC_INTERNAL
enum pcmk__type pcmk__parse_type(const char *type, enum pcmk__comparison op,
const char *value1, const char *value2);
G_GNUC_INTERNAL
enum pcmk__reference_source pcmk__parse_source(const char *source);
G_GNUC_INTERNAL
int pcmk__cmp_by_type(const char *value1, const char *value2,
enum pcmk__type type);
G_GNUC_INTERNAL
int pcmk__unpack_duration(const xmlNode *duration, const crm_time_t *start,
crm_time_t **end);
G_GNUC_INTERNAL
int pcmk__evaluate_date_spec(const xmlNode *date_spec, const crm_time_t *now);
G_GNUC_INTERNAL
int pcmk__evaluate_attr_expression(const xmlNode *expression,
const pcmk_rule_input_t *rule_input);
G_GNUC_INTERNAL
int pcmk__evaluate_rsc_expression(const xmlNode *expr,
const pcmk_rule_input_t *rule_input);
G_GNUC_INTERNAL
int pcmk__evaluate_op_expression(const xmlNode *expr,
const pcmk_rule_input_t *rule_input);
/*
* Utils
*/
#define PCMK__PW_BUFFER_LEN 500
/*
* Schemas
*/
typedef struct {
unsigned char v[2];
} pcmk__schema_version_t;
enum pcmk__schema_validator {
pcmk__schema_validator_none,
pcmk__schema_validator_rng
};
typedef struct {
int schema_index;
char *name;
/*!
* List of XSLT stylesheets for upgrading from this schema version to the
* next one. Sorted by the order in which they should be applied to the CIB.
*/
GList *transforms;
void *cache;
enum pcmk__schema_validator validator;
pcmk__schema_version_t version;
} pcmk__schema_t;
G_GNUC_INTERNAL
GList *pcmk__find_x_0_schema(void);
#ifdef __cplusplus
}
#endif
#endif // PCMK__COMMON_CRMCOMMON_PRIVATE__H
diff --git a/lib/common/ipc_client.c b/lib/common/ipc_client.c
index 1f65ace114..12a85ff607 100644
--- a/lib/common/ipc_client.c
+++ b/lib/common/ipc_client.c
@@ -1,1610 +1,1680 @@
/*
* Copyright 2004-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#include
#if defined(HAVE_UCRED) || defined(HAVE_SOCKPEERCRED)
#include
#elif defined(HAVE_GETPEERUCRED)
#include
#endif
#include
#include
#include
#include
#include /* indirectly: pcmk_err_generic */
#include
#include
#include
#include "crmcommon_private.h"
static int is_ipc_provider_expected(qb_ipcc_connection_t *qb_ipc, int sock,
uid_t refuid, gid_t refgid, pid_t *gotpid,
uid_t *gotuid, gid_t *gotgid);
/*!
* \brief Create a new object for using Pacemaker daemon IPC
*
* \param[out] api Where to store new IPC object
* \param[in] server Which Pacemaker daemon the object is for
*
* \return Standard Pacemaker result code
*
* \note The caller is responsible for freeing *api using pcmk_free_ipc_api().
* \note This is intended to supersede crm_ipc_new() but currently only
* supports the controller, pacemakerd, and schedulerd IPC API.
*/
int
pcmk_new_ipc_api(pcmk_ipc_api_t **api, enum pcmk_ipc_server server)
{
if (api == NULL) {
return EINVAL;
}
*api = calloc(1, sizeof(pcmk_ipc_api_t));
if (*api == NULL) {
return errno;
}
(*api)->server = server;
if (pcmk_ipc_name(*api, false) == NULL) {
pcmk_free_ipc_api(*api);
*api = NULL;
return EOPNOTSUPP;
}
// Set server methods
switch (server) {
case pcmk_ipc_attrd:
(*api)->cmds = pcmk__attrd_api_methods();
break;
case pcmk_ipc_based:
break;
case pcmk_ipc_controld:
(*api)->cmds = pcmk__controld_api_methods();
break;
case pcmk_ipc_execd:
break;
case pcmk_ipc_fenced:
break;
case pcmk_ipc_pacemakerd:
(*api)->cmds = pcmk__pacemakerd_api_methods();
break;
case pcmk_ipc_schedulerd:
(*api)->cmds = pcmk__schedulerd_api_methods();
break;
default: // pcmk_ipc_unknown
pcmk_free_ipc_api(*api);
*api = NULL;
return EINVAL;
}
if ((*api)->cmds == NULL) {
pcmk_free_ipc_api(*api);
*api = NULL;
return ENOMEM;
}
(*api)->ipc = crm_ipc_new(pcmk_ipc_name(*api, false), 0);
if ((*api)->ipc == NULL) {
pcmk_free_ipc_api(*api);
*api = NULL;
return ENOMEM;
}
// If daemon API has its own data to track, allocate it
if ((*api)->cmds->new_data != NULL) {
if ((*api)->cmds->new_data(*api) != pcmk_rc_ok) {
pcmk_free_ipc_api(*api);
*api = NULL;
return ENOMEM;
}
}
crm_trace("Created %s API IPC object", pcmk_ipc_name(*api, true));
return pcmk_rc_ok;
}
static void
free_daemon_specific_data(pcmk_ipc_api_t *api)
{
if ((api != NULL) && (api->cmds != NULL)) {
if ((api->cmds->free_data != NULL) && (api->api_data != NULL)) {
api->cmds->free_data(api->api_data);
api->api_data = NULL;
}
free(api->cmds);
api->cmds = NULL;
}
}
/*!
* \internal
* \brief Call an IPC API event callback, if one is registed
*
* \param[in,out] api IPC API connection
* \param[in] event_type The type of event that occurred
* \param[in] status Event status
* \param[in,out] event_data Event-specific data
*/
void
pcmk__call_ipc_callback(pcmk_ipc_api_t *api, enum pcmk_ipc_event event_type,
crm_exit_t status, void *event_data)
{
if ((api != NULL) && (api->cb != NULL)) {
api->cb(api, event_type, status, event_data, api->user_data);
}
}
/*!
* \internal
* \brief Clean up after an IPC disconnect
*
* \param[in,out] user_data IPC API connection that disconnected
*
* \note This function can be used as a main loop IPC destroy callback.
*/
static void
ipc_post_disconnect(gpointer user_data)
{
pcmk_ipc_api_t *api = user_data;
crm_info("Disconnected from %s", pcmk_ipc_name(api, true));
// Perform any daemon-specific handling needed
if ((api->cmds != NULL) && (api->cmds->post_disconnect != NULL)) {
api->cmds->post_disconnect(api);
}
// Call client's registered event callback
pcmk__call_ipc_callback(api, pcmk_ipc_event_disconnect, CRM_EX_DISCONNECT,
NULL);
/* If this is being called from a running main loop, mainloop_gio_destroy()
* will free ipc and mainloop_io immediately after calling this function.
* If this is called from a stopped main loop, these will leak, so the best
* practice is to close the connection before stopping the main loop.
*/
api->ipc = NULL;
api->mainloop_io = NULL;
if (api->free_on_disconnect) {
/* pcmk_free_ipc_api() has already been called, but did not free api
* or api->cmds because this function needed them. Do that now.
*/
free_daemon_specific_data(api);
crm_trace("Freeing IPC API object after disconnect");
free(api);
}
}
/*!
* \brief Free the contents of an IPC API object
*
* \param[in,out] api IPC API object to free
*/
void
pcmk_free_ipc_api(pcmk_ipc_api_t *api)
{
bool free_on_disconnect = false;
if (api == NULL) {
return;
}
crm_debug("Releasing %s IPC API", pcmk_ipc_name(api, true));
if (api->ipc != NULL) {
if (api->mainloop_io != NULL) {
/* We need to keep the api pointer itself around, because it is the
* user data for the IPC client destroy callback. That will be
* triggered by the pcmk_disconnect_ipc() call below, but it might
* happen later in the main loop (if still running).
*
* This flag tells the destroy callback to free the object. It can't
* do that unconditionally, because the application might call this
* function after a disconnect that happened by other means.
*/
free_on_disconnect = api->free_on_disconnect = true;
}
pcmk_disconnect_ipc(api); // Frees api if free_on_disconnect is true
}
if (!free_on_disconnect) {
free_daemon_specific_data(api);
crm_trace("Freeing IPC API object");
free(api);
}
}
/*!
* \brief Get the IPC name used with an IPC API connection
*
* \param[in] api IPC API connection
* \param[in] for_log If true, return human-friendly name instead of IPC name
*
* \return IPC API's human-friendly or connection name, or if none is available,
* "Pacemaker" if for_log is true and NULL if for_log is false
*/
const char *
pcmk_ipc_name(const pcmk_ipc_api_t *api, bool for_log)
{
if (api == NULL) {
return for_log? "Pacemaker" : NULL;
}
if (for_log) {
const char *name = pcmk__server_log_name(api->server);
return pcmk__s(name, "Pacemaker");
}
switch (api->server) {
// These servers do not have pcmk_ipc_api_t implementations yet
case pcmk_ipc_based:
case pcmk_ipc_execd:
case pcmk_ipc_fenced:
return NULL;
default:
return pcmk__server_ipc_name(api->server);
}
}
/*!
* \brief Check whether an IPC API connection is active
*
* \param[in,out] api IPC API connection
*
* \return true if IPC is connected, false otherwise
*/
bool
pcmk_ipc_is_connected(pcmk_ipc_api_t *api)
{
return (api != NULL) && crm_ipc_connected(api->ipc);
}
/*!
* \internal
* \brief Call the daemon-specific API's dispatch function
*
* Perform daemon-specific handling of IPC reply dispatch. It is the daemon
* method's responsibility to call the client's registered event callback, as
* well as allocate and free any event data.
*
* \param[in,out] api IPC API connection
* \param[in,out] message IPC reply XML to dispatch
*/
static bool
call_api_dispatch(pcmk_ipc_api_t *api, xmlNode *message)
{
crm_log_xml_trace(message, "ipc-received");
if ((api->cmds != NULL) && (api->cmds->dispatch != NULL)) {
return api->cmds->dispatch(api, message);
}
return false;
}
/*!
* \internal
* \brief Dispatch previously read IPC data
*
* \param[in] buffer Data read from IPC
* \param[in,out] api IPC object
*
* \return Standard Pacemaker return code. In particular:
*
* pcmk_rc_ok: There are no more messages expected from the server. Quit
* reading.
* EINPROGRESS: There are more messages expected from the server. Keep reading.
*
* All other values indicate an error.
*/
static int
dispatch_ipc_data(const char *buffer, pcmk_ipc_api_t *api)
{
bool more = false;
xmlNode *msg;
if (buffer == NULL) {
crm_warn("Empty message received from %s IPC",
pcmk_ipc_name(api, true));
return ENOMSG;
}
msg = pcmk__xml_parse(buffer);
if (msg == NULL) {
crm_warn("Malformed message received from %s IPC",
pcmk_ipc_name(api, true));
return EPROTO;
}
more = call_api_dispatch(api, msg);
pcmk__xml_free(msg);
if (more) {
return EINPROGRESS;
} else {
return pcmk_rc_ok;
}
}
/*!
* \internal
* \brief Dispatch data read from IPC source
*
* \param[in] buffer Data read from IPC
* \param[in] length Number of bytes of data in buffer (ignored)
* \param[in,out] user_data IPC object
*
* \return Always 0 (meaning connection is still required)
*
* \note This function can be used as a main loop IPC dispatch callback.
*/
static int
dispatch_ipc_source_data(const char *buffer, ssize_t length, gpointer user_data)
{
pcmk_ipc_api_t *api = user_data;
CRM_CHECK(api != NULL, return 0);
dispatch_ipc_data(buffer, api);
return 0;
}
/*!
* \brief Check whether an IPC connection has data available (without main loop)
*
* \param[in] api IPC API connection
* \param[in] timeout_ms If less than 0, poll indefinitely; if 0, poll once
* and return immediately; otherwise, poll for up to
* this many milliseconds
*
* \return Standard Pacemaker return code
*
* \note Callers of pcmk_connect_ipc() using pcmk_ipc_dispatch_poll should call
* this function to check whether IPC data is available. Return values of
* interest include pcmk_rc_ok meaning data is available, and EAGAIN
* meaning no data is available; all other values indicate errors.
* \todo This does not allow the caller to poll multiple file descriptors at
* once. If there is demand for that, we could add a wrapper for
* pcmk__ipc_fd(api->ipc), so the caller can call poll() themselves.
*/
int
pcmk_poll_ipc(const pcmk_ipc_api_t *api, int timeout_ms)
{
int rc;
struct pollfd pollfd = { 0, };
if ((api == NULL) || (api->dispatch_type != pcmk_ipc_dispatch_poll)) {
return EINVAL;
}
rc = pcmk__ipc_fd(api->ipc, &(pollfd.fd));
if (rc != pcmk_rc_ok) {
crm_debug("Could not obtain file descriptor for %s IPC: %s",
pcmk_ipc_name(api, true), pcmk_rc_str(rc));
return rc;
}
pollfd.events = POLLIN;
rc = poll(&pollfd, 1, timeout_ms);
if (rc < 0) {
/* Some UNIX systems return negative and set EAGAIN for failure to
* allocate memory; standardize the return code in that case
*/
return (errno == EAGAIN)? ENOMEM : errno;
} else if (rc == 0) {
return EAGAIN;
}
return pcmk_rc_ok;
}
/*!
* \brief Dispatch available messages on an IPC connection (without main loop)
*
* \param[in,out] api IPC API connection
*
* \return Standard Pacemaker return code
*
* \note Callers of pcmk_connect_ipc() using pcmk_ipc_dispatch_poll should call
* this function when IPC data is available.
*/
void
pcmk_dispatch_ipc(pcmk_ipc_api_t *api)
{
if (api == NULL) {
return;
}
while (crm_ipc_ready(api->ipc) > 0) {
if (crm_ipc_read(api->ipc) > 0) {
dispatch_ipc_data(crm_ipc_buffer(api->ipc), api);
+ pcmk__ipc_free_client_buffer(api->ipc);
}
}
}
// \return Standard Pacemaker return code
static int
connect_with_main_loop(pcmk_ipc_api_t *api)
{
int rc;
struct ipc_client_callbacks callbacks = {
.dispatch = dispatch_ipc_source_data,
.destroy = ipc_post_disconnect,
};
rc = pcmk__add_mainloop_ipc(api->ipc, G_PRIORITY_DEFAULT, api,
&callbacks, &(api->mainloop_io));
if (rc != pcmk_rc_ok) {
return rc;
}
crm_debug("Connected to %s IPC (attached to main loop)",
pcmk_ipc_name(api, true));
/* After this point, api->mainloop_io owns api->ipc, so api->ipc
* should not be explicitly freed.
*/
return pcmk_rc_ok;
}
// \return Standard Pacemaker return code
static int
connect_without_main_loop(pcmk_ipc_api_t *api)
{
int rc = pcmk__connect_generic_ipc(api->ipc);
if (rc != pcmk_rc_ok) {
crm_ipc_close(api->ipc);
} else {
crm_debug("Connected to %s IPC (without main loop)",
pcmk_ipc_name(api, true));
}
return rc;
}
/*!
* \internal
* \brief Connect to a Pacemaker daemon via IPC (retrying after soft errors)
*
* \param[in,out] api IPC API instance
* \param[in] dispatch_type How IPC replies should be dispatched
* \param[in] attempts How many times to try (in case of soft error)
*
* \return Standard Pacemaker return code
*/
int
pcmk__connect_ipc(pcmk_ipc_api_t *api, enum pcmk_ipc_dispatch dispatch_type,
int attempts)
{
int rc = pcmk_rc_ok;
if ((api == NULL) || (attempts < 1)) {
return EINVAL;
}
if (api->ipc == NULL) {
api->ipc = crm_ipc_new(pcmk_ipc_name(api, false), 0);
if (api->ipc == NULL) {
return ENOMEM;
}
}
if (crm_ipc_connected(api->ipc)) {
crm_trace("Already connected to %s", pcmk_ipc_name(api, true));
return pcmk_rc_ok;
}
api->dispatch_type = dispatch_type;
crm_debug("Attempting connection to %s (up to %d time%s)",
pcmk_ipc_name(api, true), attempts, pcmk__plural_s(attempts));
for (int remaining = attempts - 1; remaining >= 0; --remaining) {
switch (dispatch_type) {
case pcmk_ipc_dispatch_main:
rc = connect_with_main_loop(api);
break;
case pcmk_ipc_dispatch_sync:
case pcmk_ipc_dispatch_poll:
rc = connect_without_main_loop(api);
break;
}
if ((remaining == 0) || ((rc != EAGAIN) && (rc != EALREADY))) {
break; // Result is final
}
// Retry after soft error (interrupted by signal, etc.)
pcmk__sleep_ms((attempts - remaining) * 500);
crm_debug("Re-attempting connection to %s (%d attempt%s remaining)",
pcmk_ipc_name(api, true), remaining,
pcmk__plural_s(remaining));
}
if (rc != pcmk_rc_ok) {
return rc;
}
if ((api->cmds != NULL) && (api->cmds->post_connect != NULL)) {
rc = api->cmds->post_connect(api);
if (rc != pcmk_rc_ok) {
crm_ipc_close(api->ipc);
}
}
return rc;
}
/*!
* \brief Connect to a Pacemaker daemon via IPC
*
* \param[in,out] api IPC API instance
* \param[in] dispatch_type How IPC replies should be dispatched
*
* \return Standard Pacemaker return code
*/
int
pcmk_connect_ipc(pcmk_ipc_api_t *api, enum pcmk_ipc_dispatch dispatch_type)
{
int rc = pcmk__connect_ipc(api, dispatch_type, 2);
if (rc != pcmk_rc_ok) {
crm_err("Connection to %s failed: %s",
pcmk_ipc_name(api, true), pcmk_rc_str(rc));
}
return rc;
}
/*!
* \brief Disconnect an IPC API instance
*
* \param[in,out] api IPC API connection
*
* \return Standard Pacemaker return code
*
* \note If the connection is attached to a main loop, this function should be
* called before quitting the main loop, to ensure that all memory is
* freed.
*/
void
pcmk_disconnect_ipc(pcmk_ipc_api_t *api)
{
if ((api == NULL) || (api->ipc == NULL)) {
return;
}
switch (api->dispatch_type) {
case pcmk_ipc_dispatch_main:
{
mainloop_io_t *mainloop_io = api->mainloop_io;
// Make sure no code with access to api can use these again
api->mainloop_io = NULL;
api->ipc = NULL;
mainloop_del_ipc_client(mainloop_io);
// After this point api might have already been freed
}
break;
case pcmk_ipc_dispatch_poll:
case pcmk_ipc_dispatch_sync:
{
crm_ipc_t *ipc = api->ipc;
// Make sure no code with access to api can use ipc again
api->ipc = NULL;
// This should always be the case already, but to be safe
api->free_on_disconnect = false;
crm_ipc_close(ipc);
crm_ipc_destroy(ipc);
ipc_post_disconnect(api);
}
break;
}
}
/*!
* \brief Register a callback for IPC API events
*
* \param[in,out] api IPC API connection
* \param[in] callback Callback to register
* \param[in] userdata Caller data to pass to callback
*
* \note This function may be called multiple times to update the callback
* and/or user data. The caller remains responsible for freeing
* userdata in any case (after the IPC is disconnected, if the
* user data is still registered with the IPC).
*/
void
pcmk_register_ipc_callback(pcmk_ipc_api_t *api, pcmk_ipc_callback_t cb,
void *user_data)
{
if (api == NULL) {
return;
}
api->cb = cb;
api->user_data = user_data;
}
/*!
* \internal
* \brief Send an XML request across an IPC API connection
*
* \param[in,out] api IPC API connection
* \param[in] request XML request to send
*
* \return Standard Pacemaker return code
*
* \note Daemon-specific IPC API functions should call this function to send
* requests, because it handles different dispatch types appropriately.
*/
int
pcmk__send_ipc_request(pcmk_ipc_api_t *api, const xmlNode *request)
{
int rc;
xmlNode *reply = NULL;
enum crm_ipc_flags flags = crm_ipc_flags_none;
if ((api == NULL) || (api->ipc == NULL) || (request == NULL)) {
return EINVAL;
}
crm_log_xml_trace(request, "ipc-sent");
// Synchronous dispatch requires waiting for a reply
if ((api->dispatch_type == pcmk_ipc_dispatch_sync)
&& (api->cmds != NULL)
&& (api->cmds->reply_expected != NULL)
&& (api->cmds->reply_expected(api, request))) {
flags = crm_ipc_client_response;
}
/* The 0 here means a default timeout of 5 seconds
*
* @TODO Maybe add a timeout_ms member to pcmk_ipc_api_t and a
* pcmk_set_ipc_timeout() setter for it, then use it here.
*/
rc = crm_ipc_send(api->ipc, request, flags, 0, &reply);
if (rc < 0) {
return pcmk_legacy2rc(rc);
} else if (rc == 0) {
return ENODATA;
}
// With synchronous dispatch, we dispatch any reply now
if (reply != NULL) {
bool more = call_api_dispatch(api, reply);
pcmk__xml_free(reply);
while (more) {
rc = crm_ipc_read(api->ipc);
if (rc == -EAGAIN) {
continue;
} else if (rc == -ENOMSG || rc == pcmk_ok) {
return pcmk_rc_ok;
} else if (rc < 0) {
return -rc;
}
rc = dispatch_ipc_data(crm_ipc_buffer(api->ipc), api);
+ pcmk__ipc_free_client_buffer(api->ipc);
if (rc == pcmk_rc_ok) {
more = false;
} else if (rc == EINPROGRESS) {
more = true;
} else {
continue;
}
}
}
return pcmk_rc_ok;
}
/*!
* \internal
* \brief Create the XML for an IPC request to purge a node from the peer cache
*
* \param[in] api IPC API connection
* \param[in] node_name If not NULL, name of node to purge
* \param[in] nodeid If not 0, node ID of node to purge
*
* \return Newly allocated IPC request XML
*
* \note The controller, fencer, and pacemakerd use the same request syntax, but
* the attribute manager uses a different one. The CIB manager doesn't
* have any syntax for it. The executor and scheduler don't connect to the
* cluster layer and thus don't have or need any syntax for it.
*
* \todo Modify the attribute manager to accept the common syntax (as well
* as its current one, for compatibility with older clients). Modify
* the CIB manager to accept and honor the common syntax. Modify the
* executor and scheduler to accept the syntax (immediately returning
* success), just for consistency. Modify this function to use the
* common syntax with all daemons if their version supports it.
*/
static xmlNode *
create_purge_node_request(const pcmk_ipc_api_t *api, const char *node_name,
uint32_t nodeid)
{
xmlNode *request = NULL;
const char *client = crm_system_name? crm_system_name : "client";
switch (api->server) {
case pcmk_ipc_attrd:
request = pcmk__xe_create(NULL, __func__);
crm_xml_add(request, PCMK__XA_T, PCMK__VALUE_ATTRD);
crm_xml_add(request, PCMK__XA_SRC, crm_system_name);
crm_xml_add(request, PCMK_XA_TASK, PCMK__ATTRD_CMD_PEER_REMOVE);
pcmk__xe_set_bool_attr(request, PCMK__XA_REAP, true);
crm_xml_add(request, PCMK__XA_ATTR_HOST, node_name);
if (nodeid > 0) {
crm_xml_add_int(request, PCMK__XA_ATTR_HOST_ID, nodeid);
}
break;
case pcmk_ipc_controld:
case pcmk_ipc_fenced:
case pcmk_ipc_pacemakerd:
request = pcmk__new_request(api->server, client, NULL,
pcmk_ipc_name(api, false),
CRM_OP_RM_NODE_CACHE, NULL);
if (nodeid > 0) {
crm_xml_add_ll(request, PCMK_XA_ID, (long long) nodeid);
}
crm_xml_add(request, PCMK_XA_UNAME, node_name);
break;
case pcmk_ipc_based:
case pcmk_ipc_execd:
case pcmk_ipc_schedulerd:
break;
default: // pcmk_ipc_unknown (shouldn't be possible)
return NULL;
}
return request;
}
/*!
* \brief Ask a Pacemaker daemon to purge a node from its peer cache
*
* \param[in,out] api IPC API connection
* \param[in] node_name If not NULL, name of node to purge
* \param[in] nodeid If not 0, node ID of node to purge
*
* \return Standard Pacemaker return code
*
* \note At least one of node_name or nodeid must be specified.
*/
int
pcmk_ipc_purge_node(pcmk_ipc_api_t *api, const char *node_name, uint32_t nodeid)
{
int rc = 0;
xmlNode *request = NULL;
if (api == NULL) {
return EINVAL;
}
if ((node_name == NULL) && (nodeid == 0)) {
return EINVAL;
}
request = create_purge_node_request(api, node_name, nodeid);
if (request == NULL) {
return EOPNOTSUPP;
}
rc = pcmk__send_ipc_request(api, request);
pcmk__xml_free(request);
crm_debug("%s peer cache purge of node %s[%lu]: rc=%d",
pcmk_ipc_name(api, true), node_name, (unsigned long) nodeid, rc);
return rc;
}
/*
* Generic IPC API (to eventually be deprecated as public API and made internal)
*/
struct crm_ipc_s {
struct pollfd pfd;
unsigned int buf_size; // size of allocated buffer
- int msg_size;
int need_reply;
- char *buffer;
+ GByteArray *buffer;
char *server_name; // server IPC name being connected to
qb_ipcc_connection_t *ipc;
};
/*!
* \brief Create a new (legacy) object for using Pacemaker daemon IPC
*
* \param[in] name IPC system name to connect to
* \param[in] max_size Use a maximum IPC buffer size of at least this size
*
* \return Newly allocated IPC object on success, NULL otherwise
*
* \note The caller is responsible for freeing the result using
* crm_ipc_destroy().
* \note This should be considered deprecated for use with daemons supported by
* pcmk_new_ipc_api().
* \note @COMPAT Since 3.0.1, \p max_size is ignored and the default given by
* \c crm_ipc_default_buffer_size() will be used instead.
*/
crm_ipc_t *
crm_ipc_new(const char *name, size_t max_size)
{
crm_ipc_t *client = NULL;
client = calloc(1, sizeof(crm_ipc_t));
if (client == NULL) {
crm_err("Could not create IPC connection: %s", strerror(errno));
return NULL;
}
client->server_name = strdup(name);
if (client->server_name == NULL) {
crm_err("Could not create %s IPC connection: %s",
name, strerror(errno));
free(client);
return NULL;
}
- client->buf_size = crm_ipc_default_buffer_size();
- client->buffer = malloc(client->buf_size);
- if (client->buffer == NULL) {
- crm_err("Could not create %s IPC connection: %s",
- name, strerror(errno));
- free(client->server_name);
- free(client);
- return NULL;
- }
+ client->buf_size = crm_ipc_default_buffer_size();
+ client->buffer = NULL;
client->pfd.fd = -1;
client->pfd.events = POLLIN;
client->pfd.revents = 0;
return client;
}
/*!
* \internal
* \brief Connect a generic (not daemon-specific) IPC object
*
* \param[in,out] ipc Generic IPC object to connect
*
* \return Standard Pacemaker return code
*/
int
pcmk__connect_generic_ipc(crm_ipc_t *ipc)
{
uid_t cl_uid = 0;
gid_t cl_gid = 0;
pid_t found_pid = 0;
uid_t found_uid = 0;
gid_t found_gid = 0;
int rc = pcmk_rc_ok;
if (ipc == NULL) {
return EINVAL;
}
ipc->need_reply = FALSE;
ipc->ipc = qb_ipcc_connect(ipc->server_name, ipc->buf_size);
if (ipc->ipc == NULL) {
return errno;
}
rc = qb_ipcc_fd_get(ipc->ipc, &ipc->pfd.fd);
if (rc < 0) { // -errno
crm_ipc_close(ipc);
return -rc;
}
rc = pcmk_daemon_user(&cl_uid, &cl_gid);
rc = pcmk_legacy2rc(rc);
if (rc != pcmk_rc_ok) {
crm_ipc_close(ipc);
return rc;
}
rc = is_ipc_provider_expected(ipc->ipc, ipc->pfd.fd, cl_uid, cl_gid,
&found_pid, &found_uid, &found_gid);
if (rc != pcmk_rc_ok) {
if (rc == pcmk_rc_ipc_unauthorized) {
crm_info("%s IPC provider authentication failed: process %lld has "
"uid %lld (expected %lld) and gid %lld (expected %lld)",
ipc->server_name,
(long long) PCMK__SPECIAL_PID_AS_0(found_pid),
(long long) found_uid, (long long) cl_uid,
(long long) found_gid, (long long) cl_gid);
}
crm_ipc_close(ipc);
return rc;
}
return pcmk_rc_ok;
}
void
crm_ipc_close(crm_ipc_t * client)
{
if (client) {
if (client->ipc) {
qb_ipcc_connection_t *ipc = client->ipc;
client->ipc = NULL;
qb_ipcc_disconnect(ipc);
}
}
}
void
crm_ipc_destroy(crm_ipc_t * client)
{
if (client) {
if (client->ipc && qb_ipcc_is_connected(client->ipc)) {
crm_notice("Destroying active %s IPC connection",
client->server_name);
/* The next line is basically unsafe
*
* If this connection was attached to mainloop and mainloop is active,
* the 'disconnected' callback will end up back here and we'll end
* up free'ing the memory twice - something that can still happen
* even without this if we destroy a connection and it closes before
* we call exit
*/
/* crm_ipc_close(client); */
} else {
crm_trace("Destroying inactive %s IPC connection",
client->server_name);
}
- free(client->buffer);
+
+ if (client->buffer != NULL) {
+ pcmk__ipc_free_client_buffer(client);
+ }
+
free(client->server_name);
free(client);
}
}
/*!
* \internal
* \brief Get the file descriptor for a generic IPC object
*
* \param[in,out] ipc Generic IPC object to get file descriptor for
* \param[out] fd Where to store file descriptor
*
* \return Standard Pacemaker return code
*/
int
pcmk__ipc_fd(crm_ipc_t *ipc, int *fd)
{
if ((ipc == NULL) || (fd == NULL)) {
return EINVAL;
}
if ((ipc->ipc == NULL) || (ipc->pfd.fd < 0)) {
return ENOTCONN;
}
*fd = ipc->pfd.fd;
return pcmk_rc_ok;
}
int
crm_ipc_get_fd(crm_ipc_t * client)
{
int fd = -1;
if (pcmk__ipc_fd(client, &fd) != pcmk_rc_ok) {
crm_err("Could not obtain file descriptor for %s IPC",
((client == NULL)? "unspecified" : client->server_name));
errno = EINVAL;
return -EINVAL;
}
return fd;
}
bool
crm_ipc_connected(crm_ipc_t * client)
{
bool rc = FALSE;
if (client == NULL) {
crm_trace("No client");
return FALSE;
} else if (client->ipc == NULL) {
crm_trace("No connection");
return FALSE;
} else if (client->pfd.fd < 0) {
crm_trace("Bad descriptor");
return FALSE;
}
rc = qb_ipcc_is_connected(client->ipc);
if (rc == FALSE) {
client->pfd.fd = -EINVAL;
}
return rc;
}
/*!
* \brief Check whether an IPC connection is ready to be read
*
* \param[in,out] client Connection to check
*
* \return Positive value if ready to be read, 0 if not ready, -errno on error
*/
int
crm_ipc_ready(crm_ipc_t *client)
{
int rc;
pcmk__assert(client != NULL);
if (!crm_ipc_connected(client)) {
return -ENOTCONN;
}
client->pfd.revents = 0;
rc = poll(&(client->pfd), 1, 0);
return (rc < 0)? -errno : rc;
}
long
-crm_ipc_read(crm_ipc_t * client)
+crm_ipc_read(crm_ipc_t *client)
{
- pcmk__ipc_header_t *header = NULL;
+ guint8 *buffer = NULL;
+ long rc = -ENOMSG;
+
+ pcmk__assert((client != NULL) && (client->ipc != NULL));
+ buffer = g_malloc0(crm_ipc_default_buffer_size());
+
+ do {
+ pcmk__ipc_header_t *header = NULL;
+ ssize_t bytes = qb_ipcc_event_recv(client->ipc, buffer,
+ crm_ipc_default_buffer_size(), 0);
+
+ header = (pcmk__ipc_header_t *)(void *) buffer;
+
+ if (bytes <= 0) {
+ crm_trace("No message received from %s IPC: %s",
+ client->server_name, strerror(-bytes));
+
+ if (!crm_ipc_connected(client) || bytes == -ENOTCONN) {
+ crm_err("Connection to %s IPC failed", client->server_name);
+ rc = -ENOTCONN;
+ pcmk__ipc_free_client_buffer(client);
+ } else if (bytes == -EAGAIN) {
+ rc = -EAGAIN;
+ }
- pcmk__assert((client != NULL) && (client->ipc != NULL)
- && (client->buffer != NULL));
+ goto done;
- client->buffer[0] = 0;
- client->msg_size = qb_ipcc_event_recv(client->ipc, client->buffer,
- client->buf_size, 0);
- if (client->msg_size >= 0) {
- header = (pcmk__ipc_header_t *)(void*)client->buffer;
- if (!pcmk__valid_ipc_header(header)) {
- return -EBADMSG;
+ } else if (bytes != header->size + sizeof(pcmk__ipc_header_t)) {
+ crm_trace("Message size does not match header");
+ rc = -EBADMSG;
+ pcmk__ipc_free_client_buffer(client);
+ goto done;
}
- crm_trace("Received %s IPC event %d size=%u rc=%d text='%.100s'",
+ crm_trace("Received %s IPC event %" PRId32 " size=%" PRIu32 " rc=%zu",
client->server_name, header->qb.id, header->qb.size,
- client->msg_size,
- client->buffer + sizeof(pcmk__ipc_header_t));
+ bytes);
- } else {
- crm_trace("No message received from %s IPC: %s",
- client->server_name, pcmk_strerror(client->msg_size));
+ rc = pcmk__ipc_msg_append(&client->buffer, buffer);
- if (client->msg_size == -EAGAIN) {
- return -EAGAIN;
+ if (rc == pcmk_rc_ok) {
+ break;
+ } else if (rc == pcmk_rc_ipc_more) {
+ continue;
+ } else {
+ pcmk__ipc_free_client_buffer(client);
+ rc = pcmk_rc2legacy(rc);
+ goto done;
}
- }
+ } while (true);
- if (!crm_ipc_connected(client) || client->msg_size == -ENOTCONN) {
- crm_err("Connection to %s IPC failed", client->server_name);
+ if (client->buffer->len > 0) {
+ /* Data length excluding the header */
+ rc = client->buffer->len - sizeof(pcmk__ipc_header_t);
}
- if (header) {
- /* Data excluding the header */
- return header->size;
+done:
+ g_free(buffer);
+ return rc;
+}
+
+void
+pcmk__ipc_free_client_buffer(crm_ipc_t *client)
+{
+ pcmk__assert(client != NULL);
+
+ if (client->buffer != NULL) {
+ g_byte_array_free(client->buffer, TRUE);
+ client->buffer = NULL;
}
- return -ENOMSG;
}
const char *
crm_ipc_buffer(crm_ipc_t * client)
{
pcmk__assert(client != NULL);
- return client->buffer + sizeof(pcmk__ipc_header_t);
+ CRM_CHECK(client->buffer != NULL, return NULL);
+ return (const char *) (client->buffer->data + sizeof(pcmk__ipc_header_t));
}
uint32_t
crm_ipc_buffer_flags(crm_ipc_t * client)
{
pcmk__ipc_header_t *header = NULL;
pcmk__assert(client != NULL);
if (client->buffer == NULL) {
return 0;
}
- header = (pcmk__ipc_header_t *)(void*)client->buffer;
+ header = (pcmk__ipc_header_t *)(void*) client->buffer->data;
return header->flags;
}
const char *
crm_ipc_name(crm_ipc_t * client)
{
pcmk__assert(client != NULL);
return client->server_name;
}
// \return Standard Pacemaker return code
static int
internal_ipc_get_reply(crm_ipc_t *client, int request_id, int ms_timeout,
ssize_t *bytes, xmlNode **reply)
{
+ guint8 *buffer = NULL;
pcmk__ipc_header_t *hdr = NULL;
time_t timeout = 0;
int32_t qb_timeout = -1;
int rc = pcmk_rc_ok;
if (ms_timeout > 0) {
timeout = time(NULL) + 1 + pcmk__timeout_ms2s(ms_timeout);
qb_timeout = 1000;
}
/* get the reply */
crm_trace("Expecting reply to %s IPC message %d", client->server_name,
request_id);
+ buffer = g_malloc0(client->buf_size);
+
do {
+ guint8 *data = NULL;
xmlNode *xml = NULL;
- *bytes = qb_ipcc_recv(client->ipc, client->buffer, client->buf_size,
+ *bytes = qb_ipcc_recv(client->ipc, buffer, client->buf_size,
qb_timeout);
+ hdr = (pcmk__ipc_header_t *) (void *) buffer;
+
if (*bytes <= 0) {
if (!crm_ipc_connected(client)) {
crm_err("%s IPC provider disconnected while waiting for message %d",
client->server_name, request_id);
break;
}
continue;
- }
- hdr = (pcmk__ipc_header_t *)(void*) client->buffer;
+ } else if (*bytes != hdr->size + sizeof(pcmk__ipc_header_t)) {
+ crm_trace("Message size does not match header");
+ rc = -EBADMSG;
+ break;
+ }
if (hdr->qb.id == request_id) {
/* Got the reply we were expecting. */
+ if (client->buffer != NULL) {
+ pcmk__ipc_free_client_buffer(client);
+ }
+
+ client->buffer = g_byte_array_sized_new(client->buf_size);
+ g_byte_array_append(client->buffer, (const guint8 *) buffer, *bytes);
break;
}
- xml = pcmk__xml_parse(crm_ipc_buffer(client));
+ data = buffer + sizeof(pcmk__ipc_header_t);
+ xml = pcmk__xml_parse((const char *) data);
if (hdr->qb.id < request_id) {
crm_err("Discarding old reply %d (need %d)", hdr->qb.id, request_id);
crm_log_xml_notice(xml, "OldIpcReply");
} else if (hdr->qb.id > request_id) {
crm_err("Discarding newer reply %d (need %d)", hdr->qb.id, request_id);
crm_log_xml_notice(xml, "ImpossibleReply");
pcmk__assert(hdr->qb.id <= request_id);
}
} while (time(NULL) < timeout || (timeout == 0 && *bytes == -EAGAIN));
if (*bytes > 0) {
crm_trace("Received %zd-byte reply %" PRId32 " to %s IPC %d: %.100s",
*bytes, hdr->qb.id, client->server_name, request_id,
crm_ipc_buffer(client));
if (reply != NULL) {
*reply = pcmk__xml_parse(crm_ipc_buffer(client));
}
} else if (*bytes < 0) {
rc = (int) -*bytes; // System errno
crm_trace("No reply to %s IPC %d: %s " QB_XS " rc=%d",
client->server_name, request_id, pcmk_rc_str(rc), rc);
}
/* If bytes == 0, we'll return that to crm_ipc_send which will interpret
* that as pcmk_rc_ok, log that the IPC request failed (since we did not
* give it a valid reply), and return that 0 to its callers. It's up to
* the callers to take appropriate action after that.
*/
+ /* Once we've parsed the client buffer as XML and saved it to reply,
+ * there's no need to keep the client buffer around anymore. Free it here
+ * to avoid having to do this anywhere crm_ipc_send is called.
+ */
+ pcmk__ipc_free_client_buffer(client);
+ g_free(buffer);
return rc;
}
/*!
* \brief Send an IPC XML message
*
* \param[in,out] client Connection to IPC server
* \param[in] message XML message to send
* \param[in] flags Bitmask of crm_ipc_flags
* \param[in] ms_timeout Give up if not sent within this much time
* (5 seconds if 0, or no timeout if negative)
* \param[out] reply Reply from server (or NULL if none)
*
* \return Negative errno on error, otherwise size of reply received in bytes
* if reply was needed, otherwise number of bytes sent
*/
int
crm_ipc_send(crm_ipc_t *client, const xmlNode *message,
enum crm_ipc_flags flags, int32_t ms_timeout, xmlNode **reply)
{
int rc = 0;
time_t timeout = 0;
ssize_t qb_rc = 0;
ssize_t bytes = 0;
struct iovec *iov;
static uint32_t id = 0;
pcmk__ipc_header_t *header;
+ GString *iov_buffer = NULL;
if (client == NULL) {
crm_notice("Can't send IPC request without connection (bug?): %.100s",
message);
return -ENOTCONN;
} else if (!crm_ipc_connected(client)) {
/* Don't even bother */
crm_notice("Can't send %s IPC requests: Connection closed",
client->server_name);
return -ENOTCONN;
}
if (ms_timeout == 0) {
ms_timeout = 5000;
}
+ /* This loop exists only to clear out any old replies that we haven't
+ * yet read. We don't care about their contents since it's too late to
+ * do anything with them, so we just read and throw them away.
+ */
if (client->need_reply) {
- qb_rc = qb_ipcc_recv(client->ipc, client->buffer, client->buf_size, ms_timeout);
+ char *buffer = pcmk__assert_alloc(crm_ipc_default_buffer_size(),
+ sizeof(char));
+
+ qb_rc = qb_ipcc_recv(client->ipc, buffer, client->buf_size, ms_timeout);
+ free(buffer);
+
if (qb_rc < 0) {
crm_warn("Sending %s IPC disabled until pending reply received",
client->server_name);
return -EALREADY;
} else {
crm_notice("Sending %s IPC re-enabled after pending reply received",
client->server_name);
client->need_reply = FALSE;
}
}
id++;
CRM_LOG_ASSERT(id != 0); /* Crude wrap-around detection */
- rc = pcmk__ipc_prepare_iov(id, message, &iov, &bytes);
- if (rc != pcmk_rc_ok) {
+
+ iov_buffer = g_string_sized_new(1024);
+ pcmk__xml_string(message, 0, iov_buffer, 0);
+ rc = pcmk__ipc_prepare_iov(id, iov_buffer, 0, &iov, &bytes);
+
+ if ((rc != pcmk_rc_ok) && (rc != pcmk_rc_ipc_more)) {
crm_warn("Couldn't prepare %s IPC request: %s " QB_XS " rc=%d",
client->server_name, pcmk_rc_str(rc), rc);
+ g_string_free(iov_buffer, TRUE);
return pcmk_rc2legacy(rc);
}
header = iov[0].iov_base;
pcmk__set_ipc_flags(header->flags, client->server_name, flags);
if (pcmk_is_set(flags, crm_ipc_proxied)) {
/* Don't look for a synchronous response */
pcmk__clear_ipc_flags(flags, "client", crm_ipc_client_response);
}
crm_trace("Sending %s IPC request %d of %u bytes using %dms timeout",
client->server_name, header->qb.id, header->qb.size, ms_timeout);
/* Send the IPC request, respecting any timeout we were passed */
if (ms_timeout > 0) {
timeout = time(NULL) + 1 + pcmk__timeout_ms2s(ms_timeout);
}
do {
qb_rc = qb_ipcc_sendv(client->ipc, iov, 2);
} while ((qb_rc == -EAGAIN) && ((timeout == 0) || (time(NULL) < timeout)));
rc = (int) qb_rc; // Negative of system errno, or bytes sent
if (qb_rc <= 0) {
goto send_cleanup;
}
/* If we should not wait for a response, bail now */
if (!pcmk_is_set(flags, crm_ipc_client_response)) {
crm_trace("Not waiting for reply to %s IPC request %d",
client->server_name, header->qb.id);
goto send_cleanup;
}
rc = internal_ipc_get_reply(client, header->qb.id, ms_timeout, &bytes, reply);
if (rc == pcmk_rc_ok) {
rc = (int) bytes; // Size of reply received
} else {
/* rc is either a positive system errno or a negative standard Pacemaker
* return code. If it's an errno, we need to convert it back to a
* negative number for comparison and return at the end of this function.
*/
rc = pcmk_rc2legacy(rc);
if (ms_timeout > 0) {
/* We didn't get the reply in time, so disable future sends for now.
* The only alternative would be to close the connection since we
* don't know how to detect and discard out-of-sequence replies.
*
* @TODO Implement out-of-sequence detection
*/
client->need_reply = TRUE;
}
}
send_cleanup:
if (!crm_ipc_connected(client)) {
crm_notice("Couldn't send %s IPC request %d: Connection closed "
QB_XS " rc=%d", client->server_name, header->qb.id, rc);
} else if (rc == -ETIMEDOUT) {
crm_warn("%s IPC request %d failed: %s after %dms " QB_XS " rc=%d",
client->server_name, header->qb.id, pcmk_strerror(rc),
ms_timeout, rc);
crm_write_blackbox(0, NULL);
} else if (rc <= 0) {
crm_warn("%s IPC request %d failed: %s " QB_XS " rc=%d",
client->server_name, header->qb.id,
((rc == 0)? "No bytes sent" : pcmk_strerror(rc)), rc);
}
+ g_string_free(iov_buffer, TRUE);
pcmk_free_ipc_event(iov);
return rc;
}
/*!
* \brief Ensure an IPC provider has expected user or group
*
* \param[in] qb_ipc libqb client connection if available
* \param[in] sock Connected Unix socket for IPC
* \param[in] refuid Expected user ID
* \param[in] refgid Expected group ID
* \param[out] gotpid If not NULL, where to store provider's actual process ID
* (or 1 on platforms where ID is not available)
* \param[out] gotuid If not NULL, where to store provider's actual user ID
* \param[out] gotgid If not NULL, where to store provider's actual group ID
*
* \return Standard Pacemaker return code
* \note An actual user ID of 0 (root) will always be considered authorized,
* regardless of the expected values provided. The caller can use the
* output arguments to be stricter than this function.
*/
static int
is_ipc_provider_expected(qb_ipcc_connection_t *qb_ipc, int sock,
uid_t refuid, gid_t refgid,
pid_t *gotpid, uid_t *gotuid, gid_t *gotgid)
{
int rc = EOPNOTSUPP;
pid_t found_pid = 0;
uid_t found_uid = 0;
gid_t found_gid = 0;
#ifdef HAVE_QB_IPCC_AUTH_GET
if (qb_ipc != NULL) {
rc = qb_ipcc_auth_get(qb_ipc, &found_pid, &found_uid, &found_gid);
rc = -rc; // libqb returns 0 or -errno
if (rc == pcmk_rc_ok) {
goto found;
}
}
#endif
#ifdef HAVE_UCRED
{
struct ucred ucred;
socklen_t ucred_len = sizeof(ucred);
if (getsockopt(sock, SOL_SOCKET, SO_PEERCRED, &ucred, &ucred_len) < 0) {
rc = errno;
} else if (ucred_len != sizeof(ucred)) {
rc = EOPNOTSUPP;
} else {
found_pid = ucred.pid;
found_uid = ucred.uid;
found_gid = ucred.gid;
goto found;
}
}
#endif
#ifdef HAVE_SOCKPEERCRED
{
struct sockpeercred sockpeercred;
socklen_t sockpeercred_len = sizeof(sockpeercred);
if (getsockopt(sock, SOL_SOCKET, SO_PEERCRED,
&sockpeercred, &sockpeercred_len) < 0) {
rc = errno;
} else if (sockpeercred_len != sizeof(sockpeercred)) {
rc = EOPNOTSUPP;
} else {
found_pid = sockpeercred.pid;
found_uid = sockpeercred.uid;
found_gid = sockpeercred.gid;
goto found;
}
}
#endif
#ifdef HAVE_GETPEEREID // For example, FreeBSD
if (getpeereid(sock, &found_uid, &found_gid) < 0) {
rc = errno;
} else {
found_pid = PCMK__SPECIAL_PID;
goto found;
}
#endif
#ifdef HAVE_GETPEERUCRED
{
ucred_t *ucred = NULL;
if (getpeerucred(sock, &ucred) < 0) {
rc = errno;
} else {
found_pid = ucred_getpid(ucred);
found_uid = ucred_geteuid(ucred);
found_gid = ucred_getegid(ucred);
ucred_free(ucred);
goto found;
}
}
#endif
return rc; // If we get here, nothing succeeded
found:
if (gotpid != NULL) {
*gotpid = found_pid;
}
if (gotuid != NULL) {
*gotuid = found_uid;
}
if (gotgid != NULL) {
*gotgid = found_gid;
}
if ((found_uid != 0) && (found_uid != refuid) && (found_gid != refgid)) {
return pcmk_rc_ipc_unauthorized;
}
return pcmk_rc_ok;
}
int
crm_ipc_is_authentic_process(int sock, uid_t refuid, gid_t refgid,
pid_t *gotpid, uid_t *gotuid, gid_t *gotgid)
{
int ret = is_ipc_provider_expected(NULL, sock, refuid, refgid,
gotpid, gotuid, gotgid);
/* The old function had some very odd return codes*/
if (ret == 0) {
return 1;
} else if (ret == pcmk_rc_ipc_unauthorized) {
return 0;
} else {
return pcmk_rc2legacy(ret);
}
}
int
pcmk__ipc_is_authentic_process_active(const char *name, uid_t refuid,
gid_t refgid, pid_t *gotpid)
{
static char last_asked_name[PATH_MAX / 2] = ""; /* log spam prevention */
int fd;
int rc = pcmk_rc_ipc_unresponsive;
int auth_rc = 0;
int32_t qb_rc;
pid_t found_pid = 0; uid_t found_uid = 0; gid_t found_gid = 0;
qb_ipcc_connection_t *c;
#ifdef HAVE_QB_IPCC_CONNECT_ASYNC
struct pollfd pollfd = { 0, };
int poll_rc;
c = qb_ipcc_connect_async(name, 0,
&(pollfd.fd));
#else
c = qb_ipcc_connect(name, 0);
#endif
if (c == NULL) {
crm_info("Could not connect to %s IPC: %s", name, strerror(errno));
rc = pcmk_rc_ipc_unresponsive;
goto bail;
}
#ifdef HAVE_QB_IPCC_CONNECT_ASYNC
pollfd.events = POLLIN;
do {
poll_rc = poll(&pollfd, 1, 2000);
} while ((poll_rc == -1) && (errno == EINTR));
/* If poll() failed, given that disconnect function is not registered yet,
* qb_ipcc_disconnect() won't clean up the socket. In any case, call
* qb_ipcc_connect_continue() here so that it may fail and do the cleanup
* for us.
*/
if (qb_ipcc_connect_continue(c) != 0) {
crm_info("Could not connect to %s IPC: %s", name,
(poll_rc == 0)?"timeout":strerror(errno));
rc = pcmk_rc_ipc_unresponsive;
c = NULL; // qb_ipcc_connect_continue cleaned up for us
goto bail;
}
#endif
qb_rc = qb_ipcc_fd_get(c, &fd);
if (qb_rc != 0) {
rc = (int) -qb_rc; // System errno
crm_err("Could not get fd from %s IPC: %s " QB_XS " rc=%d",
name, pcmk_rc_str(rc), rc);
goto bail;
}
auth_rc = is_ipc_provider_expected(c, fd, refuid, refgid,
&found_pid, &found_uid, &found_gid);
if (auth_rc == pcmk_rc_ipc_unauthorized) {
crm_err("Daemon (IPC %s) effectively blocked with unauthorized"
" process %lld (uid: %lld, gid: %lld)",
name, (long long) PCMK__SPECIAL_PID_AS_0(found_pid),
(long long) found_uid, (long long) found_gid);
rc = pcmk_rc_ipc_unauthorized;
goto bail;
}
if (auth_rc != pcmk_rc_ok) {
rc = auth_rc;
crm_err("Could not get peer credentials from %s IPC: %s "
QB_XS " rc=%d", name, pcmk_rc_str(rc), rc);
goto bail;
}
if (gotpid != NULL) {
*gotpid = found_pid;
}
rc = pcmk_rc_ok;
if ((found_uid != refuid || found_gid != refgid)
&& strncmp(last_asked_name, name, sizeof(last_asked_name))) {
if ((found_uid == 0) && (refuid != 0)) {
crm_warn("Daemon (IPC %s) runs as root, whereas the expected"
" credentials are %lld:%lld, hazard of violating"
" the least privilege principle",
name, (long long) refuid, (long long) refgid);
} else {
crm_notice("Daemon (IPC %s) runs as %lld:%lld, whereas the"
" expected credentials are %lld:%lld, which may"
" mean a different set of privileges than expected",
name, (long long) found_uid, (long long) found_gid,
(long long) refuid, (long long) refgid);
}
memccpy(last_asked_name, name, '\0', sizeof(last_asked_name));
}
bail:
if (c != NULL) {
qb_ipcc_disconnect(c);
}
return rc;
}
// Deprecated functions kept only for backward API compatibility
// LCOV_EXCL_START
#include
bool
crm_ipc_connect(crm_ipc_t *client)
{
int rc = pcmk__connect_generic_ipc(client);
if (rc == pcmk_rc_ok) {
return true;
}
if ((client != NULL) && (client->ipc == NULL)) {
errno = (rc > 0)? rc : ENOTCONN;
crm_debug("Could not establish %s IPC connection: %s (%d)",
client->server_name, pcmk_rc_str(errno), errno);
} else if (rc == pcmk_rc_ipc_unauthorized) {
crm_err("%s IPC provider authentication failed",
(client == NULL)? "Pacemaker" : client->server_name);
errno = ECONNABORTED;
} else {
crm_err("Could not verify authenticity of %s IPC provider",
(client == NULL)? "Pacemaker" : client->server_name);
errno = ENOTCONN;
}
return false;
}
// LCOV_EXCL_STOP
// End deprecated API
diff --git a/lib/common/ipc_common.c b/lib/common/ipc_common.c
index 4996ec6d28..3e1e9b398e 100644
--- a/lib/common/ipc_common.c
+++ b/lib/common/ipc_common.c
@@ -1,72 +1,190 @@
/*
* Copyright 2004-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#include
#include
#include // uint64_t
#include
#include
#include "crmcommon_private.h"
/* The IPC buffer is always 128k. If we are asked to send a message larger
* than that size, it will be split into multiple messages that must be
* reassembled on the other end.
*/
#define BUFFER_SIZE (128*1024) // 128k
/*!
* \brief Return pacemaker's IPC buffer size
*
* \return IPC buffer size in bytes
*/
unsigned int
crm_ipc_default_buffer_size(void)
{
return BUFFER_SIZE;
}
/*!
* \internal
* \brief Check whether an IPC header is valid
*
* \param[in] header IPC header to check
*
* \return true if IPC header has a supported version, false otherwise
*/
bool
pcmk__valid_ipc_header(const pcmk__ipc_header_t *header)
{
if (header == NULL) {
crm_err("IPC message without header");
return false;
} else if (header->version > PCMK__IPC_VERSION) {
crm_err("Filtering incompatible v%d IPC message (only versions <= %d supported)",
header->version, PCMK__IPC_VERSION);
return false;
}
return true;
}
const char *
pcmk__client_type_str(uint64_t client_type)
{
switch (client_type) {
case pcmk__client_ipc:
return "IPC";
case pcmk__client_tcp:
return "TCP";
case pcmk__client_tls:
return "TLS";
default:
return "unknown";
}
}
+
+/*!
+ * \internal
+ * \brief Add more data to a received partial IPC message
+ *
+ * This function can be called repeatedly to build up a complete IPC message
+ * from smaller parts. It does this by inspecting flags on the message.
+ * Most of the time, IPC messages will be small enough where this function
+ * won't get called more than once, but more complex clusters can end up with
+ * very large IPC messages that don't fit in a single buffer.
+ *
+ * Important return values:
+ *
+ * - EBADMSG - Something was wrong with the data.
+ * - pcmk_rc_ipc_more - \p data was a chunk of a partial message and there is
+ * more to come. The caller should not process the message
+ * yet and should continue reading from the IPC connection.
+ * - pcmk_rc_ok - We have the complete message. The caller should process
+ * it and free the buffer to prepare for the next message.
+ *
+ * \param[in,out] buffer The buffer to add this data to
+ * \param[in] data The received IPC message or message portion.
+ *
+ * \return Standard Pacemaker return code
+ */
+int
+pcmk__ipc_msg_append(GByteArray **buffer, guint8 *data)
+{
+ pcmk__ipc_header_t *full_header = NULL;
+ pcmk__ipc_header_t *header = (pcmk__ipc_header_t *) (void *) data;
+ const guint8 *payload = (guint8 *) data + sizeof(pcmk__ipc_header_t);
+ int rc = pcmk_rc_ok;
+
+ if (!pcmk__valid_ipc_header(header)) {
+ return EBADMSG;
+ }
+
+ if (pcmk_is_set(header->flags, crm_ipc_multipart_end)) {
+ full_header = (pcmk__ipc_header_t *) (void *) (*buffer)->data;
+
+ /* This is the end of a multipart IPC message. Add the payload of the
+ * received data (so, don't include the header) to the partial buffer.
+ * Remember that this needs to include the null terminating character.
+ */
+ CRM_CHECK(buffer != NULL && *buffer != NULL && header->part_id != 0,
+ return EINVAL);
+ CRM_CHECK(full_header->qb.id == header->qb.id, return EBADMSG);
+ g_byte_array_append(*buffer, payload, header->size);
+
+ crm_trace("Received IPC message %" PRId32 " (final part %" PRIu16 ") of %"
+ PRId32 " bytes",
+ header->qb.id, header->part_id, header->qb.size);
+
+ } else if (pcmk_is_set(header->flags, crm_ipc_multipart)) {
+ if (header->part_id == 0) {
+ /* This is the first part of a multipart IPC message. Initialize
+ * the buffer with the entire message, including its header. Do
+ * not include the null terminating character.
+ */
+ CRM_CHECK(buffer != NULL && *buffer == NULL, return EINVAL);
+ *buffer = g_byte_array_new();
+
+ /* Clear any multipart flags from the header of the incoming part
+ * so they'll be clear in the fully reassembled message. This
+ * message is passed to pcmk__client_data2xml, which will extract
+ * the header flags and return them. Those flags can then be used
+ * when constructing a reply, including ACKs. We don't want these
+ * specific incoming flags to influence the reply.
+ */
+ pcmk__clear_ipc_flags(header->flags, "server", crm_ipc_multipart);
+
+ g_byte_array_append(*buffer, data,
+ sizeof(pcmk__ipc_header_t) + header->size - 1);
+
+ } else {
+ full_header = (pcmk__ipc_header_t *) (void *) (*buffer)->data;
+
+ /* This is some intermediate part of a multipart message. Add
+ * the payload of the received data (so, don't include the header)
+ * to the partial buffer and return. Do not include the null
+ * terminating character.
+ */
+ CRM_CHECK(buffer != NULL && *buffer != NULL, return EINVAL);
+ CRM_CHECK(full_header->qb.id == header->qb.id, return EBADMSG);
+ g_byte_array_append(*buffer, payload, header->size - 1);
+ }
+
+ crm_trace("Received IPC message %" PRId32 " (part %" PRIu16 ") of %"
+ PRId32 " bytes",
+ header->qb.id, header->part_id, header->qb.size);
+
+ rc = pcmk_rc_ipc_more;
+
+ } else {
+ /* This is a standalone IPC message. For simplicity in the caller,
+ * copy the entire message over into a byte array so it can be handled
+ * the same as a multipart message.
+ */
+ CRM_CHECK(buffer != NULL && *buffer == NULL, return EINVAL);
+ *buffer = g_byte_array_new();
+ g_byte_array_append(*buffer, data,
+ sizeof(pcmk__ipc_header_t) + header->size);
+
+ crm_trace("Received IPC message %" PRId32 " of %" PRId32 " bytes",
+ header->qb.id, header->qb.size);
+ }
+
+ crm_trace("Text = %s", payload);
+ crm_trace("Buffer = %s", (*buffer)->data + sizeof(pcmk__ipc_header_t));
+
+ /* The buffer's header should have a size that matches the full size of
+ * the received message, not just the last chunk of it.
+ */
+ full_header = (pcmk__ipc_header_t *) (void *) (*buffer)->data;
+ full_header->size = (*buffer)->len - sizeof(pcmk__ipc_header_t);
+
+ return rc;
+}
diff --git a/lib/common/ipc_server.c b/lib/common/ipc_server.c
index f6bb0bddcc..ebc113bde9 100644
--- a/lib/common/ipc_server.c
+++ b/lib/common/ipc_server.c
@@ -1,947 +1,1094 @@
/*
* Copyright 2004-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include "crmcommon_private.h"
/* Evict clients whose event queue grows this large (by default) */
#define PCMK_IPC_DEFAULT_QUEUE_MAX 500
static GHashTable *client_connections = NULL;
/*!
* \internal
* \brief Count IPC clients
*
* \return Number of active IPC client connections
*/
guint
pcmk__ipc_client_count(void)
{
return client_connections? g_hash_table_size(client_connections) : 0;
}
/*!
* \internal
* \brief Execute a function for each active IPC client connection
*
* \param[in] func Function to call
* \param[in,out] user_data Pointer to pass to function
*
* \note The parameters are the same as for g_hash_table_foreach().
*/
void
pcmk__foreach_ipc_client(GHFunc func, gpointer user_data)
{
if ((func != NULL) && (client_connections != NULL)) {
g_hash_table_foreach(client_connections, func, user_data);
}
}
pcmk__client_t *
pcmk__find_client(const qb_ipcs_connection_t *c)
{
if (client_connections) {
return g_hash_table_lookup(client_connections, c);
}
crm_trace("No client found for %p", c);
return NULL;
}
pcmk__client_t *
pcmk__find_client_by_id(const char *id)
{
if ((client_connections != NULL) && (id != NULL)) {
gpointer key;
pcmk__client_t *client = NULL;
GHashTableIter iter;
g_hash_table_iter_init(&iter, client_connections);
while (g_hash_table_iter_next(&iter, &key, (gpointer *) & client)) {
if (strcmp(client->id, id) == 0) {
return client;
}
}
}
crm_trace("No client found with id='%s'", pcmk__s(id, ""));
return NULL;
}
/*!
* \internal
* \brief Get a client identifier for use in log messages
*
* \param[in] c Client
*
* \return Client's name, client's ID, or a string literal, as available
* \note This is intended to be used in format strings like "client %s".
*/
const char *
pcmk__client_name(const pcmk__client_t *c)
{
if (c == NULL) {
return "(unspecified)";
} else if (c->name != NULL) {
return c->name;
} else if (c->id != NULL) {
return c->id;
} else {
return "(unidentified)";
}
}
void
pcmk__client_cleanup(void)
{
if (client_connections != NULL) {
int active = g_hash_table_size(client_connections);
if (active > 0) {
crm_warn("Exiting with %d active IPC client%s",
active, pcmk__plural_s(active));
}
g_hash_table_destroy(client_connections);
client_connections = NULL;
}
}
void
pcmk__drop_all_clients(qb_ipcs_service_t *service)
{
qb_ipcs_connection_t *c = NULL;
if (service == NULL) {
return;
}
c = qb_ipcs_connection_first_get(service);
while (c != NULL) {
qb_ipcs_connection_t *last = c;
c = qb_ipcs_connection_next_get(service, last);
/* There really shouldn't be anyone connected at this point */
crm_notice("Disconnecting client %p, pid=%d...",
last, pcmk__client_pid(last));
qb_ipcs_disconnect(last);
qb_ipcs_connection_unref(last);
}
}
/*!
* \internal
* \brief Allocate a new pcmk__client_t object based on an IPC connection
*
* \param[in] c IPC connection (NULL to allocate generic client)
* \param[in] key Connection table key (NULL to use sane default)
* \param[in] uid_client UID corresponding to c (ignored if c is NULL)
*
* \return Pointer to new pcmk__client_t (guaranteed not to be \c NULL)
*/
static pcmk__client_t *
client_from_connection(qb_ipcs_connection_t *c, void *key, uid_t uid_client)
{
pcmk__client_t *client = pcmk__assert_alloc(1, sizeof(pcmk__client_t));
if (c) {
client->user = pcmk__uid2username(uid_client);
if (client->user == NULL) {
client->user = pcmk__str_copy("#unprivileged");
crm_err("Unable to enforce ACLs for user ID %d, assuming unprivileged",
uid_client);
}
client->ipcs = c;
pcmk__set_client_flags(client, pcmk__client_ipc);
client->pid = pcmk__client_pid(c);
if (key == NULL) {
key = c;
}
}
client->id = crm_generate_uuid();
if (key == NULL) {
key = client->id;
}
if (client_connections == NULL) {
crm_trace("Creating IPC client table");
client_connections = g_hash_table_new(g_direct_hash, g_direct_equal);
}
g_hash_table_insert(client_connections, key, client);
return client;
}
/*!
* \brief Allocate a new pcmk__client_t object and generate its ID
*
* \param[in] key What to use as connections hash table key (NULL to use ID)
*
* \return Pointer to new pcmk__client_t (asserts on failure)
*/
pcmk__client_t *
pcmk__new_unauth_client(void *key)
{
return client_from_connection(NULL, key, 0);
}
pcmk__client_t *
pcmk__new_client(qb_ipcs_connection_t *c, uid_t uid_client, gid_t gid_client)
{
gid_t uid_cluster = 0;
gid_t gid_cluster = 0;
pcmk__client_t *client = NULL;
CRM_CHECK(c != NULL, return NULL);
if (pcmk_daemon_user(&uid_cluster, &gid_cluster) < 0) {
static bool need_log = TRUE;
if (need_log) {
crm_warn("Could not find user and group IDs for user %s",
CRM_DAEMON_USER);
need_log = FALSE;
}
}
if (uid_client != 0) {
crm_trace("Giving group %u access to new IPC connection", gid_cluster);
/* Passing -1 to chown(2) means don't change */
qb_ipcs_connection_auth_set(c, -1, gid_cluster, S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP);
}
/* TODO: Do our own auth checking, return NULL if unauthorized */
client = client_from_connection(c, NULL, uid_client);
if ((uid_client == 0) || (uid_client == uid_cluster)) {
/* Remember when a connection came from root or hacluster */
pcmk__set_client_flags(client, pcmk__client_privileged);
}
crm_debug("New IPC client %s for PID %u with uid %d and gid %d",
client->id, client->pid, uid_client, gid_client);
return client;
}
static struct iovec *
pcmk__new_ipc_event(void)
{
return (struct iovec *) pcmk__assert_alloc(2, sizeof(struct iovec));
}
/*!
* \brief Free an I/O vector created by pcmk__ipc_prepare_iov()
*
* \param[in,out] event I/O vector to free
*/
void
pcmk_free_ipc_event(struct iovec *event)
{
if (event != NULL) {
free(event[0].iov_base);
free(event[1].iov_base);
free(event);
}
}
static void
free_event(gpointer data)
{
pcmk_free_ipc_event((struct iovec *) data);
}
static void
add_event(pcmk__client_t *c, struct iovec *iov)
{
if (c->event_queue == NULL) {
c->event_queue = g_queue_new();
}
g_queue_push_tail(c->event_queue, iov);
}
void
pcmk__free_client(pcmk__client_t *c)
{
if (c == NULL) {
return;
}
if (client_connections) {
if (c->ipcs) {
crm_trace("Destroying %p/%p (%d remaining)",
c, c->ipcs, g_hash_table_size(client_connections) - 1);
g_hash_table_remove(client_connections, c->ipcs);
} else {
crm_trace("Destroying remote connection %p (%d remaining)",
c, g_hash_table_size(client_connections) - 1);
g_hash_table_remove(client_connections, c->id);
}
}
if (c->event_timer) {
g_source_remove(c->event_timer);
}
if (c->event_queue) {
crm_debug("Destroying %d events", g_queue_get_length(c->event_queue));
g_queue_free_full(c->event_queue, free_event);
}
free(c->id);
free(c->name);
free(c->user);
if (c->remote) {
if (c->remote->auth_timeout) {
g_source_remove(c->remote->auth_timeout);
}
if (c->remote->tls_session != NULL) {
/* @TODO Reduce duplication at callers. Put here everything
* necessary to tear down and free tls_session.
*/
gnutls_deinit(c->remote->tls_session);
}
free(c->remote->buffer);
free(c->remote);
}
free(c);
}
/*!
* \internal
* \brief Raise IPC eviction threshold for a client, if allowed
*
* \param[in,out] client Client to modify
* \param[in] qmax New threshold
*/
void
pcmk__set_client_queue_max(pcmk__client_t *client, const char *qmax)
{
int rc = pcmk_rc_ok;
long long qmax_ll = 0LL;
unsigned int orig_value = 0U;
CRM_CHECK(client != NULL, return);
orig_value = client->queue_max;
if (pcmk_is_set(client->flags, pcmk__client_privileged)) {
rc = pcmk__scan_ll(qmax, &qmax_ll, 0LL);
if (rc == pcmk_rc_ok) {
if ((qmax_ll <= 0LL) || (qmax_ll > UINT_MAX)) {
rc = ERANGE;
} else {
client->queue_max = (unsigned int) qmax_ll;
}
}
} else {
rc = EACCES;
}
if (rc != pcmk_rc_ok) {
crm_info("Could not set IPC threshold for client %s[%u] to %s: %s",
pcmk__client_name(client), client->pid,
pcmk__s(qmax, "default"), pcmk_rc_str(rc));
} else if (client->queue_max != orig_value) {
crm_debug("IPC threshold for client %s[%u] is now %u (was %u)",
pcmk__client_name(client), client->pid,
client->queue_max, orig_value);
}
}
int
pcmk__client_pid(qb_ipcs_connection_t *c)
{
struct qb_ipcs_connection_stats stats;
stats.client_pid = 0;
qb_ipcs_connection_stats_get(c, &stats, 0);
return stats.client_pid;
}
/*!
* \internal
* \brief Retrieve message XML from data read from client IPC
*
* \param[in,out] c IPC client connection
* \param[in] data Data read from client connection
* \param[out] id Where to store message ID from libqb header
* \param[out] flags Where to store flags from libqb header
*
* \return Message XML on success, NULL otherwise
*/
xmlNode *
pcmk__client_data2xml(pcmk__client_t *c, void *data, uint32_t *id,
uint32_t *flags)
{
xmlNode *xml = NULL;
char *text = ((char *)data) + sizeof(pcmk__ipc_header_t);
pcmk__ipc_header_t *header = data;
if (!pcmk__valid_ipc_header(header)) {
return NULL;
}
if (id) {
*id = ((struct qb_ipc_response_header *)data)->id;
}
if (flags) {
*flags = header->flags;
}
if (pcmk_is_set(header->flags, crm_ipc_proxied)) {
/* Mark this client as being the endpoint of a proxy connection.
* Proxy connections responses are sent on the event channel, to avoid
* blocking the controller serving as proxy.
*/
pcmk__set_client_flags(c, pcmk__client_proxied);
}
pcmk__assert(text[header->size - 1] == 0);
xml = pcmk__xml_parse(text);
crm_log_xml_trace(xml, "[IPC received]");
return xml;
}
static int crm_ipcs_flush_events(pcmk__client_t *c);
static gboolean
crm_ipcs_flush_events_cb(gpointer data)
{
pcmk__client_t *c = data;
c->event_timer = 0;
crm_ipcs_flush_events(c);
return FALSE;
}
/*!
* \internal
* \brief Add progressive delay before next event queue flush
*
* \param[in,out] c Client connection to add delay to
* \param[in] queue_len Current event queue length
*/
static inline void
delay_next_flush(pcmk__client_t *c, unsigned int queue_len)
{
/* Delay a maximum of 1.5 seconds */
guint delay = (queue_len < 5)? (1000 + 100 * queue_len) : 1500;
c->event_timer = pcmk__create_timer(delay, crm_ipcs_flush_events_cb, c);
}
/*!
* \internal
* \brief Send client any messages in its queue
*
* \param[in,out] c Client to flush
*
* \return Standard Pacemaker return value
*/
static int
crm_ipcs_flush_events(pcmk__client_t *c)
{
int rc = pcmk_rc_ok;
ssize_t qb_rc = 0;
unsigned int sent = 0;
unsigned int queue_len = 0;
if (c == NULL) {
return rc;
} else if (c->event_timer) {
/* There is already a timer, wait until it goes off */
crm_trace("Timer active for %p - %d", c->ipcs, c->event_timer);
return rc;
}
if (c->event_queue) {
queue_len = g_queue_get_length(c->event_queue);
}
while (sent < 100) {
pcmk__ipc_header_t *header = NULL;
struct iovec *event = NULL;
if (c->event_queue) {
// We don't pop unless send is successful
event = g_queue_peek_head(c->event_queue);
}
if (event == NULL) { // Queue is empty
break;
}
qb_rc = qb_ipcs_event_sendv(c->ipcs, event, 2);
if (qb_rc < 0) {
rc = (int) -qb_rc;
break;
}
event = g_queue_pop_head(c->event_queue);
sent++;
header = event[0].iov_base;
crm_trace("Event %" PRId32 " to %p[%u] (%zd bytes) sent: %.120s",
header->qb.id, c->ipcs, c->pid, qb_rc,
(char *) (event[1].iov_base));
pcmk_free_ipc_event(event);
}
queue_len -= sent;
if (sent > 0 || queue_len) {
crm_trace("Sent %u events (%u remaining) for %p[%d]: %s (%zd)",
sent, queue_len, c->ipcs, c->pid, pcmk_rc_str(rc), qb_rc);
}
if (queue_len) {
/* Allow clients to briefly fall behind on processing incoming messages,
* but drop completely unresponsive clients so the connection doesn't
* consume resources indefinitely.
*/
if (queue_len > QB_MAX(c->queue_max, PCMK_IPC_DEFAULT_QUEUE_MAX)) {
if ((c->queue_backlog <= 1) || (queue_len < c->queue_backlog)) {
/* Don't evict for a new or shrinking backlog */
crm_warn("Client with process ID %u has a backlog of %u messages "
QB_XS " %p", c->pid, queue_len, c->ipcs);
} else {
crm_err("Evicting client with process ID %u due to backlog of %u messages "
QB_XS " %p", c->pid, queue_len, c->ipcs);
c->queue_backlog = 0;
qb_ipcs_disconnect(c->ipcs);
return rc;
}
}
c->queue_backlog = queue_len;
delay_next_flush(c, queue_len);
} else {
/* Event queue is empty, there is no backlog */
c->queue_backlog = 0;
}
return rc;
}
/*!
* \internal
* \brief Create an I/O vector for sending an IPC XML message
*
- * \param[in] request Identifier for libqb response header
- * \param[in] message XML message to send
- * \param[out] result Where to store prepared I/O vector - NULL
- * on error
- * \param[out] bytes Size of prepared data in bytes
+ * If the message is too large to fit into a single buffer, this function will
+ * prepare an I/O vector that only holds as much as fits. The remainder can be
+ * prepared in a separate call by keeping a running count of the number of times
+ * this function has been called and passing that in for \p index.
+ *
+ * \param[in] request Identifier for libqb response header
+ * \param[in] message Message to send
+ * \param[in] index How many times this function has been called - basically,
+ * a count of how many chunks of \p message have already
+ * been sent
+ * \param[out] result Where to store prepared I/O vector - NULL on error
+ * \param[out] bytes Size of prepared data in bytes (includes header)
*
* \return Standard Pacemaker return code
*/
int
-pcmk__ipc_prepare_iov(uint32_t request, const xmlNode *message,
+pcmk__ipc_prepare_iov(uint32_t request, const GString *message, uint16_t index,
struct iovec **result, ssize_t *bytes)
{
- struct iovec *iov;
+ struct iovec *iov = NULL;
+ unsigned int payload_size = 0;
unsigned int total = 0;
unsigned int max_send_size = crm_ipc_default_buffer_size();
- GString *buffer = NULL;
+ unsigned int max_chunk_size = 0;
+ size_t offset = 0;
pcmk__ipc_header_t *header = NULL;
int rc = pcmk_rc_ok;
if ((message == NULL) || (result == NULL)) {
rc = EINVAL;
goto done;
}
header = calloc(1, sizeof(pcmk__ipc_header_t));
if (header == NULL) {
rc = ENOMEM;
goto done;
}
- buffer = g_string_sized_new(1024);
- pcmk__xml_string(message, 0, buffer, 0);
-
*result = NULL;
iov = pcmk__new_ipc_event();
iov[0].iov_len = sizeof(pcmk__ipc_header_t);
iov[0].iov_base = header;
header->version = PCMK__IPC_VERSION;
- header->size = buffer->len + 1;
- total = iov[0].iov_len + header->size;
+
+ /* We are passed an index, which is basically how many times this function
+ * has been called. This is how we support multi-part IPC messages. We
+ * need to convert that into an offset into the buffer that we want to start
+ * reading from.
+ *
+ * Each call to this function can send max_send_size, but this also includes
+ * the header and a null terminator character for the end of the payload.
+ * We need to subtract those out here.
+ */
+ max_chunk_size = max_send_size - iov[0].iov_len - 1;
+ offset = index * max_chunk_size;
+
+ /* How much of message is left to send? This does not include the null
+ * terminator character.
+ */
+ payload_size = message->len - offset;
+
+ /* How much would be transmitted, including the header size and null
+ * terminator character for the buffer?
+ */
+ total = iov[0].iov_len + payload_size + 1;
if (total >= max_send_size) {
- crm_log_xml_trace(message, "EMSGSIZE");
- crm_err("Could not transmit message; message size %" PRIu32" bytes is "
- "larger than the maximum of %" PRIu32, header->size,
- max_send_size);
- rc = EMSGSIZE;
- pcmk_free_ipc_event(iov);
- goto done;
- }
+ /* The entire packet is too big to fit in a single buffer. Calculate
+ * how much of it we can send - buffer size, minus header size, minus
+ * one for the null terminator.
+ */
+ payload_size = max_chunk_size;
+
+ header->size = payload_size + 1;
+
+ iov[1].iov_base = strndup(message->str + offset, payload_size);
+ if (iov[1].iov_base == NULL) {
+ rc = ENOMEM;
+ goto done;
+ }
- iov[1].iov_base = pcmk__str_copy(buffer->str);
- iov[1].iov_len = header->size;
+ iov[1].iov_len = header->size;
+ rc = pcmk_rc_ipc_more;
+ } else {
+ /* The entire packet fits in a single buffer. We can copy the entirety
+ * of it into the payload.
+ */
+ header->size = payload_size + 1;
+
+ iov[1].iov_base = pcmk__str_copy(message->str + offset);
+ iov[1].iov_len = header->size;
+ }
+
+ header->part_id = index;
header->qb.size = iov[0].iov_len + iov[1].iov_len;
header->qb.id = (int32_t)request; /* Replying to a specific request */
+ if ((rc == pcmk_rc_ok) && (index != 0)) {
+ pcmk__set_ipc_flags(header->flags, "multipart ipc",
+ crm_ipc_multipart | crm_ipc_multipart_end);
+ } else if (rc == pcmk_rc_ipc_more) {
+ pcmk__set_ipc_flags(header->flags, "multipart ipc",
+ crm_ipc_multipart);
+ }
+
*result = iov;
pcmk__assert(header->qb.size > 0);
if (bytes != NULL) {
*bytes = header->qb.size;
}
done:
- if (buffer != NULL) {
- g_string_free(buffer, TRUE);
+ if ((rc != pcmk_rc_ok) && (rc != pcmk_rc_ipc_more)) {
+ pcmk_free_ipc_event(iov);
}
+
return rc;
}
+/* Return the next available ID for a server event.
+ *
+ * For the parts of a multipart event, all parts should have the same ID as
+ * the first part.
+ */
+static uint32_t
+id_for_server_event(pcmk__ipc_header_t *header)
+{
+ static uint32_t id = 1;
+
+ if (pcmk_is_set(header->flags, crm_ipc_multipart) && (header->part_id != 0)) {
+ return id;
+ } else {
+ id++;
+ return id;
+ }
+}
+
int
pcmk__ipc_send_iov(pcmk__client_t *c, struct iovec *iov, uint32_t flags)
{
int rc = pcmk_rc_ok;
- static uint32_t id = 1;
pcmk__ipc_header_t *header = iov[0].iov_base;
- if (c->flags & pcmk__client_proxied) {
- /* _ALL_ replies to proxied connections need to be sent as events */
- if (!pcmk_is_set(flags, crm_ipc_server_event)) {
- /* The proxied flag lets us know this was originally meant to be a
- * response, even though we're sending it over the event channel.
- */
- pcmk__set_ipc_flags(flags, "server event",
- crm_ipc_server_event
- |crm_ipc_proxied_relay_response);
- }
+ /* _ALL_ replies to proxied connections need to be sent as events */
+ if (pcmk_is_set(c->flags, pcmk__client_proxied)
+ && !pcmk_is_set(flags, crm_ipc_server_event)) {
+ /* The proxied flag lets us know this was originally meant to be a
+ * response, even though we're sending it over the event channel.
+ */
+ pcmk__set_ipc_flags(flags, "server event",
+ crm_ipc_server_event|crm_ipc_proxied_relay_response);
}
pcmk__set_ipc_flags(header->flags, "server event", flags);
- if (flags & crm_ipc_server_event) {
- header->qb.id = id++; /* We don't really use it, but doesn't hurt to set one */
+ if (pcmk_is_set(flags, crm_ipc_server_event)) {
+ /* Server events don't use an ID, though we do set one in
+ * pcmk__ipc_prepare_iov if the event is in response to a particular
+ * request. In that case, we don't want to set a new ID here that
+ * overwrites that one.
+ *
+ * @TODO: Since server event IDs aren't used anywhere, do we really
+ * need to set this for any reason other than ease of logging?
+ */
+ if (header->qb.id == 0) {
+ header->qb.id = id_for_server_event(header);
+ }
- if (flags & crm_ipc_server_free) {
+ if (pcmk_is_set(flags, crm_ipc_server_free)) {
crm_trace("Sending the original to %p[%d]", c->ipcs, c->pid);
add_event(c, iov);
} else {
struct iovec *iov_copy = pcmk__new_ipc_event();
crm_trace("Sending a copy to %p[%d]", c->ipcs, c->pid);
iov_copy[0].iov_len = iov[0].iov_len;
iov_copy[0].iov_base = malloc(iov[0].iov_len);
memcpy(iov_copy[0].iov_base, iov[0].iov_base, iov[0].iov_len);
iov_copy[1].iov_len = iov[1].iov_len;
iov_copy[1].iov_base = malloc(iov[1].iov_len);
memcpy(iov_copy[1].iov_base, iov[1].iov_base, iov[1].iov_len);
add_event(c, iov_copy);
}
+ rc = crm_ipcs_flush_events(c);
+
} else {
ssize_t qb_rc;
CRM_LOG_ASSERT(header->qb.id != 0); /* Replying to a specific request */
qb_rc = qb_ipcs_response_sendv(c->ipcs, iov, 2);
if (qb_rc < header->qb.size) {
if (qb_rc < 0) {
rc = (int) -qb_rc;
}
crm_notice("Response %" PRId32 " to pid %u failed: %s "
QB_XS " bytes=%" PRId32 " rc=%zd ipcs=%p",
header->qb.id, c->pid, pcmk_rc_str(rc),
header->qb.size, qb_rc, c->ipcs);
} else {
crm_trace("Response %" PRId32 " sent, %zd bytes to %p[%u]",
header->qb.id, qb_rc, c->ipcs, c->pid);
}
- if (flags & crm_ipc_server_free) {
+ if (pcmk_is_set(flags, crm_ipc_server_free)) {
pcmk_free_ipc_event(iov);
}
- }
- if (flags & crm_ipc_server_event) {
- rc = crm_ipcs_flush_events(c);
- } else {
crm_ipcs_flush_events(c);
}
if ((rc == EPIPE) || (rc == ENOTCONN)) {
crm_trace("Client %p disconnected", c->ipcs);
}
+
return rc;
}
int
pcmk__ipc_send_xml(pcmk__client_t *c, uint32_t request, const xmlNode *message,
uint32_t flags)
{
struct iovec *iov = NULL;
int rc = pcmk_rc_ok;
+ GString *iov_buffer = NULL;
if (c == NULL) {
return EINVAL;
}
- rc = pcmk__ipc_prepare_iov(request, message, &iov, NULL);
- if (rc == pcmk_rc_ok) {
- pcmk__set_ipc_flags(flags, "send data", crm_ipc_server_free);
- rc = pcmk__ipc_send_iov(c, iov, flags);
+
+ iov_buffer = g_string_sized_new(1024);
+ pcmk__xml_string(message, 0, iov_buffer, 0);
+
+ /* Testing crm_ipc_server_event is obvious. pcmk__client_proxied is less
+ * obvious. According to pcmk__ipc_send_iov, replies to proxied connections
+ * need to be sent as events. However, do_local_notify (which calls this
+ * function) will clear all flags so we can't go just by crm_ipc_server_event.
+ *
+ * Changing do_local_notify to check for a proxied connection first results
+ * in processes on the Pacemaker Remote node (like cibadmin or crm_mon)
+ * timing out when waiting for a reply.
+ *
+ * Currently, server events support multipart IPC messages, while other
+ * IPC message types do not.
+ */
+ if (pcmk_is_set(flags, crm_ipc_server_event)
+ || pcmk_is_set(c->flags, pcmk__client_proxied)) {
+ uint16_t index = 0;
+
+ do {
+ rc = pcmk__ipc_prepare_iov(request, iov_buffer, index, &iov, NULL);
+
+ switch (rc) {
+ case pcmk_rc_ok:
+ /* No more chunks to send after this one */
+ pcmk__set_ipc_flags(flags, "send data", crm_ipc_server_free);
+ rc = pcmk__ipc_send_iov(c, iov, flags);
+
+ /* Return pcmk_rc_ok instead so callers don't have to know
+ * whether they passed an event or not when interpreting the
+ * return code.
+ */
+ if (rc == EAGAIN) {
+ rc = pcmk_rc_ok;
+ }
+
+ goto done;
+
+ case pcmk_rc_ipc_more:
+ /* There are more chunks to send after this one */
+ pcmk__set_ipc_flags(flags, "send data", crm_ipc_server_free);
+ rc = pcmk__ipc_send_iov(c, iov, flags);
+
+ /* Did an error occur during transmission? EAGAIN is not
+ * an error for server events. The event will be queued for
+ * transmission and we will attempt sending it again the
+ * next time pcmk__ipc_send_iov is called, or when the
+ * crm_ipcs_flush_events_cb happens.
+ */
+ if ((rc != pcmk_rc_ok) && (rc != EAGAIN)) {
+ goto done;
+ }
+
+ index++;
+ break;
+
+ default:
+ /* An error occurred during preparation */
+ goto done;
+ }
+ } while (true);
+
} else {
- crm_notice("IPC message to pid %d failed: %s " QB_XS " rc=%d",
+ rc = pcmk__ipc_prepare_iov(request, iov_buffer, 0, &iov, NULL);
+
+ if ((rc == pcmk_rc_ok) || (rc == pcmk_rc_ipc_more)) {
+ pcmk__set_ipc_flags(flags, "send data", crm_ipc_server_free);
+ rc = pcmk__ipc_send_iov(c, iov, flags);
+ }
+ }
+
+done:
+ if ((rc != pcmk_rc_ok) && (rc != EAGAIN)) {
+ crm_notice("IPC message to pid %u failed: %s " QB_XS " rc=%d",
c->pid, pcmk_rc_str(rc), rc);
}
+
+ g_string_free(iov_buffer, TRUE);
return rc;
}
/*!
* \internal
* \brief Create an acknowledgement with a status code to send to a client
*
* \param[in] function Calling function
* \param[in] line Source file line within calling function
* \param[in] flags IPC flags to use when sending
* \param[in] tag Element name to use for acknowledgement
* \param[in] ver IPC protocol version (can be NULL)
* \param[in] status Exit status code to add to ack
*
* \return Newly created XML for ack
*
* \note The caller is responsible for freeing the return value with
* \c pcmk__xml_free().
*/
xmlNode *
pcmk__ipc_create_ack_as(const char *function, int line, uint32_t flags,
const char *tag, const char *ver, crm_exit_t status)
{
xmlNode *ack = NULL;
if (pcmk_is_set(flags, crm_ipc_client_response)) {
ack = pcmk__xe_create(NULL, tag);
crm_xml_add(ack, PCMK_XA_FUNCTION, function);
crm_xml_add_int(ack, PCMK__XA_LINE, line);
crm_xml_add_int(ack, PCMK_XA_STATUS, (int) status);
crm_xml_add(ack, PCMK__XA_IPC_PROTO_VERSION, ver);
}
return ack;
}
/*!
* \internal
* \brief Send an acknowledgement with a status code to a client
*
* \param[in] function Calling function
* \param[in] line Source file line within calling function
* \param[in] c Client to send ack to
* \param[in] request Request ID being replied to
* \param[in] flags IPC flags to use when sending
* \param[in] tag Element name to use for acknowledgement
* \param[in] ver IPC protocol version (can be NULL)
* \param[in] status Status code to send with acknowledgement
*
* \return Standard Pacemaker return code
*/
int
pcmk__ipc_send_ack_as(const char *function, int line, pcmk__client_t *c,
uint32_t request, uint32_t flags, const char *tag,
const char *ver, crm_exit_t status)
{
int rc = pcmk_rc_ok;
xmlNode *ack = pcmk__ipc_create_ack_as(function, line, flags, tag, ver, status);
if (ack != NULL) {
crm_trace("Ack'ing IPC message from client %s as <%s status=%d>",
pcmk__client_name(c), tag, status);
crm_log_xml_trace(ack, "sent-ack");
c->request_id = 0;
rc = pcmk__ipc_send_xml(c, request, ack, flags);
pcmk__xml_free(ack);
}
return rc;
}
/*!
* \internal
* \brief Add an IPC server to the main loop for the CIB manager API
*
* \param[out] ipcs_ro New IPC server for read-only CIB manager API
* \param[out] ipcs_rw New IPC server for read/write CIB manager API
* \param[out] ipcs_shm New IPC server for shared-memory CIB manager API
* \param[in] ro_cb IPC callbacks for read-only API
* \param[in] rw_cb IPC callbacks for read/write and shared-memory APIs
*
* \note This function exits fatally if unable to create the servers.
* \note There is no actual difference between the three IPC endpoints other
* than their names.
*/
void pcmk__serve_based_ipc(qb_ipcs_service_t **ipcs_ro,
qb_ipcs_service_t **ipcs_rw,
qb_ipcs_service_t **ipcs_shm,
struct qb_ipcs_service_handlers *ro_cb,
struct qb_ipcs_service_handlers *rw_cb)
{
*ipcs_ro = mainloop_add_ipc_server(PCMK__SERVER_BASED_RO,
QB_IPC_NATIVE, ro_cb);
*ipcs_rw = mainloop_add_ipc_server(PCMK__SERVER_BASED_RW,
QB_IPC_NATIVE, rw_cb);
*ipcs_shm = mainloop_add_ipc_server(PCMK__SERVER_BASED_SHM,
QB_IPC_SHM, rw_cb);
if (*ipcs_ro == NULL || *ipcs_rw == NULL || *ipcs_shm == NULL) {
crm_err("Failed to create the CIB manager: exiting and inhibiting respawn");
crm_warn("Verify pacemaker and pacemaker_remote are not both enabled");
crm_exit(CRM_EX_FATAL);
}
}
/*!
* \internal
* \brief Destroy IPC servers for the CIB manager API
*
* \param[out] ipcs_ro IPC server for read-only the CIB manager API
* \param[out] ipcs_rw IPC server for read/write the CIB manager API
* \param[out] ipcs_shm IPC server for shared-memory the CIB manager API
*
* \note This is a convenience function for calling qb_ipcs_destroy() for each
* argument.
*/
void
pcmk__stop_based_ipc(qb_ipcs_service_t *ipcs_ro,
qb_ipcs_service_t *ipcs_rw,
qb_ipcs_service_t *ipcs_shm)
{
qb_ipcs_destroy(ipcs_ro);
qb_ipcs_destroy(ipcs_rw);
qb_ipcs_destroy(ipcs_shm);
}
/*!
* \internal
* \brief Add an IPC server to the main loop for the controller API
*
* \param[in] cb IPC callbacks
*
* \return Newly created IPC server
*/
qb_ipcs_service_t *
pcmk__serve_controld_ipc(struct qb_ipcs_service_handlers *cb)
{
return mainloop_add_ipc_server(CRM_SYSTEM_CRMD, QB_IPC_NATIVE, cb);
}
/*!
* \internal
* \brief Add an IPC server to the main loop for the attribute manager API
*
* \param[out] ipcs Where to store newly created IPC server
* \param[in] cb IPC callbacks
*
* \note This function exits fatally if unable to create the servers.
*/
void
pcmk__serve_attrd_ipc(qb_ipcs_service_t **ipcs,
struct qb_ipcs_service_handlers *cb)
{
*ipcs = mainloop_add_ipc_server(PCMK__VALUE_ATTRD, QB_IPC_NATIVE, cb);
if (*ipcs == NULL) {
crm_crit("Exiting fatally because unable to serve " PCMK__SERVER_ATTRD
" IPC (verify pacemaker and pacemaker_remote are not both "
"enabled)");
crm_exit(CRM_EX_FATAL);
}
}
/*!
* \internal
* \brief Add an IPC server to the main loop for the fencer API
*
* \param[out] ipcs Where to store newly created IPC server
* \param[in] cb IPC callbacks
*
* \note This function exits fatally if unable to create the servers.
*/
void
pcmk__serve_fenced_ipc(qb_ipcs_service_t **ipcs,
struct qb_ipcs_service_handlers *cb)
{
*ipcs = mainloop_add_ipc_server_with_prio("stonith-ng", QB_IPC_NATIVE, cb,
QB_LOOP_HIGH);
if (*ipcs == NULL) {
crm_err("Failed to create fencer: exiting and inhibiting respawn.");
crm_warn("Verify pacemaker and pacemaker_remote are not both enabled.");
crm_exit(CRM_EX_FATAL);
}
}
/*!
* \internal
* \brief Add an IPC server to the main loop for the pacemakerd API
*
* \param[out] ipcs Where to store newly created IPC server
* \param[in] cb IPC callbacks
*
* \note This function exits with CRM_EX_OSERR if unable to create the servers.
*/
void
pcmk__serve_pacemakerd_ipc(qb_ipcs_service_t **ipcs,
struct qb_ipcs_service_handlers *cb)
{
*ipcs = mainloop_add_ipc_server(CRM_SYSTEM_MCP, QB_IPC_NATIVE, cb);
if (*ipcs == NULL) {
crm_err("Couldn't start pacemakerd IPC server");
crm_warn("Verify pacemaker and pacemaker_remote are not both enabled.");
/* sub-daemons are observed by pacemakerd. Thus we exit CRM_EX_FATAL
* if we want to prevent pacemakerd from restarting them.
* With pacemakerd we leave the exit-code shown to e.g. systemd
* to what it was prior to moving the code here from pacemakerd.c
*/
crm_exit(CRM_EX_OSERR);
}
}
/*!
* \internal
* \brief Add an IPC server to the main loop for the scheduler API
*
* \param[in] cb IPC callbacks
*
* \return Newly created IPC server
* \note This function exits fatally if unable to create the servers.
*/
qb_ipcs_service_t *
pcmk__serve_schedulerd_ipc(struct qb_ipcs_service_handlers *cb)
{
return mainloop_add_ipc_server(CRM_SYSTEM_PENGINE, QB_IPC_NATIVE, cb);
}
diff --git a/lib/common/mainloop.c b/lib/common/mainloop.c
index dce30a030c..fc115e7dff 100644
--- a/lib/common/mainloop.c
+++ b/lib/common/mainloop.c
@@ -1,1465 +1,1471 @@
/*
* Copyright 2004-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
struct mainloop_child_s {
pid_t pid;
char *desc;
unsigned timerid;
gboolean timeout;
void *privatedata;
enum mainloop_child_flags flags;
/* Called when a process dies */
void (*callback) (mainloop_child_t * p, pid_t pid, int core, int signo, int exitcode);
};
struct trigger_s {
GSource source;
gboolean running;
gboolean trigger;
void *user_data;
guint id;
};
struct mainloop_timer_s {
guint id;
guint period_ms;
bool repeat;
char *name;
GSourceFunc cb;
void *userdata;
};
static gboolean
crm_trigger_prepare(GSource * source, gint * timeout)
{
crm_trigger_t *trig = (crm_trigger_t *) source;
/* cluster-glue's FD and IPC related sources make use of
* g_source_add_poll() but do not set a timeout in their prepare
* functions
*
* This means mainloop's poll() will block until an event for one
* of these sources occurs - any /other/ type of source, such as
* this one or g_idle_*, that doesn't use g_source_add_poll() is
* S-O-L and won't be processed until there is something fd-based
* happens.
*
* Luckily the timeout we can set here affects all sources and
* puts an upper limit on how long poll() can take.
*
* So unconditionally set a small-ish timeout, not too small that
* we're in constant motion, which will act as an upper bound on
* how long the signal handling might be delayed for.
*/
*timeout = 500; /* Timeout in ms */
return trig->trigger;
}
static gboolean
crm_trigger_check(GSource * source)
{
crm_trigger_t *trig = (crm_trigger_t *) source;
return trig->trigger;
}
/*!
* \internal
* \brief GSource dispatch function for crm_trigger_t
*
* \param[in] source crm_trigger_t being dispatched
* \param[in] callback Callback passed at source creation
* \param[in,out] userdata User data passed at source creation
*
* \return G_SOURCE_REMOVE to remove source, G_SOURCE_CONTINUE to keep it
*/
static gboolean
crm_trigger_dispatch(GSource *source, GSourceFunc callback, gpointer userdata)
{
gboolean rc = G_SOURCE_CONTINUE;
crm_trigger_t *trig = (crm_trigger_t *) source;
if (trig->running) {
/* Wait until the existing job is complete before starting the next one */
return G_SOURCE_CONTINUE;
}
trig->trigger = FALSE;
if (callback) {
int callback_rc = callback(trig->user_data);
if (callback_rc < 0) {
crm_trace("Trigger handler %p not yet complete", trig);
trig->running = TRUE;
} else if (callback_rc == 0) {
rc = G_SOURCE_REMOVE;
}
}
return rc;
}
static void
crm_trigger_finalize(GSource * source)
{
crm_trace("Trigger %p destroyed", source);
}
static GSourceFuncs crm_trigger_funcs = {
crm_trigger_prepare,
crm_trigger_check,
crm_trigger_dispatch,
crm_trigger_finalize,
};
static crm_trigger_t *
mainloop_setup_trigger(GSource * source, int priority, int (*dispatch) (gpointer user_data),
gpointer userdata)
{
crm_trigger_t *trigger = NULL;
trigger = (crm_trigger_t *) source;
trigger->id = 0;
trigger->trigger = FALSE;
trigger->user_data = userdata;
if (dispatch) {
g_source_set_callback(source, dispatch, trigger, NULL);
}
g_source_set_priority(source, priority);
g_source_set_can_recurse(source, FALSE);
trigger->id = g_source_attach(source, NULL);
return trigger;
}
void
mainloop_trigger_complete(crm_trigger_t * trig)
{
crm_trace("Trigger handler %p complete", trig);
trig->running = FALSE;
}
/*!
* \brief Create a trigger to be used as a mainloop source
*
* \param[in] priority Relative priority of source (lower number is higher priority)
* \param[in] dispatch Trigger dispatch function (should return 0 to remove the
* trigger from the mainloop, -1 if the trigger should be
* kept but the job is still running and not complete, and
* 1 if the trigger should be kept and the job is complete)
* \param[in] userdata Pointer to pass to \p dispatch
*
* \return Newly allocated mainloop source for trigger
*/
crm_trigger_t *
mainloop_add_trigger(int priority, int (*dispatch) (gpointer user_data),
gpointer userdata)
{
GSource *source = NULL;
pcmk__assert(sizeof(crm_trigger_t) > sizeof(GSource));
source = g_source_new(&crm_trigger_funcs, sizeof(crm_trigger_t));
return mainloop_setup_trigger(source, priority, dispatch, userdata);
}
void
mainloop_set_trigger(crm_trigger_t * source)
{
if(source) {
source->trigger = TRUE;
}
}
gboolean
mainloop_destroy_trigger(crm_trigger_t * source)
{
GSource *gs = NULL;
if(source == NULL) {
return TRUE;
}
gs = (GSource *)source;
g_source_destroy(gs); /* Remove from mainloop, ref_count-- */
g_source_unref(gs); /* The caller no longer carries a reference to source
*
* At this point the source should be free'd,
* unless we're currently processing said
* source, in which case mainloop holds an
* additional reference and it will be free'd
* once our processing completes
*/
return TRUE;
}
// Define a custom glib source for signal handling
// Data structure for custom glib source
typedef struct signal_s {
crm_trigger_t trigger; // trigger that invoked source (must be first)
void (*handler) (int sig); // signal handler
int signal; // signal that was received
} crm_signal_t;
// Table to associate signal handlers with signal numbers
static crm_signal_t *crm_signals[NSIG];
/*!
* \internal
* \brief Dispatch an event from custom glib source for signals
*
* Given an signal event, clear the event trigger and call any registered
* signal handler.
*
* \param[in] source glib source that triggered this dispatch
* \param[in] callback (ignored)
* \param[in] userdata (ignored)
*/
static gboolean
crm_signal_dispatch(GSource *source, GSourceFunc callback, gpointer userdata)
{
crm_signal_t *sig = (crm_signal_t *) source;
if(sig->signal != SIGCHLD) {
crm_notice("Caught '%s' signal " QB_XS " %d (%s handler)",
strsignal(sig->signal), sig->signal,
(sig->handler? "invoking" : "no"));
}
sig->trigger.trigger = FALSE;
if (sig->handler) {
sig->handler(sig->signal);
}
return TRUE;
}
/*!
* \internal
* \brief Handle a signal by setting a trigger for signal source
*
* \param[in] sig Signal number that was received
*
* \note This is the true signal handler for the mainloop signal source, and
* must be async-safe.
*/
static void
mainloop_signal_handler(int sig)
{
if (sig > 0 && sig < NSIG && crm_signals[sig] != NULL) {
mainloop_set_trigger((crm_trigger_t *) crm_signals[sig]);
}
}
// Functions implementing our custom glib source for signal handling
static GSourceFuncs crm_signal_funcs = {
crm_trigger_prepare,
crm_trigger_check,
crm_signal_dispatch,
crm_trigger_finalize,
};
/*!
* \internal
* \brief Set a true signal handler
*
* signal()-like interface to sigaction()
*
* \param[in] sig Signal number to register handler for
* \param[in] dispatch Signal handler
*
* \return The previous value of the signal handler, or SIG_ERR on error
* \note The dispatch function must be async-safe.
*/
sighandler_t
crm_signal_handler(int sig, sighandler_t dispatch)
{
sigset_t mask;
struct sigaction sa;
struct sigaction old;
if (sigemptyset(&mask) < 0) {
crm_err("Could not set handler for signal %d: %s",
sig, pcmk_rc_str(errno));
return SIG_ERR;
}
memset(&sa, 0, sizeof(struct sigaction));
sa.sa_handler = dispatch;
sa.sa_flags = SA_RESTART;
sa.sa_mask = mask;
if (sigaction(sig, &sa, &old) < 0) {
crm_err("Could not set handler for signal %d: %s",
sig, pcmk_rc_str(errno));
return SIG_ERR;
}
return old.sa_handler;
}
static void
mainloop_destroy_signal_entry(int sig)
{
crm_signal_t *tmp = crm_signals[sig];
if (tmp != NULL) {
crm_signals[sig] = NULL;
crm_trace("Unregistering mainloop handler for signal %d", sig);
mainloop_destroy_trigger((crm_trigger_t *) tmp);
}
}
/*!
* \internal
* \brief Add a signal handler to a mainloop
*
* \param[in] sig Signal number to handle
* \param[in] dispatch Signal handler function
*
* \note The true signal handler merely sets a mainloop trigger to call this
* dispatch function via the mainloop. Therefore, the dispatch function
* does not need to be async-safe.
*/
gboolean
mainloop_add_signal(int sig, void (*dispatch) (int sig))
{
GSource *source = NULL;
int priority = G_PRIORITY_HIGH - 1;
if (sig == SIGTERM) {
/* TERM is higher priority than other signals,
* signals are higher priority than other ipc.
* Yes, minus: smaller is "higher"
*/
priority--;
}
if (sig >= NSIG || sig < 0) {
crm_err("Signal %d is out of range", sig);
return FALSE;
} else if (crm_signals[sig] != NULL && crm_signals[sig]->handler == dispatch) {
crm_trace("Signal handler for %d is already installed", sig);
return TRUE;
} else if (crm_signals[sig] != NULL) {
crm_err("Different signal handler for %d is already installed", sig);
return FALSE;
}
pcmk__assert(sizeof(crm_signal_t) > sizeof(GSource));
source = g_source_new(&crm_signal_funcs, sizeof(crm_signal_t));
crm_signals[sig] = (crm_signal_t *) mainloop_setup_trigger(source, priority, NULL, NULL);
pcmk__assert(crm_signals[sig] != NULL);
crm_signals[sig]->handler = dispatch;
crm_signals[sig]->signal = sig;
if (crm_signal_handler(sig, mainloop_signal_handler) == SIG_ERR) {
mainloop_destroy_signal_entry(sig);
return FALSE;
}
return TRUE;
}
gboolean
mainloop_destroy_signal(int sig)
{
if (sig >= NSIG || sig < 0) {
crm_err("Signal %d is out of range", sig);
return FALSE;
} else if (crm_signal_handler(sig, NULL) == SIG_ERR) {
crm_perror(LOG_ERR, "Could not uninstall signal handler for signal %d", sig);
return FALSE;
} else if (crm_signals[sig] == NULL) {
return TRUE;
}
mainloop_destroy_signal_entry(sig);
return TRUE;
}
static qb_array_t *gio_map = NULL;
void
mainloop_cleanup(void)
{
if (gio_map != NULL) {
qb_array_free(gio_map);
gio_map = NULL;
}
for (int sig = 0; sig < NSIG; ++sig) {
mainloop_destroy_signal_entry(sig);
}
}
/*
* libqb...
*/
struct gio_to_qb_poll {
int32_t is_used;
guint source;
int32_t events;
void *data;
qb_ipcs_dispatch_fn_t fn;
enum qb_loop_priority p;
};
static gboolean
gio_read_socket(GIOChannel * gio, GIOCondition condition, gpointer data)
{
struct gio_to_qb_poll *adaptor = (struct gio_to_qb_poll *)data;
gint fd = g_io_channel_unix_get_fd(gio);
crm_trace("%p.%d %d", data, fd, condition);
/* if this assert get's hit, then there is a race condition between
* when we destroy a fd and when mainloop actually gives it up */
pcmk__assert(adaptor->is_used > 0);
return (adaptor->fn(fd, condition, adaptor->data) == 0);
}
static void
gio_poll_destroy(gpointer data)
{
struct gio_to_qb_poll *adaptor = (struct gio_to_qb_poll *)data;
adaptor->is_used--;
pcmk__assert(adaptor->is_used >= 0);
if (adaptor->is_used == 0) {
crm_trace("Marking adaptor %p unused", adaptor);
adaptor->source = 0;
}
}
/*!
* \internal
* \brief Convert libqb's poll priority into GLib's one
*
* \param[in] prio libqb's poll priority (#QB_LOOP_MED assumed as fallback)
*
* \return best matching GLib's priority
*/
static gint
conv_prio_libqb2glib(enum qb_loop_priority prio)
{
switch (prio) {
case QB_LOOP_LOW: return G_PRIORITY_LOW;
case QB_LOOP_HIGH: return G_PRIORITY_HIGH;
default: return G_PRIORITY_DEFAULT; // QB_LOOP_MED
}
}
/*!
* \internal
* \brief Convert libqb's poll priority to rate limiting spec
*
* \param[in] prio libqb's poll priority (#QB_LOOP_MED assumed as fallback)
*
* \return best matching rate limiting spec
* \note This is the inverse of libqb's qb_ipcs_request_rate_limit().
*/
static enum qb_ipcs_rate_limit
conv_libqb_prio2ratelimit(enum qb_loop_priority prio)
{
switch (prio) {
case QB_LOOP_LOW: return QB_IPCS_RATE_SLOW;
case QB_LOOP_HIGH: return QB_IPCS_RATE_FAST;
default: return QB_IPCS_RATE_NORMAL; // QB_LOOP_MED
}
}
static int32_t
gio_poll_dispatch_update(enum qb_loop_priority p, int32_t fd, int32_t evts,
void *data, qb_ipcs_dispatch_fn_t fn, int32_t add)
{
struct gio_to_qb_poll *adaptor;
GIOChannel *channel;
int32_t res = 0;
res = qb_array_index(gio_map, fd, (void **)&adaptor);
if (res < 0) {
crm_err("Array lookup failed for fd=%d: %d", fd, res);
return res;
}
crm_trace("Adding fd=%d to mainloop as adaptor %p", fd, adaptor);
if (add && adaptor->source) {
crm_err("Adaptor for descriptor %d is still in-use", fd);
return -EEXIST;
}
if (!add && !adaptor->is_used) {
crm_err("Adaptor for descriptor %d is not in-use", fd);
return -ENOENT;
}
/* channel is created with ref_count = 1 */
channel = g_io_channel_unix_new(fd);
if (!channel) {
crm_err("No memory left to add fd=%d", fd);
return -ENOMEM;
}
if (adaptor->source) {
g_source_remove(adaptor->source);
adaptor->source = 0;
}
/* Because unlike the poll() API, glib doesn't tell us about HUPs by default */
evts |= (G_IO_HUP | G_IO_NVAL | G_IO_ERR);
adaptor->fn = fn;
adaptor->events = evts;
adaptor->data = data;
adaptor->p = p;
adaptor->is_used++;
adaptor->source =
g_io_add_watch_full(channel, conv_prio_libqb2glib(p), evts,
gio_read_socket, adaptor, gio_poll_destroy);
/* Now that mainloop now holds a reference to channel,
* thanks to g_io_add_watch_full(), drop ours from g_io_channel_unix_new().
*
* This means that channel will be free'd by:
* g_main_context_dispatch()
* -> g_source_destroy_internal()
* -> g_source_callback_unref()
* shortly after gio_poll_destroy() completes
*/
g_io_channel_unref(channel);
crm_trace("Added to mainloop with gsource id=%d", adaptor->source);
if (adaptor->source > 0) {
return 0;
}
return -EINVAL;
}
static int32_t
gio_poll_dispatch_add(enum qb_loop_priority p, int32_t fd, int32_t evts,
void *data, qb_ipcs_dispatch_fn_t fn)
{
return gio_poll_dispatch_update(p, fd, evts, data, fn, QB_TRUE);
}
static int32_t
gio_poll_dispatch_mod(enum qb_loop_priority p, int32_t fd, int32_t evts,
void *data, qb_ipcs_dispatch_fn_t fn)
{
return gio_poll_dispatch_update(p, fd, evts, data, fn, QB_FALSE);
}
static int32_t
gio_poll_dispatch_del(int32_t fd)
{
struct gio_to_qb_poll *adaptor;
crm_trace("Looking for fd=%d", fd);
if (qb_array_index(gio_map, fd, (void **)&adaptor) == 0) {
if (adaptor->source) {
g_source_remove(adaptor->source);
adaptor->source = 0;
}
}
return 0;
}
struct qb_ipcs_poll_handlers gio_poll_funcs = {
.job_add = NULL,
.dispatch_add = gio_poll_dispatch_add,
.dispatch_mod = gio_poll_dispatch_mod,
.dispatch_del = gio_poll_dispatch_del,
};
static enum qb_ipc_type
pick_ipc_type(enum qb_ipc_type requested)
{
const char *env = pcmk__env_option(PCMK__ENV_IPC_TYPE);
if (env && strcmp("shared-mem", env) == 0) {
return QB_IPC_SHM;
} else if (env && strcmp("socket", env) == 0) {
return QB_IPC_SOCKET;
} else if (env && strcmp("posix", env) == 0) {
return QB_IPC_POSIX_MQ;
} else if (env && strcmp("sysv", env) == 0) {
return QB_IPC_SYSV_MQ;
} else if (requested == QB_IPC_NATIVE) {
/* We prefer shared memory because the server never blocks on
* send. If part of a message fits into the socket, libqb
* needs to block until the remainder can be sent also.
* Otherwise the client will wait forever for the remaining
* bytes.
*/
return QB_IPC_SHM;
}
return requested;
}
qb_ipcs_service_t *
mainloop_add_ipc_server(const char *name, enum qb_ipc_type type,
struct qb_ipcs_service_handlers *callbacks)
{
return mainloop_add_ipc_server_with_prio(name, type, callbacks, QB_LOOP_MED);
}
qb_ipcs_service_t *
mainloop_add_ipc_server_with_prio(const char *name, enum qb_ipc_type type,
struct qb_ipcs_service_handlers *callbacks,
enum qb_loop_priority prio)
{
int rc = 0;
qb_ipcs_service_t *server = NULL;
if (gio_map == NULL) {
gio_map = qb_array_create_2(64, sizeof(struct gio_to_qb_poll), 1);
}
server = qb_ipcs_create(name, 0, pick_ipc_type(type), callbacks);
if (server == NULL) {
crm_err("Could not create %s IPC server: %s (%d)",
name, pcmk_rc_str(errno), errno);
return NULL;
}
if (prio != QB_LOOP_MED) {
qb_ipcs_request_rate_limit(server, conv_libqb_prio2ratelimit(prio));
}
// Enforce a minimum IPC buffer size on all clients
qb_ipcs_enforce_buffer_size(server, crm_ipc_default_buffer_size());
qb_ipcs_poll_handlers_set(server, &gio_poll_funcs);
rc = qb_ipcs_run(server);
if (rc < 0) {
crm_err("Could not start %s IPC server: %s (%d)", name, pcmk_strerror(rc), rc);
return NULL; // qb_ipcs_run() destroys server on failure
}
return server;
}
void
mainloop_del_ipc_server(qb_ipcs_service_t * server)
{
if (server) {
qb_ipcs_destroy(server);
}
}
struct mainloop_io_s {
char *name;
void *userdata;
int fd;
guint source;
crm_ipc_t *ipc;
GIOChannel *channel;
int (*dispatch_fn_ipc) (const char *buffer, ssize_t length, gpointer userdata);
int (*dispatch_fn_io) (gpointer userdata);
void (*destroy_fn) (gpointer userdata);
};
/*!
* \internal
* \brief I/O watch callback function (GIOFunc)
*
* \param[in] gio I/O channel being watched
* \param[in] condition I/O condition satisfied
* \param[in] data User data passed when source was created
*
* \return G_SOURCE_REMOVE to remove source, G_SOURCE_CONTINUE to keep it
*/
static gboolean
mainloop_gio_callback(GIOChannel *gio, GIOCondition condition, gpointer data)
{
gboolean rc = G_SOURCE_CONTINUE;
mainloop_io_t *client = data;
pcmk__assert(client->fd == g_io_channel_unix_get_fd(gio));
if (condition & G_IO_IN) {
if (client->ipc) {
long read_rc = 0L;
int max = 10;
do {
read_rc = crm_ipc_read(client->ipc);
if (read_rc <= 0) {
crm_trace("Could not read IPC message from %s: %s (%ld)",
client->name, pcmk_strerror(read_rc), read_rc);
+ if (read_rc == -EAGAIN) {
+ continue;
+ }
+
} else if (client->dispatch_fn_ipc) {
const char *buffer = crm_ipc_buffer(client->ipc);
crm_trace("New %ld-byte IPC message from %s "
"after I/O condition %d",
read_rc, client->name, (int) condition);
if (client->dispatch_fn_ipc(buffer, read_rc, client->userdata) < 0) {
crm_trace("Connection to %s no longer required", client->name);
rc = G_SOURCE_REMOVE;
}
}
+ pcmk__ipc_free_client_buffer(client->ipc);
+
} while ((rc == G_SOURCE_CONTINUE) && (read_rc > 0) && --max > 0);
} else {
crm_trace("New I/O event for %s after I/O condition %d",
client->name, (int) condition);
if (client->dispatch_fn_io) {
if (client->dispatch_fn_io(client->userdata) < 0) {
crm_trace("Connection to %s no longer required", client->name);
rc = G_SOURCE_REMOVE;
}
}
}
}
if (client->ipc && !crm_ipc_connected(client->ipc)) {
crm_err("Connection to %s closed " QB_XS " client=%p condition=%d",
client->name, client, condition);
rc = G_SOURCE_REMOVE;
} else if (condition & (G_IO_HUP | G_IO_NVAL | G_IO_ERR)) {
crm_trace("The connection %s[%p] has been closed (I/O condition=%d)",
client->name, client, condition);
rc = G_SOURCE_REMOVE;
} else if ((condition & G_IO_IN) == 0) {
/*
#define GLIB_SYSDEF_POLLIN =1
#define GLIB_SYSDEF_POLLPRI =2
#define GLIB_SYSDEF_POLLOUT =4
#define GLIB_SYSDEF_POLLERR =8
#define GLIB_SYSDEF_POLLHUP =16
#define GLIB_SYSDEF_POLLNVAL =32
typedef enum
{
G_IO_IN GLIB_SYSDEF_POLLIN,
G_IO_OUT GLIB_SYSDEF_POLLOUT,
G_IO_PRI GLIB_SYSDEF_POLLPRI,
G_IO_ERR GLIB_SYSDEF_POLLERR,
G_IO_HUP GLIB_SYSDEF_POLLHUP,
G_IO_NVAL GLIB_SYSDEF_POLLNVAL
} GIOCondition;
A bitwise combination representing a condition to watch for on an event source.
G_IO_IN There is data to read.
G_IO_OUT Data can be written (without blocking).
G_IO_PRI There is urgent data to read.
G_IO_ERR Error condition.
G_IO_HUP Hung up (the connection has been broken, usually for pipes and sockets).
G_IO_NVAL Invalid request. The file descriptor is not open.
*/
crm_err("Strange condition: %d", condition);
}
/* G_SOURCE_REMOVE results in mainloop_gio_destroy() being called
* just before the source is removed from mainloop
*/
return rc;
}
static void
mainloop_gio_destroy(gpointer c)
{
mainloop_io_t *client = c;
char *c_name = strdup(client->name);
/* client->source is valid but about to be destroyed (ref_count == 0) in gmain.c
* client->channel will still have ref_count > 0... should be == 1
*/
crm_trace("Destroying client %s[%p]", c_name, c);
if (client->ipc) {
crm_ipc_close(client->ipc);
}
if (client->destroy_fn) {
void (*destroy_fn) (gpointer userdata) = client->destroy_fn;
client->destroy_fn = NULL;
destroy_fn(client->userdata);
}
if (client->ipc) {
crm_ipc_t *ipc = client->ipc;
client->ipc = NULL;
crm_ipc_destroy(ipc);
}
crm_trace("Destroyed client %s[%p]", c_name, c);
free(client->name); client->name = NULL;
free(client);
free(c_name);
}
/*!
* \brief Connect to IPC and add it as a main loop source
*
* \param[in,out] ipc IPC connection to add
* \param[in] priority Event source priority to use for connection
* \param[in] userdata Data to register with callbacks
* \param[in] callbacks Dispatch and destroy callbacks for connection
* \param[out] source Newly allocated event source
*
* \return Standard Pacemaker return code
*
* \note On failure, the caller is still responsible for ipc. On success, the
* caller should call mainloop_del_ipc_client() when source is no longer
* needed, which will lead to the disconnection of the IPC later in the
* main loop if it is connected. However the IPC disconnects,
* mainloop_gio_destroy() will free ipc and source after calling the
* destroy callback.
*/
int
pcmk__add_mainloop_ipc(crm_ipc_t *ipc, int priority, void *userdata,
const struct ipc_client_callbacks *callbacks,
mainloop_io_t **source)
{
int rc = pcmk_rc_ok;
int fd = -1;
const char *ipc_name = NULL;
CRM_CHECK((ipc != NULL) && (callbacks != NULL), return EINVAL);
ipc_name = pcmk__s(crm_ipc_name(ipc), "Pacemaker");
rc = pcmk__connect_generic_ipc(ipc);
if (rc != pcmk_rc_ok) {
crm_debug("Connection to %s failed: %s", ipc_name, pcmk_rc_str(rc));
return rc;
}
rc = pcmk__ipc_fd(ipc, &fd);
if (rc != pcmk_rc_ok) {
crm_debug("Could not obtain file descriptor for %s IPC: %s",
ipc_name, pcmk_rc_str(rc));
crm_ipc_close(ipc);
return rc;
}
*source = mainloop_add_fd(ipc_name, priority, fd, userdata, NULL);
if (*source == NULL) {
rc = errno;
crm_ipc_close(ipc);
return rc;
}
(*source)->ipc = ipc;
(*source)->destroy_fn = callbacks->destroy;
(*source)->dispatch_fn_ipc = callbacks->dispatch;
return pcmk_rc_ok;
}
/*!
* \brief Get period for mainloop timer
*
* \param[in] timer Timer
*
* \return Period in ms
*/
guint
pcmk__mainloop_timer_get_period(const mainloop_timer_t *timer)
{
if (timer) {
return timer->period_ms;
}
return 0;
}
mainloop_io_t *
mainloop_add_ipc_client(const char *name, int priority, size_t max_size,
void *userdata, struct ipc_client_callbacks *callbacks)
{
crm_ipc_t *ipc = crm_ipc_new(name, 0);
mainloop_io_t *source = NULL;
int rc = pcmk__add_mainloop_ipc(ipc, priority, userdata, callbacks,
&source);
if (rc != pcmk_rc_ok) {
if (crm_log_level == LOG_STDOUT) {
fprintf(stderr, "Connection to %s failed: %s",
name, pcmk_rc_str(rc));
}
crm_ipc_destroy(ipc);
if (rc > 0) {
errno = rc;
} else {
errno = ENOTCONN;
}
return NULL;
}
return source;
}
void
mainloop_del_ipc_client(mainloop_io_t * client)
{
mainloop_del_fd(client);
}
crm_ipc_t *
mainloop_get_ipc_client(mainloop_io_t * client)
{
if (client) {
return client->ipc;
}
return NULL;
}
mainloop_io_t *
mainloop_add_fd(const char *name, int priority, int fd, void *userdata,
struct mainloop_fd_callbacks * callbacks)
{
mainloop_io_t *client = NULL;
if (fd >= 0) {
client = calloc(1, sizeof(mainloop_io_t));
if (client == NULL) {
return NULL;
}
client->name = strdup(name);
client->userdata = userdata;
if (callbacks) {
client->destroy_fn = callbacks->destroy;
client->dispatch_fn_io = callbacks->dispatch;
}
client->fd = fd;
client->channel = g_io_channel_unix_new(fd);
client->source =
g_io_add_watch_full(client->channel, priority,
(G_IO_IN | G_IO_HUP | G_IO_NVAL | G_IO_ERR), mainloop_gio_callback,
client, mainloop_gio_destroy);
/* Now that mainloop now holds a reference to channel,
* thanks to g_io_add_watch_full(), drop ours from g_io_channel_unix_new().
*
* This means that channel will be free'd by:
* g_main_context_dispatch() or g_source_remove()
* -> g_source_destroy_internal()
* -> g_source_callback_unref()
* shortly after mainloop_gio_destroy() completes
*/
g_io_channel_unref(client->channel);
crm_trace("Added connection %d for %s[%p].%d", client->source, client->name, client, fd);
} else {
errno = EINVAL;
}
return client;
}
void
mainloop_del_fd(mainloop_io_t * client)
{
if (client != NULL) {
crm_trace("Removing client %s[%p]", client->name, client);
if (client->source) {
/* Results in mainloop_gio_destroy() being called just
* before the source is removed from mainloop
*/
g_source_remove(client->source);
}
}
}
static GList *child_list = NULL;
pid_t
mainloop_child_pid(mainloop_child_t * child)
{
return child->pid;
}
const char *
mainloop_child_name(mainloop_child_t * child)
{
return child->desc;
}
int
mainloop_child_timeout(mainloop_child_t * child)
{
return child->timeout;
}
void *
mainloop_child_userdata(mainloop_child_t * child)
{
return child->privatedata;
}
void
mainloop_clear_child_userdata(mainloop_child_t * child)
{
child->privatedata = NULL;
}
/* good function name */
static void
child_free(mainloop_child_t *child)
{
if (child->timerid != 0) {
crm_trace("Removing timer %d", child->timerid);
g_source_remove(child->timerid);
child->timerid = 0;
}
free(child->desc);
free(child);
}
/* terrible function name */
static int
child_kill_helper(mainloop_child_t *child)
{
int rc;
if (child->flags & mainloop_leave_pid_group) {
crm_debug("Kill pid %d only. leave group intact.", child->pid);
rc = kill(child->pid, SIGKILL);
} else {
crm_debug("Kill pid %d's group", child->pid);
rc = kill(-child->pid, SIGKILL);
}
if (rc < 0) {
if (errno != ESRCH) {
crm_perror(LOG_ERR, "kill(%d, KILL) failed", child->pid);
}
return -errno;
}
return 0;
}
static gboolean
child_timeout_callback(gpointer p)
{
mainloop_child_t *child = p;
int rc = 0;
child->timerid = 0;
if (child->timeout) {
crm_warn("%s process (PID %d) will not die!", child->desc, (int)child->pid);
return FALSE;
}
rc = child_kill_helper(child);
if (rc == -ESRCH) {
/* Nothing left to do. pid doesn't exist */
return FALSE;
}
child->timeout = TRUE;
crm_debug("%s process (PID %d) timed out", child->desc, (int)child->pid);
child->timerid = pcmk__create_timer(5000, child_timeout_callback, child);
return FALSE;
}
static bool
child_waitpid(mainloop_child_t *child, int flags)
{
int rc = 0;
int core = 0;
int signo = 0;
int status = 0;
int exitcode = 0;
bool callback_needed = true;
rc = waitpid(child->pid, &status, flags);
if (rc == 0) { // WNOHANG in flags, and child status is not available
crm_trace("Child process %d (%s) still active",
child->pid, child->desc);
callback_needed = false;
} else if (rc != child->pid) {
/* According to POSIX, possible conditions:
* - child->pid was non-positive (process group or any child),
* and rc is specific child
* - errno ECHILD (pid does not exist or is not child)
* - errno EINVAL (invalid flags)
* - errno EINTR (caller interrupted by signal)
*
* @TODO Handle these cases more specifically.
*/
signo = SIGCHLD;
exitcode = 1;
crm_notice("Wait for child process %d (%s) interrupted: %s",
child->pid, child->desc, pcmk_rc_str(errno));
} else if (WIFEXITED(status)) {
exitcode = WEXITSTATUS(status);
crm_trace("Child process %d (%s) exited with status %d",
child->pid, child->desc, exitcode);
} else if (WIFSIGNALED(status)) {
signo = WTERMSIG(status);
crm_trace("Child process %d (%s) exited with signal %d (%s)",
child->pid, child->desc, signo, strsignal(signo));
#ifdef WCOREDUMP // AIX, SunOS, maybe others
} else if (WCOREDUMP(status)) {
core = 1;
crm_err("Child process %d (%s) dumped core",
child->pid, child->desc);
#endif
} else { // flags must contain WUNTRACED and/or WCONTINUED to reach this
crm_trace("Child process %d (%s) stopped or continued",
child->pid, child->desc);
callback_needed = false;
}
if (callback_needed && child->callback) {
child->callback(child, child->pid, core, signo, exitcode);
}
return callback_needed;
}
static void
child_death_dispatch(int signal)
{
for (GList *iter = child_list; iter; ) {
GList *saved = iter;
mainloop_child_t *child = iter->data;
iter = iter->next;
if (child_waitpid(child, WNOHANG)) {
crm_trace("Removing completed process %d from child list",
child->pid);
child_list = g_list_remove_link(child_list, saved);
g_list_free(saved);
child_free(child);
}
}
}
static gboolean
child_signal_init(gpointer p)
{
crm_trace("Installed SIGCHLD handler");
/* Do NOT use g_child_watch_add() and friends, they rely on pthreads */
mainloop_add_signal(SIGCHLD, child_death_dispatch);
/* In case they terminated before the signal handler was installed */
child_death_dispatch(SIGCHLD);
return FALSE;
}
gboolean
mainloop_child_kill(pid_t pid)
{
GList *iter;
mainloop_child_t *child = NULL;
mainloop_child_t *match = NULL;
/* It is impossible to block SIGKILL, this allows us to
* call waitpid without WNOHANG flag.*/
int waitflags = 0, rc = 0;
for (iter = child_list; iter != NULL && match == NULL; iter = iter->next) {
child = iter->data;
if (pid == child->pid) {
match = child;
}
}
if (match == NULL) {
return FALSE;
}
rc = child_kill_helper(match);
if(rc == -ESRCH) {
/* It's gone, but hasn't shown up in waitpid() yet. Wait until we get
* SIGCHLD and let handler clean it up as normal (so we get the correct
* return code/status). The blocking alternative would be to call
* child_waitpid(match, 0).
*/
crm_trace("Waiting for signal that child process %d completed",
match->pid);
return TRUE;
} else if(rc != 0) {
/* If KILL for some other reason set the WNOHANG flag since we
* can't be certain what happened.
*/
waitflags = WNOHANG;
}
if (!child_waitpid(match, waitflags)) {
/* not much we can do if this occurs */
return FALSE;
}
child_list = g_list_remove(child_list, match);
child_free(match);
return TRUE;
}
/* Create/Log a new tracked process
* To track a process group, use -pid
*
* @TODO Using a non-positive pid (i.e. any child, or process group) would
* likely not be useful since we will free the child after the first
* completed process.
*/
void
mainloop_child_add_with_flags(pid_t pid, int timeout, const char *desc, void *privatedata, enum mainloop_child_flags flags,
void (*callback) (mainloop_child_t * p, pid_t pid, int core, int signo, int exitcode))
{
static bool need_init = TRUE;
mainloop_child_t *child = pcmk__assert_alloc(1, sizeof(mainloop_child_t));
child->pid = pid;
child->timerid = 0;
child->timeout = FALSE;
child->privatedata = privatedata;
child->callback = callback;
child->flags = flags;
child->desc = pcmk__str_copy(desc);
if (timeout) {
child->timerid = pcmk__create_timer(timeout, child_timeout_callback, child);
}
child_list = g_list_append(child_list, child);
if(need_init) {
need_init = FALSE;
/* SIGCHLD processing has to be invoked from mainloop.
* We do not want it to be possible to both add a child pid
* to mainloop, and have the pid's exit callback invoked within
* the same callstack. */
pcmk__create_timer(1, child_signal_init, NULL);
}
}
void
mainloop_child_add(pid_t pid, int timeout, const char *desc, void *privatedata,
void (*callback) (mainloop_child_t * p, pid_t pid, int core, int signo, int exitcode))
{
mainloop_child_add_with_flags(pid, timeout, desc, privatedata, 0, callback);
}
static gboolean
mainloop_timer_cb(gpointer user_data)
{
int id = 0;
bool repeat = FALSE;
struct mainloop_timer_s *t = user_data;
pcmk__assert(t != NULL);
id = t->id;
t->id = 0; /* Ensure it's unset during callbacks so that
* mainloop_timer_running() works as expected
*/
if(t->cb) {
crm_trace("Invoking callbacks for timer %s", t->name);
repeat = t->repeat;
if(t->cb(t->userdata) == FALSE) {
crm_trace("Timer %s complete", t->name);
repeat = FALSE;
}
}
if(repeat) {
/* Restore if repeating */
t->id = id;
}
return repeat;
}
bool
mainloop_timer_running(mainloop_timer_t *t)
{
if(t && t->id != 0) {
return TRUE;
}
return FALSE;
}
void
mainloop_timer_start(mainloop_timer_t *t)
{
mainloop_timer_stop(t);
if(t && t->period_ms > 0) {
crm_trace("Starting timer %s", t->name);
t->id = pcmk__create_timer(t->period_ms, mainloop_timer_cb, t);
}
}
void
mainloop_timer_stop(mainloop_timer_t *t)
{
if(t && t->id != 0) {
crm_trace("Stopping timer %s", t->name);
g_source_remove(t->id);
t->id = 0;
}
}
guint
mainloop_timer_set_period(mainloop_timer_t *t, guint period_ms)
{
guint last = 0;
if(t) {
last = t->period_ms;
t->period_ms = period_ms;
}
if(t && t->id != 0 && last != t->period_ms) {
mainloop_timer_start(t);
}
return last;
}
mainloop_timer_t *
mainloop_timer_add(const char *name, guint period_ms, bool repeat, GSourceFunc cb, void *userdata)
{
mainloop_timer_t *t = pcmk__assert_alloc(1, sizeof(mainloop_timer_t));
if (name != NULL) {
t->name = crm_strdup_printf("%s-%u-%d", name, period_ms, repeat);
} else {
t->name = crm_strdup_printf("%p-%u-%d", t, period_ms, repeat);
}
t->id = 0;
t->period_ms = period_ms;
t->repeat = repeat;
t->cb = cb;
t->userdata = userdata;
crm_trace("Created timer %s with %p %p", t->name, userdata, t->userdata);
return t;
}
void
mainloop_timer_del(mainloop_timer_t *t)
{
if(t) {
crm_trace("Destroying timer %s", t->name);
mainloop_timer_stop(t);
free(t->name);
free(t);
}
}
/*
* Helpers to make sure certain events aren't lost at shutdown
*/
static gboolean
drain_timeout_cb(gpointer user_data)
{
bool *timeout_popped = (bool*) user_data;
*timeout_popped = TRUE;
return FALSE;
}
/*!
* \brief Drain some remaining main loop events then quit it
*
* \param[in,out] mloop Main loop to drain and quit
* \param[in] n Drain up to this many pending events
*/
void
pcmk_quit_main_loop(GMainLoop *mloop, unsigned int n)
{
if ((mloop != NULL) && g_main_loop_is_running(mloop)) {
GMainContext *ctx = g_main_loop_get_context(mloop);
/* Drain up to n events in case some memory clean-up is pending
* (helpful to reduce noise in valgrind output).
*/
for (int i = 0; (i < n) && g_main_context_pending(ctx); ++i) {
g_main_context_dispatch(ctx);
}
g_main_loop_quit(mloop);
}
}
/*!
* \brief Process main loop events while a certain condition is met
*
* \param[in,out] mloop Main loop to process
* \param[in] timer_ms Don't process longer than this amount of time
* \param[in] check Function that returns true if events should be
* processed
*
* \note This function is intended to be called at shutdown if certain important
* events should not be missed. The caller would likely quit the main loop
* or exit after calling this function. The check() function will be
* passed the remaining timeout in milliseconds.
*/
void
pcmk_drain_main_loop(GMainLoop *mloop, guint timer_ms, bool (*check)(guint))
{
bool timeout_popped = FALSE;
guint timer = 0;
GMainContext *ctx = NULL;
CRM_CHECK(mloop && check, return);
ctx = g_main_loop_get_context(mloop);
if (ctx) {
time_t start_time = time(NULL);
timer = pcmk__create_timer(timer_ms, drain_timeout_cb, &timeout_popped);
while (!timeout_popped
&& check(timer_ms - (time(NULL) - start_time) * 1000)) {
g_main_context_iteration(ctx, TRUE);
}
}
if (!timeout_popped && (timer > 0)) {
g_source_remove(timer);
}
}
diff --git a/lib/common/results.c b/lib/common/results.c
index 568b8d448f..4001abf1ae 100644
--- a/lib/common/results.c
+++ b/lib/common/results.c
@@ -1,1205 +1,1209 @@
/*
* Copyright 2004-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
G_DEFINE_QUARK(pcmk-rc-error-quark, pcmk__rc_error)
G_DEFINE_QUARK(pcmk-exitc-error-quark, pcmk__exitc_error)
// General (all result code types)
/*!
* \brief Get the name and description of a given result code
*
* A result code can be interpreted as a member of any one of several families.
*
* \param[in] code The result code to look up
* \param[in] type How \p code should be interpreted
* \param[out] name Where to store the result code's name
* \param[out] desc Where to store the result code's description
*
* \return Standard Pacemaker return code
*/
int
pcmk_result_get_strings(int code, enum pcmk_result_type type, const char **name,
const char **desc)
{
const char *code_name = NULL;
const char *code_desc = NULL;
switch (type) {
case pcmk_result_legacy:
code_name = pcmk_errorname(code);
code_desc = pcmk_strerror(code);
break;
case pcmk_result_rc:
code_name = pcmk_rc_name(code);
code_desc = pcmk_rc_str(code);
break;
case pcmk_result_exitcode:
code_name = crm_exit_name(code);
code_desc = crm_exit_str((crm_exit_t) code);
break;
default:
return pcmk_rc_undetermined;
}
if (name != NULL) {
*name = code_name;
}
if (desc != NULL) {
*desc = code_desc;
}
return pcmk_rc_ok;
}
/*!
* \internal
* \brief Get the lower and upper bounds of a result code family
*
* \param[in] type Type of result code
* \param[out] lower Where to store the lower bound
* \param[out] upper Where to store the upper bound
*
* \return Standard Pacemaker return code
*
* \note There is no true upper bound on standard Pacemaker return codes or
* legacy return codes. All system \p errno values are valid members of
* these result code families, and there is no global upper limit nor a
* constant by which to refer to the highest \p errno value on a given
* system.
*/
int
pcmk__result_bounds(enum pcmk_result_type type, int *lower, int *upper)
{
pcmk__assert((lower != NULL) && (upper != NULL));
switch (type) {
case pcmk_result_legacy:
*lower = pcmk_ok;
*upper = 256; // should be enough for almost any system error code
break;
case pcmk_result_rc:
*lower = pcmk_rc_error - pcmk__n_rc + 1;
*upper = 256;
break;
case pcmk_result_exitcode:
*lower = CRM_EX_OK;
*upper = CRM_EX_MAX;
break;
default:
*lower = 0;
*upper = -1;
return pcmk_rc_undetermined;
}
return pcmk_rc_ok;
}
/*!
* \internal
* \brief Log a failed assertion
*
* \param[in] file File making the assertion
* \param[in] function Function making the assertion
* \param[in] line Line of file making the assertion
* \param[in] assert_condition String representation of assertion
*/
static void
log_assertion_as(const char *file, const char *function, int line,
const char *assert_condition)
{
if (!pcmk__is_daemon) {
crm_enable_stderr(TRUE); // Make sure command-line user sees message
}
crm_err("%s: Triggered fatal assertion at %s:%d : %s",
function, file, line, assert_condition);
}
/* coverity[+kill] */
/*!
* \internal
* \brief Log a failed assertion and abort
*
* \param[in] file File making the assertion
* \param[in] function Function making the assertion
* \param[in] line Line of file making the assertion
* \param[in] assert_condition String representation of assertion
*
* \note This does not return
*/
_Noreturn void
pcmk__abort_as(const char *file, const char *function, int line,
const char *assert_condition)
{
log_assertion_as(file, function, line, assert_condition);
abort();
}
/* coverity[+kill] */
/*!
* \internal
* \brief Handle a failed assertion
*
* When called by a daemon, fork a child that aborts (to dump core), otherwise
* abort the current process.
*
* \param[in] file File making the assertion
* \param[in] function Function making the assertion
* \param[in] line Line of file making the assertion
* \param[in] assert_condition String representation of assertion
*/
static void
fail_assert_as(const char *file, const char *function, int line,
const char *assert_condition)
{
int status = 0;
pid_t pid = 0;
if (!pcmk__is_daemon) {
pcmk__abort_as(file, function, line, assert_condition); // No return
}
pid = fork();
switch (pid) {
case -1: // Fork failed
crm_warn("%s: Cannot dump core for non-fatal assertion at %s:%d "
": %s", function, file, line, assert_condition);
break;
case 0: // Child process: just abort to dump core
abort();
break;
default: // Parent process: wait for child
crm_err("%s: Forked child [%d] to record non-fatal assertion at "
"%s:%d : %s", function, pid, file, line, assert_condition);
crm_write_blackbox(SIGTRAP, NULL);
do {
if (waitpid(pid, &status, 0) == pid) {
return; // Child finished dumping core
}
} while (errno == EINTR);
if (errno == ECHILD) {
// crm_mon ignores SIGCHLD
crm_trace("Cannot wait on forked child [%d] "
"(SIGCHLD is probably ignored)", pid);
} else {
crm_err("Cannot wait on forked child [%d]: %s",
pid, pcmk_rc_str(errno));
}
break;
}
}
/* coverity[+kill] */
void
crm_abort(const char *file, const char *function, int line,
const char *assert_condition, gboolean do_core, gboolean do_fork)
{
if (!do_fork) {
pcmk__abort_as(file, function, line, assert_condition); // No return
} else if (do_core) {
fail_assert_as(file, function, line, assert_condition);
} else {
log_assertion_as(file, function, line, assert_condition);
}
}
// @COMPAT Legacy function return codes
//! \deprecated Use standard return codes and pcmk_rc_name() instead
const char *
pcmk_errorname(int rc)
{
rc = abs(rc);
switch (rc) {
case pcmk_err_generic: return "pcmk_err_generic";
case pcmk_err_no_quorum: return "pcmk_err_no_quorum";
case pcmk_err_schema_validation: return "pcmk_err_schema_validation";
case pcmk_err_transform_failed: return "pcmk_err_transform_failed";
case pcmk_err_old_data: return "pcmk_err_old_data";
case pcmk_err_diff_failed: return "pcmk_err_diff_failed";
case pcmk_err_diff_resync: return "pcmk_err_diff_resync";
case pcmk_err_cib_modified: return "pcmk_err_cib_modified";
case pcmk_err_cib_backup: return "pcmk_err_cib_backup";
case pcmk_err_cib_save: return "pcmk_err_cib_save";
case pcmk_err_cib_corrupt: return "pcmk_err_cib_corrupt";
case pcmk_err_multiple: return "pcmk_err_multiple";
case pcmk_err_node_unknown: return "pcmk_err_node_unknown";
case pcmk_err_already: return "pcmk_err_already";
case pcmk_err_bad_nvpair: return "pcmk_err_bad_nvpair";
case pcmk_err_unknown_format: return "pcmk_err_unknown_format";
default: return pcmk_rc_name(rc); // system errno
}
}
//! \deprecated Use standard return codes and pcmk_rc_str() instead
const char *
pcmk_strerror(int rc)
{
return pcmk_rc_str(pcmk_legacy2rc(rc));
}
// Standard Pacemaker API return codes
/* This array is used only for nonzero values of pcmk_rc_e. Its values must be
* kept in the exact reverse order of the enum value numbering (i.e. add new
* values to the end of the array).
*/
static const struct pcmk__rc_info {
const char *name;
const char *desc;
int legacy_rc;
} pcmk__rcs[] = {
{ "pcmk_rc_error",
"Error",
-pcmk_err_generic,
},
{ "pcmk_rc_unknown_format",
"Unknown output format",
-pcmk_err_unknown_format,
},
{ "pcmk_rc_bad_nvpair",
"Bad name/value pair given",
-pcmk_err_bad_nvpair,
},
{ "pcmk_rc_already",
"Already in requested state",
-pcmk_err_already,
},
{ "pcmk_rc_node_unknown",
"Node not found",
-pcmk_err_node_unknown,
},
{ "pcmk_rc_multiple",
"Resource active on multiple nodes",
-pcmk_err_multiple,
},
{ "pcmk_rc_cib_corrupt",
"Could not parse on-disk configuration",
-pcmk_err_cib_corrupt,
},
{ "pcmk_rc_cib_save",
"Could not save new configuration to disk",
-pcmk_err_cib_save,
},
{ "pcmk_rc_cib_backup",
"Could not archive previous configuration",
-pcmk_err_cib_backup,
},
{ "pcmk_rc_cib_modified",
"On-disk configuration was manually modified",
-pcmk_err_cib_modified,
},
{ "pcmk_rc_diff_resync",
"Application of update diff failed, requesting full refresh",
-pcmk_err_diff_resync,
},
{ "pcmk_rc_diff_failed",
"Application of update diff failed",
-pcmk_err_diff_failed,
},
{ "pcmk_rc_old_data",
"Update was older than existing configuration",
-pcmk_err_old_data,
},
{ "pcmk_rc_transform_failed",
"Schema transform failed",
-pcmk_err_transform_failed,
},
{ "pcmk_rc_schema_unchanged",
"Schema is already the latest available",
-pcmk_err_schema_unchanged,
},
{ "pcmk_rc_schema_validation",
"Update does not conform to the configured schema",
-pcmk_err_schema_validation,
},
{ "pcmk_rc_no_quorum",
"Operation requires quorum",
-pcmk_err_no_quorum,
},
{ "pcmk_rc_ipc_unauthorized",
"IPC server is blocked by unauthorized process",
-pcmk_err_generic,
},
{ "pcmk_rc_ipc_unresponsive",
"IPC server is unresponsive",
-pcmk_err_generic,
},
{ "pcmk_rc_ipc_pid_only",
"IPC server process is active but not accepting connections",
-pcmk_err_generic,
},
{ "pcmk_rc_op_unsatisfied",
"Not applicable under current conditions",
-pcmk_err_generic,
},
{ "pcmk_rc_undetermined",
"Result undetermined",
-pcmk_err_generic,
},
{ "pcmk_rc_before_range",
"Result occurs before given range",
-pcmk_err_generic,
},
{ "pcmk_rc_within_range",
"Result occurs within given range",
-pcmk_err_generic,
},
{ "pcmk_rc_after_range",
"Result occurs after given range",
-pcmk_err_generic,
},
{ "pcmk_rc_no_output",
"Output message produced no output",
-pcmk_err_generic,
},
{ "pcmk_rc_no_input",
"Input file not available",
-pcmk_err_generic,
},
{ "pcmk_rc_underflow",
"Value too small to be stored in data type",
-pcmk_err_generic,
},
{ "pcmk_rc_dot_error",
"Error writing dot(1) file",
-pcmk_err_generic,
},
{ "pcmk_rc_graph_error",
"Error writing graph file",
-pcmk_err_generic,
},
{ "pcmk_rc_invalid_transition",
"Cluster simulation produced invalid transition",
-pcmk_err_generic,
},
{ "pcmk_rc_unpack_error",
"Unable to parse CIB XML",
-pcmk_err_generic,
},
{ "pcmk_rc_duplicate_id",
"Two or more XML elements have the same ID",
-pcmk_err_generic,
},
{ "pcmk_rc_disabled",
"Disabled",
-pcmk_err_generic,
},
{ "pcmk_rc_bad_input",
"Bad input value provided",
-pcmk_err_generic,
},
{ "pcmk_rc_bad_xml_patch",
"Bad XML patch format",
-pcmk_err_generic,
},
{ "pcmk_rc_no_transaction",
"No active transaction found",
-pcmk_err_generic,
},
{ "pcmk_rc_ns_resolution",
"Nameserver resolution error",
-pcmk_err_generic,
},
{ "pcmk_rc_compression",
"Compression/decompression error",
-pcmk_err_generic,
},
{ "pcmk_rc_no_dc",
"DC is not yet elected",
-pcmk_err_generic,
},
+ { "pcmk_rc_ipc_more",
+ "More IPC message fragments to send",
+ -pcmk_err_generic,
+ },
};
/*!
* \internal
* \brief The number of enum pcmk_rc_e values, excluding \c pcmk_rc_ok
*
* This constant stores the number of negative standard Pacemaker return codes.
* These represent Pacemaker-custom error codes. The count does not include
* positive system error numbers, nor does it include \c pcmk_rc_ok (success).
*/
const size_t pcmk__n_rc = PCMK__NELEM(pcmk__rcs);
/*!
* \brief Get a return code constant name as a string
*
* \param[in] rc Integer return code to convert
*
* \return String of constant name corresponding to rc
*/
const char *
pcmk_rc_name(int rc)
{
if ((rc <= pcmk_rc_error) && ((pcmk_rc_error - rc) < pcmk__n_rc)) {
return pcmk__rcs[pcmk_rc_error - rc].name;
}
switch (rc) {
case pcmk_rc_ok: return "pcmk_rc_ok";
case E2BIG: return "E2BIG";
case EACCES: return "EACCES";
case EADDRINUSE: return "EADDRINUSE";
case EADDRNOTAVAIL: return "EADDRNOTAVAIL";
case EAFNOSUPPORT: return "EAFNOSUPPORT";
case EAGAIN: return "EAGAIN";
case EALREADY: return "EALREADY";
case EBADF: return "EBADF";
case EBADMSG: return "EBADMSG";
case EBUSY: return "EBUSY";
case ECANCELED: return "ECANCELED";
case ECHILD: return "ECHILD";
case ECOMM: return "ECOMM";
case ECONNABORTED: return "ECONNABORTED";
case ECONNREFUSED: return "ECONNREFUSED";
case ECONNRESET: return "ECONNRESET";
/* case EDEADLK: return "EDEADLK"; */
case EDESTADDRREQ: return "EDESTADDRREQ";
case EDOM: return "EDOM";
case EDQUOT: return "EDQUOT";
case EEXIST: return "EEXIST";
case EFAULT: return "EFAULT";
case EFBIG: return "EFBIG";
case EHOSTDOWN: return "EHOSTDOWN";
case EHOSTUNREACH: return "EHOSTUNREACH";
case EIDRM: return "EIDRM";
case EILSEQ: return "EILSEQ";
case EINPROGRESS: return "EINPROGRESS";
case EINTR: return "EINTR";
case EINVAL: return "EINVAL";
case EIO: return "EIO";
case EISCONN: return "EISCONN";
case EISDIR: return "EISDIR";
case ELIBACC: return "ELIBACC";
case ELOOP: return "ELOOP";
case EMFILE: return "EMFILE";
case EMLINK: return "EMLINK";
case EMSGSIZE: return "EMSGSIZE";
#ifdef EMULTIHOP // Not available on OpenBSD
case EMULTIHOP: return "EMULTIHOP";
#endif
case ENAMETOOLONG: return "ENAMETOOLONG";
case ENETDOWN: return "ENETDOWN";
case ENETRESET: return "ENETRESET";
case ENETUNREACH: return "ENETUNREACH";
case ENFILE: return "ENFILE";
case ENOBUFS: return "ENOBUFS";
case ENODATA: return "ENODATA";
case ENODEV: return "ENODEV";
case ENOENT: return "ENOENT";
case ENOEXEC: return "ENOEXEC";
case ENOKEY: return "ENOKEY";
case ENOLCK: return "ENOLCK";
#ifdef ENOLINK // Not available on OpenBSD
case ENOLINK: return "ENOLINK";
#endif
case ENOMEM: return "ENOMEM";
case ENOMSG: return "ENOMSG";
case ENOPROTOOPT: return "ENOPROTOOPT";
case ENOSPC: return "ENOSPC";
#ifdef ENOSR
case ENOSR: return "ENOSR";
#endif
#ifdef ENOSTR
case ENOSTR: return "ENOSTR";
#endif
case ENOSYS: return "ENOSYS";
case ENOTBLK: return "ENOTBLK";
case ENOTCONN: return "ENOTCONN";
case ENOTDIR: return "ENOTDIR";
case ENOTEMPTY: return "ENOTEMPTY";
case ENOTSOCK: return "ENOTSOCK";
#if ENOTSUP != EOPNOTSUPP
case ENOTSUP: return "ENOTSUP";
#endif
case ENOTTY: return "ENOTTY";
case ENOTUNIQ: return "ENOTUNIQ";
case ENXIO: return "ENXIO";
case EOPNOTSUPP: return "EOPNOTSUPP";
case EOVERFLOW: return "EOVERFLOW";
case EPERM: return "EPERM";
case EPFNOSUPPORT: return "EPFNOSUPPORT";
case EPIPE: return "EPIPE";
case EPROTO: return "EPROTO";
case EPROTONOSUPPORT: return "EPROTONOSUPPORT";
case EPROTOTYPE: return "EPROTOTYPE";
case ERANGE: return "ERANGE";
case EREMOTE: return "EREMOTE";
case EREMOTEIO: return "EREMOTEIO";
case EROFS: return "EROFS";
case ESHUTDOWN: return "ESHUTDOWN";
case ESPIPE: return "ESPIPE";
case ESOCKTNOSUPPORT: return "ESOCKTNOSUPPORT";
case ESRCH: return "ESRCH";
case ESTALE: return "ESTALE";
case ETIME: return "ETIME";
case ETIMEDOUT: return "ETIMEDOUT";
case ETXTBSY: return "ETXTBSY";
#ifdef EUNATCH
case EUNATCH: return "EUNATCH";
#endif
case EUSERS: return "EUSERS";
/* case EWOULDBLOCK: return "EWOULDBLOCK"; */
case EXDEV: return "EXDEV";
#ifdef EBADE // Not available on OS X
case EBADE: return "EBADE";
case EBADFD: return "EBADFD";
case EBADSLT: return "EBADSLT";
case EDEADLOCK: return "EDEADLOCK";
case EBADR: return "EBADR";
case EBADRQC: return "EBADRQC";
case ECHRNG: return "ECHRNG";
#ifdef EISNAM // Not available on OS X, Illumos, Solaris
case EISNAM: return "EISNAM";
case EKEYEXPIRED: return "EKEYEXPIRED";
case EKEYREVOKED: return "EKEYREVOKED";
#endif
case EKEYREJECTED: return "EKEYREJECTED";
case EL2HLT: return "EL2HLT";
case EL2NSYNC: return "EL2NSYNC";
case EL3HLT: return "EL3HLT";
case EL3RST: return "EL3RST";
case ELIBBAD: return "ELIBBAD";
case ELIBMAX: return "ELIBMAX";
case ELIBSCN: return "ELIBSCN";
case ELIBEXEC: return "ELIBEXEC";
#ifdef ENOMEDIUM // Not available on OS X, Illumos, Solaris
case ENOMEDIUM: return "ENOMEDIUM";
case EMEDIUMTYPE: return "EMEDIUMTYPE";
#endif
case ENONET: return "ENONET";
case ENOPKG: return "ENOPKG";
case EREMCHG: return "EREMCHG";
case ERESTART: return "ERESTART";
case ESTRPIPE: return "ESTRPIPE";
#ifdef EUCLEAN // Not available on OS X, Illumos, Solaris
case EUCLEAN: return "EUCLEAN";
#endif
case EXFULL: return "EXFULL";
#endif // EBADE
default: return "Unknown";
}
}
/*!
* \brief Get a user-friendly description of a return code
*
* \param[in] rc Integer return code to convert
*
* \return String description of rc
*/
const char *
pcmk_rc_str(int rc)
{
if (rc == pcmk_rc_ok) {
return "OK";
}
if ((rc <= pcmk_rc_error) && ((pcmk_rc_error - rc) < pcmk__n_rc)) {
return pcmk__rcs[pcmk_rc_error - rc].desc;
}
if (rc < 0) {
return "Error";
}
// Handle values that could be defined by system or by portability.h
switch (rc) {
#ifdef PCMK__ENOTUNIQ
case ENOTUNIQ: return "Name not unique on network";
#endif
#ifdef PCMK__ECOMM
case ECOMM: return "Communication error on send";
#endif
#ifdef PCMK__ELIBACC
case ELIBACC: return "Can not access a needed shared library";
#endif
#ifdef PCMK__EREMOTEIO
case EREMOTEIO: return "Remote I/O error";
#endif
#ifdef PCMK__ENOKEY
case ENOKEY: return "Required key not available";
#endif
#ifdef PCMK__ENODATA
case ENODATA: return "No data available";
#endif
#ifdef PCMK__ETIME
case ETIME: return "Timer expired";
#endif
#ifdef PCMK__EKEYREJECTED
case EKEYREJECTED: return "Key was rejected by service";
#endif
default: return strerror(rc);
}
}
// This returns negative values for errors
//! \deprecated Use standard return codes instead
int
pcmk_rc2legacy(int rc)
{
if (rc >= 0) {
return -rc; // OK or system errno
}
if ((rc <= pcmk_rc_error) && ((pcmk_rc_error - rc) < pcmk__n_rc)) {
return pcmk__rcs[pcmk_rc_error - rc].legacy_rc;
}
return -pcmk_err_generic;
}
//! \deprecated Use standard return codes instead
int
pcmk_legacy2rc(int legacy_rc)
{
legacy_rc = abs(legacy_rc);
switch (legacy_rc) {
case pcmk_err_no_quorum: return pcmk_rc_no_quorum;
case pcmk_err_schema_validation: return pcmk_rc_schema_validation;
case pcmk_err_schema_unchanged: return pcmk_rc_schema_unchanged;
case pcmk_err_transform_failed: return pcmk_rc_transform_failed;
case pcmk_err_old_data: return pcmk_rc_old_data;
case pcmk_err_diff_failed: return pcmk_rc_diff_failed;
case pcmk_err_diff_resync: return pcmk_rc_diff_resync;
case pcmk_err_cib_modified: return pcmk_rc_cib_modified;
case pcmk_err_cib_backup: return pcmk_rc_cib_backup;
case pcmk_err_cib_save: return pcmk_rc_cib_save;
case pcmk_err_cib_corrupt: return pcmk_rc_cib_corrupt;
case pcmk_err_multiple: return pcmk_rc_multiple;
case pcmk_err_node_unknown: return pcmk_rc_node_unknown;
case pcmk_err_already: return pcmk_rc_already;
case pcmk_err_bad_nvpair: return pcmk_rc_bad_nvpair;
case pcmk_err_unknown_format: return pcmk_rc_unknown_format;
case pcmk_err_generic: return pcmk_rc_error;
case pcmk_ok: return pcmk_rc_ok;
default: return legacy_rc; // system errno
}
}
// Exit status codes
const char *
crm_exit_name(crm_exit_t exit_code)
{
switch (exit_code) {
case CRM_EX_OK: return "CRM_EX_OK";
case CRM_EX_ERROR: return "CRM_EX_ERROR";
case CRM_EX_INVALID_PARAM: return "CRM_EX_INVALID_PARAM";
case CRM_EX_UNIMPLEMENT_FEATURE: return "CRM_EX_UNIMPLEMENT_FEATURE";
case CRM_EX_INSUFFICIENT_PRIV: return "CRM_EX_INSUFFICIENT_PRIV";
case CRM_EX_NOT_INSTALLED: return "CRM_EX_NOT_INSTALLED";
case CRM_EX_NOT_CONFIGURED: return "CRM_EX_NOT_CONFIGURED";
case CRM_EX_NOT_RUNNING: return "CRM_EX_NOT_RUNNING";
case CRM_EX_PROMOTED: return "CRM_EX_PROMOTED";
case CRM_EX_FAILED_PROMOTED: return "CRM_EX_FAILED_PROMOTED";
case CRM_EX_USAGE: return "CRM_EX_USAGE";
case CRM_EX_DATAERR: return "CRM_EX_DATAERR";
case CRM_EX_NOINPUT: return "CRM_EX_NOINPUT";
case CRM_EX_NOUSER: return "CRM_EX_NOUSER";
case CRM_EX_NOHOST: return "CRM_EX_NOHOST";
case CRM_EX_UNAVAILABLE: return "CRM_EX_UNAVAILABLE";
case CRM_EX_SOFTWARE: return "CRM_EX_SOFTWARE";
case CRM_EX_OSERR: return "CRM_EX_OSERR";
case CRM_EX_OSFILE: return "CRM_EX_OSFILE";
case CRM_EX_CANTCREAT: return "CRM_EX_CANTCREAT";
case CRM_EX_IOERR: return "CRM_EX_IOERR";
case CRM_EX_TEMPFAIL: return "CRM_EX_TEMPFAIL";
case CRM_EX_PROTOCOL: return "CRM_EX_PROTOCOL";
case CRM_EX_NOPERM: return "CRM_EX_NOPERM";
case CRM_EX_CONFIG: return "CRM_EX_CONFIG";
case CRM_EX_FATAL: return "CRM_EX_FATAL";
case CRM_EX_PANIC: return "CRM_EX_PANIC";
case CRM_EX_DISCONNECT: return "CRM_EX_DISCONNECT";
case CRM_EX_DIGEST: return "CRM_EX_DIGEST";
case CRM_EX_NOSUCH: return "CRM_EX_NOSUCH";
case CRM_EX_QUORUM: return "CRM_EX_QUORUM";
case CRM_EX_UNSAFE: return "CRM_EX_UNSAFE";
case CRM_EX_EXISTS: return "CRM_EX_EXISTS";
case CRM_EX_MULTIPLE: return "CRM_EX_MULTIPLE";
case CRM_EX_EXPIRED: return "CRM_EX_EXPIRED";
case CRM_EX_NOT_YET_IN_EFFECT: return "CRM_EX_NOT_YET_IN_EFFECT";
case CRM_EX_INDETERMINATE: return "CRM_EX_INDETERMINATE";
case CRM_EX_UNSATISFIED: return "CRM_EX_UNSATISFIED";
case CRM_EX_NO_DC: return "CRM_EX_NO_DC";
case CRM_EX_OLD: return "CRM_EX_OLD";
case CRM_EX_TIMEOUT: return "CRM_EX_TIMEOUT";
case CRM_EX_DEGRADED: return "CRM_EX_DEGRADED";
case CRM_EX_DEGRADED_PROMOTED: return "CRM_EX_DEGRADED_PROMOTED";
case CRM_EX_NONE: return "CRM_EX_NONE";
case CRM_EX_MAX: return "CRM_EX_UNKNOWN";
}
return "CRM_EX_UNKNOWN";
}
const char *
crm_exit_str(crm_exit_t exit_code)
{
switch (exit_code) {
case CRM_EX_OK: return "OK";
case CRM_EX_ERROR: return "Error occurred";
case CRM_EX_INVALID_PARAM: return "Invalid parameter";
case CRM_EX_UNIMPLEMENT_FEATURE: return "Unimplemented";
case CRM_EX_INSUFFICIENT_PRIV: return "Insufficient privileges";
case CRM_EX_NOT_INSTALLED: return "Not installed";
case CRM_EX_NOT_CONFIGURED: return "Not configured";
case CRM_EX_NOT_RUNNING: return "Not running";
case CRM_EX_PROMOTED: return "Promoted";
case CRM_EX_FAILED_PROMOTED: return "Failed in promoted role";
case CRM_EX_USAGE: return "Incorrect usage";
case CRM_EX_DATAERR: return "Invalid data given";
case CRM_EX_NOINPUT: return "Input file not available";
case CRM_EX_NOUSER: return "User does not exist";
case CRM_EX_NOHOST: return "Host does not exist";
case CRM_EX_UNAVAILABLE: return "Necessary service unavailable";
case CRM_EX_SOFTWARE: return "Internal software bug";
case CRM_EX_OSERR: return "Operating system error occurred";
case CRM_EX_OSFILE: return "System file not available";
case CRM_EX_CANTCREAT: return "Cannot create output file";
case CRM_EX_IOERR: return "I/O error occurred";
case CRM_EX_TEMPFAIL: return "Temporary failure, try again";
case CRM_EX_PROTOCOL: return "Protocol violated";
case CRM_EX_NOPERM: return "Insufficient privileges";
case CRM_EX_CONFIG: return "Invalid configuration";
case CRM_EX_FATAL: return "Fatal error occurred, will not respawn";
case CRM_EX_PANIC: return "System panic required";
case CRM_EX_DISCONNECT: return "Not connected";
case CRM_EX_DIGEST: return "Digest mismatch";
case CRM_EX_NOSUCH: return "No such object";
case CRM_EX_QUORUM: return "Quorum required";
case CRM_EX_UNSAFE: return "Operation not safe";
case CRM_EX_EXISTS: return "Requested item already exists";
case CRM_EX_MULTIPLE: return "Multiple items match request";
case CRM_EX_EXPIRED: return "Requested item has expired";
case CRM_EX_NOT_YET_IN_EFFECT: return "Requested item is not yet in effect";
case CRM_EX_INDETERMINATE: return "Could not determine status";
case CRM_EX_UNSATISFIED: return "Not applicable under current conditions";
case CRM_EX_NO_DC: return "DC is not yet elected";
case CRM_EX_OLD: return "Update was older than existing configuration";
case CRM_EX_TIMEOUT: return "Timeout occurred";
case CRM_EX_DEGRADED: return "Service is active but might fail soon";
case CRM_EX_DEGRADED_PROMOTED: return "Service is promoted but might fail soon";
case CRM_EX_NONE: return "No exit status available";
case CRM_EX_MAX: return "Error occurred";
}
if ((exit_code > 128) && (exit_code < CRM_EX_MAX)) {
return "Interrupted by signal";
}
return "Unknown exit status";
}
/*!
* \brief Map a function return code to the most similar exit code
*
* \param[in] rc Function return code
*
* \return Most similar exit code
*/
crm_exit_t
pcmk_rc2exitc(int rc)
{
switch (rc) {
case pcmk_rc_ok:
case pcmk_rc_no_output: // quiet mode, or nothing to output
return CRM_EX_OK;
case pcmk_rc_no_quorum:
return CRM_EX_QUORUM;
case pcmk_rc_old_data:
return CRM_EX_OLD;
case pcmk_rc_cib_corrupt:
case pcmk_rc_schema_validation:
case pcmk_rc_transform_failed:
case pcmk_rc_unpack_error:
return CRM_EX_CONFIG;
case pcmk_rc_bad_nvpair:
return CRM_EX_INVALID_PARAM;
case EACCES:
return CRM_EX_INSUFFICIENT_PRIV;
case EBADF:
case EINVAL:
case EFAULT:
case ENOSYS:
case EOVERFLOW:
case pcmk_rc_underflow:
case pcmk_rc_compression:
return CRM_EX_SOFTWARE;
case EBADMSG:
case EMSGSIZE:
case ENOMSG:
case ENOPROTOOPT:
case EPROTO:
case EPROTONOSUPPORT:
case EPROTOTYPE:
return CRM_EX_PROTOCOL;
case ECOMM:
case ENOMEM:
return CRM_EX_OSERR;
case ECONNABORTED:
case ECONNREFUSED:
case ECONNRESET:
case ENOTCONN:
return CRM_EX_DISCONNECT;
case EEXIST:
case pcmk_rc_already:
return CRM_EX_EXISTS;
case EIO:
case pcmk_rc_dot_error:
case pcmk_rc_graph_error:
return CRM_EX_IOERR;
case ENOTSUP:
#if EOPNOTSUPP != ENOTSUP
case EOPNOTSUPP:
#endif
return CRM_EX_UNIMPLEMENT_FEATURE;
case ENOTUNIQ:
case pcmk_rc_multiple:
return CRM_EX_MULTIPLE;
case ENODEV:
case ENOENT:
case ENXIO:
case pcmk_rc_no_transaction:
case pcmk_rc_unknown_format:
return CRM_EX_NOSUCH;
case pcmk_rc_node_unknown:
case pcmk_rc_ns_resolution:
return CRM_EX_NOHOST;
case ETIME:
case ETIMEDOUT:
return CRM_EX_TIMEOUT;
case EAGAIN:
case EBUSY:
return CRM_EX_UNSATISFIED;
case pcmk_rc_before_range:
return CRM_EX_NOT_YET_IN_EFFECT;
case pcmk_rc_after_range:
return CRM_EX_EXPIRED;
case pcmk_rc_undetermined:
return CRM_EX_INDETERMINATE;
case pcmk_rc_op_unsatisfied:
return CRM_EX_UNSATISFIED;
case pcmk_rc_within_range:
return CRM_EX_OK;
case pcmk_rc_no_input:
return CRM_EX_NOINPUT;
case pcmk_rc_duplicate_id:
return CRM_EX_MULTIPLE;
case pcmk_rc_bad_input:
case pcmk_rc_bad_xml_patch:
return CRM_EX_DATAERR;
case pcmk_rc_no_dc:
return CRM_EX_NO_DC;
default:
return CRM_EX_ERROR;
}
}
/*!
* \brief Map a function return code to the most similar OCF exit code
*
* \param[in] rc Function return code
*
* \return Most similar OCF exit code
*/
enum ocf_exitcode
pcmk_rc2ocf(int rc)
{
switch (rc) {
case pcmk_rc_ok:
return PCMK_OCF_OK;
case pcmk_rc_bad_nvpair:
return PCMK_OCF_INVALID_PARAM;
case EACCES:
return PCMK_OCF_INSUFFICIENT_PRIV;
case ENOTSUP:
#if EOPNOTSUPP != ENOTSUP
case EOPNOTSUPP:
#endif
return PCMK_OCF_UNIMPLEMENT_FEATURE;
default:
return PCMK_OCF_UNKNOWN_ERROR;
}
}
// Other functions
/*!
* \brief Map a getaddrinfo() return code to the most similar Pacemaker
* return code
*
* \param[in] gai getaddrinfo() return code
*
* \return Most similar Pacemaker return code
*/
int
pcmk__gaierror2rc(int gai)
{
switch (gai) {
case 0:
return pcmk_rc_ok;
case EAI_AGAIN:
return EAGAIN;
case EAI_BADFLAGS:
case EAI_SERVICE:
return EINVAL;
case EAI_FAMILY:
return EAFNOSUPPORT;
case EAI_MEMORY:
return ENOMEM;
case EAI_NONAME:
return pcmk_rc_node_unknown;
case EAI_SOCKTYPE:
return ESOCKTNOSUPPORT;
case EAI_SYSTEM:
return errno;
default:
return pcmk_rc_ns_resolution;
}
}
/*!
* \brief Map a bz2 return code to the most similar Pacemaker return code
*
* \param[in] bz2 bz2 return code
*
* \return Most similar Pacemaker return code
*/
int
pcmk__bzlib2rc(int bz2)
{
switch (bz2) {
case BZ_OK:
case BZ_RUN_OK:
case BZ_FLUSH_OK:
case BZ_FINISH_OK:
case BZ_STREAM_END:
return pcmk_rc_ok;
case BZ_MEM_ERROR:
return ENOMEM;
case BZ_DATA_ERROR:
case BZ_DATA_ERROR_MAGIC:
case BZ_UNEXPECTED_EOF:
return pcmk_rc_bad_input;
case BZ_IO_ERROR:
return EIO;
case BZ_OUTBUFF_FULL:
return EFBIG;
default:
return pcmk_rc_compression;
}
}
crm_exit_t
crm_exit(crm_exit_t exit_status)
{
/* A compiler could theoretically use any type for crm_exit_t, but an int
* should always hold it, so cast to int to keep static analysis happy.
*/
if ((((int) exit_status) < 0) || (((int) exit_status) > CRM_EX_MAX)) {
exit_status = CRM_EX_ERROR;
}
crm_info("Exiting %s " QB_XS " with status %d (%s: %s)",
pcmk__s(crm_system_name, "process"), exit_status,
crm_exit_name(exit_status), crm_exit_str(exit_status));
pcmk_common_cleanup();
exit(exit_status);
}
/*
* External action results
*/
/*!
* \internal
* \brief Set the result of an action
*
* \param[out] result Where to set action result
* \param[in] exit_status OCF exit status to set
* \param[in] exec_status Execution status to set
* \param[in] exit_reason Human-friendly description of event to set
*/
void
pcmk__set_result(pcmk__action_result_t *result, int exit_status,
enum pcmk_exec_status exec_status, const char *exit_reason)
{
if (result == NULL) {
return;
}
result->exit_status = exit_status;
result->execution_status = exec_status;
if (!pcmk__str_eq(result->exit_reason, exit_reason, pcmk__str_none)) {
free(result->exit_reason);
result->exit_reason = (exit_reason == NULL)? NULL : strdup(exit_reason);
}
}
/*!
* \internal
* \brief Set the result of an action, with a formatted exit reason
*
* \param[out] result Where to set action result
* \param[in] exit_status OCF exit status to set
* \param[in] exec_status Execution status to set
* \param[in] format printf-style format for a human-friendly
* description of reason for result
* \param[in] ... arguments for \p format
*/
G_GNUC_PRINTF(4, 5)
void
pcmk__format_result(pcmk__action_result_t *result, int exit_status,
enum pcmk_exec_status exec_status,
const char *format, ...)
{
va_list ap;
int len = 0;
char *reason = NULL;
if (result == NULL) {
return;
}
result->exit_status = exit_status;
result->execution_status = exec_status;
if (format != NULL) {
va_start(ap, format);
len = vasprintf(&reason, format, ap);
pcmk__assert(len > 0);
va_end(ap);
}
free(result->exit_reason);
result->exit_reason = reason;
}
/*!
* \internal
* \brief Set the output of an action
*
* \param[out] result Action result to set output for
* \param[in] out Action output to set (must be dynamically
* allocated)
* \param[in] err Action error output to set (must be dynamically
* allocated)
*
* \note \p result will take ownership of \p out and \p err, so the caller
* should not free them.
*/
void
pcmk__set_result_output(pcmk__action_result_t *result, char *out, char *err)
{
if (result == NULL) {
return;
}
free(result->action_stdout);
result->action_stdout = out;
free(result->action_stderr);
result->action_stderr = err;
}
/*!
* \internal
* \brief Clear a result's exit reason, output, and error output
*
* \param[in,out] result Result to reset
*/
void
pcmk__reset_result(pcmk__action_result_t *result)
{
if (result == NULL) {
return;
}
free(result->exit_reason);
result->exit_reason = NULL;
free(result->action_stdout);
result->action_stdout = NULL;
free(result->action_stderr);
result->action_stderr = NULL;
}
/*!
* \internal
* \brief Copy the result of an action
*
* \param[in] src Result to copy
* \param[out] dst Where to copy \p src to
*/
void
pcmk__copy_result(const pcmk__action_result_t *src, pcmk__action_result_t *dst)
{
CRM_CHECK((src != NULL) && (dst != NULL), return);
dst->exit_status = src->exit_status;
dst->execution_status = src->execution_status;
dst->exit_reason = pcmk__str_copy(src->exit_reason);
dst->action_stdout = pcmk__str_copy(src->action_stdout);
dst->action_stderr = pcmk__str_copy(src->action_stderr);
}
diff --git a/lib/fencing/st_client.c b/lib/fencing/st_client.c
index 8122af17d9..41bc56c93d 100644
--- a/lib/fencing/st_client.c
+++ b/lib/fencing/st_client.c
@@ -1,2942 +1,2943 @@
/*
* Copyright 2004-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include // xmlNode
#include // xmlXPathObject, etc.
#include
#include
#include
#include
#include
#include "fencing_private.h"
CRM_TRACE_INIT_DATA(stonith);
// Used as stonith_t:st_private
typedef struct stonith_private_s {
char *token;
crm_ipc_t *ipc;
mainloop_io_t *source;
GHashTable *stonith_op_callback_table;
GList *notify_list;
int notify_refcnt;
bool notify_deletes;
void (*op_callback) (stonith_t * st, stonith_callback_data_t * data);
} stonith_private_t;
// Used as stonith_event_t:opaque
struct event_private {
pcmk__action_result_t result;
};
typedef struct stonith_notify_client_s {
const char *event;
const char *obj_id; /* implement one day */
const char *obj_type; /* implement one day */
void (*notify) (stonith_t * st, stonith_event_t * e);
bool delete;
} stonith_notify_client_t;
typedef struct stonith_callback_client_s {
void (*callback) (stonith_t * st, stonith_callback_data_t * data);
const char *id;
void *user_data;
gboolean only_success;
gboolean allow_timeout_updates;
struct timer_rec_s *timer;
} stonith_callback_client_t;
struct notify_blob_s {
stonith_t *stonith;
xmlNode *xml;
};
struct timer_rec_s {
int call_id;
int timeout;
guint ref;
stonith_t *stonith;
};
typedef int (*stonith_op_t) (const char *, int, const char *, xmlNode *,
xmlNode *, xmlNode *, xmlNode **, xmlNode **);
xmlNode *stonith_create_op(int call_id, const char *token, const char *op, xmlNode * data,
int call_options);
static int stonith_send_command(stonith_t *stonith, const char *op,
xmlNode *data, xmlNode **output_data,
int call_options, int timeout);
static void stonith_connection_destroy(gpointer user_data);
static void stonith_send_notification(gpointer data, gpointer user_data);
static int stonith_api_del_notification(stonith_t *stonith,
const char *event);
/*!
* \internal
* \brief Parse fence agent namespace from a string
*
* \param[in] namespace_s Name of namespace as string
*
* \return enum value parsed from \p namespace_s
*/
static enum stonith_namespace
parse_namespace(const char *namespace_s)
{
if (pcmk__str_eq(namespace_s, "any", pcmk__str_null_matches)) {
return st_namespace_any;
}
/* @TODO Is "redhat" still necessary except for stonith_text2namespace()
* backward compatibility?
*/
if (pcmk__str_any_of(namespace_s, "redhat", "stonith-ng", NULL)) {
return st_namespace_rhcs;
}
if (pcmk__str_eq(namespace_s, "internal", pcmk__str_none)) {
return st_namespace_internal;
}
if (pcmk__str_eq(namespace_s, "heartbeat", pcmk__str_none)) {
return st_namespace_lha;
}
return st_namespace_invalid;
}
/*!
* \internal
* \brief Get name of a fence agent namespace as a string
*
* \param[in] st_namespace Namespace as enum value
*
* \return Name of \p st_namespace as a string
*/
static const char *
namespace_text(enum stonith_namespace st_namespace)
{
switch (st_namespace) {
case st_namespace_any:
return "any";
case st_namespace_rhcs:
return "stonith-ng";
case st_namespace_internal:
return "internal";
case st_namespace_lha:
return "heartbeat";
default:
return "unsupported";
}
}
/*!
* \internal
* \brief Determine fence agent namespace from agent name
*
* This involves external checks (for example, checking the existence of a file
* or calling an external library function).
*
* \param[in] agent Fence agent name
*
* \return Namespace to which \p agent belongs, or \c st_namespace_invalid if
* not found
*/
static enum stonith_namespace
get_namespace_from_agent(const char *agent)
{
if (stonith__agent_is_rhcs(agent)) {
return st_namespace_rhcs;
}
#if HAVE_STONITH_STONITH_H
if (stonith__agent_is_lha(agent)) {
return st_namespace_lha;
}
#endif // HAVE_STONITH_STONITH_H
return st_namespace_invalid;
}
gboolean
stonith__watchdog_fencing_enabled_for_node_api(stonith_t *st, const char *node)
{
gboolean rv = FALSE;
stonith_t *stonith_api = (st != NULL)? st : stonith__api_new();
char *list = NULL;
if(stonith_api) {
if (stonith_api->state == stonith_disconnected) {
int rc = stonith_api->cmds->connect(stonith_api, "stonith-api", NULL);
if (rc != pcmk_ok) {
crm_err("Failed connecting to Stonith-API for watchdog-fencing-query.");
}
}
if (stonith_api->state != stonith_disconnected) {
/* caveat!!!
* this might fail when when stonithd is just updating the device-list
* probably something we should fix as well for other api-calls */
int rc = stonith_api->cmds->list(stonith_api, st_opt_sync_call, STONITH_WATCHDOG_ID, &list, 0);
if ((rc != pcmk_ok) || (list == NULL)) {
/* due to the race described above it can happen that
* we drop in here - so as not to make remote nodes
* panic on that answer
*/
if (rc == -ENODEV) {
crm_notice("Cluster does not have watchdog fencing device");
} else {
crm_warn("Could not check for watchdog fencing device: %s",
pcmk_strerror(rc));
}
} else if (list[0] == '\0') {
rv = TRUE;
} else {
GList *targets = stonith__parse_targets(list);
rv = pcmk__str_in_list(node, targets, pcmk__str_casei);
g_list_free_full(targets, free);
}
free(list);
if (!st) {
/* if we're provided the api we still might have done the
* connection - but let's assume the caller won't bother
*/
stonith_api->cmds->disconnect(stonith_api);
}
}
if (!st) {
stonith__api_free(stonith_api);
}
} else {
crm_err("Stonith-API for watchdog-fencing-query couldn't be created.");
}
crm_trace("Pacemaker assumes node %s %sto do watchdog-fencing.",
node, rv?"":"not ");
return rv;
}
gboolean
stonith__watchdog_fencing_enabled_for_node(const char *node)
{
return stonith__watchdog_fencing_enabled_for_node_api(NULL, node);
}
/* when cycling through the list we don't want to delete items
so just mark them and when we know nobody is using the list
loop over it to remove the marked items
*/
static void
foreach_notify_entry (stonith_private_t *private,
GFunc func,
gpointer user_data)
{
private->notify_refcnt++;
g_list_foreach(private->notify_list, func, user_data);
private->notify_refcnt--;
if ((private->notify_refcnt == 0) &&
private->notify_deletes) {
GList *list_item = private->notify_list;
private->notify_deletes = FALSE;
while (list_item != NULL)
{
stonith_notify_client_t *list_client = list_item->data;
GList *next = g_list_next(list_item);
if (list_client->delete) {
free(list_client);
private->notify_list =
g_list_delete_link(private->notify_list, list_item);
}
list_item = next;
}
}
}
static void
stonith_connection_destroy(gpointer user_data)
{
stonith_t *stonith = user_data;
stonith_private_t *native = NULL;
struct notify_blob_s blob;
crm_trace("Sending destroyed notification");
blob.stonith = stonith;
blob.xml = pcmk__xe_create(NULL, PCMK__XE_NOTIFY);
native = stonith->st_private;
native->ipc = NULL;
native->source = NULL;
free(native->token); native->token = NULL;
stonith->state = stonith_disconnected;
crm_xml_add(blob.xml, PCMK__XA_T, PCMK__VALUE_ST_NOTIFY);
crm_xml_add(blob.xml, PCMK__XA_SUBT, PCMK__VALUE_ST_NOTIFY_DISCONNECT);
foreach_notify_entry(native, stonith_send_notification, &blob);
pcmk__xml_free(blob.xml);
}
xmlNode *
create_device_registration_xml(const char *id, enum stonith_namespace standard,
const char *agent,
const stonith_key_value_t *params,
const char *rsc_provides)
{
xmlNode *data = pcmk__xe_create(NULL, PCMK__XE_ST_DEVICE_ID);
xmlNode *args = pcmk__xe_create(data, PCMK__XE_ATTRIBUTES);
#if HAVE_STONITH_STONITH_H
if (standard == st_namespace_any) {
standard = get_namespace_from_agent(agent);
}
if (standard == st_namespace_lha) {
hash2field((gpointer) "plugin", (gpointer) agent, args);
agent = "fence_legacy";
}
#endif
crm_xml_add(data, PCMK_XA_ID, id);
crm_xml_add(data, PCMK__XA_ST_ORIGIN, __func__);
crm_xml_add(data, PCMK_XA_AGENT, agent);
if ((standard != st_namespace_any) && (standard != st_namespace_invalid)) {
crm_xml_add(data, PCMK__XA_NAMESPACE, namespace_text(standard));
}
if (rsc_provides) {
crm_xml_add(data, PCMK__XA_RSC_PROVIDES, rsc_provides);
}
for (; params; params = params->next) {
hash2field((gpointer) params->key, (gpointer) params->value, args);
}
return data;
}
static int
stonith_api_register_device(stonith_t *st, int call_options,
const char *id, const char *namespace_s,
const char *agent,
const stonith_key_value_t *params)
{
int rc = 0;
xmlNode *data = NULL;
data = create_device_registration_xml(id, parse_namespace(namespace_s),
agent, params, NULL);
rc = stonith_send_command(st, STONITH_OP_DEVICE_ADD, data, NULL, call_options, 0);
pcmk__xml_free(data);
return rc;
}
static int
stonith_api_remove_device(stonith_t * st, int call_options, const char *name)
{
int rc = 0;
xmlNode *data = NULL;
data = pcmk__xe_create(NULL, PCMK__XE_ST_DEVICE_ID);
crm_xml_add(data, PCMK__XA_ST_ORIGIN, __func__);
crm_xml_add(data, PCMK_XA_ID, name);
rc = stonith_send_command(st, STONITH_OP_DEVICE_DEL, data, NULL, call_options, 0);
pcmk__xml_free(data);
return rc;
}
static int
stonith_api_remove_level_full(stonith_t *st, int options,
const char *node, const char *pattern,
const char *attr, const char *value, int level)
{
int rc = 0;
xmlNode *data = NULL;
CRM_CHECK(node || pattern || (attr && value), return -EINVAL);
data = pcmk__xe_create(NULL, PCMK_XE_FENCING_LEVEL);
crm_xml_add(data, PCMK__XA_ST_ORIGIN, __func__);
if (node) {
crm_xml_add(data, PCMK_XA_TARGET, node);
} else if (pattern) {
crm_xml_add(data, PCMK_XA_TARGET_PATTERN, pattern);
} else {
crm_xml_add(data, PCMK_XA_TARGET_ATTRIBUTE, attr);
crm_xml_add(data, PCMK_XA_TARGET_VALUE, value);
}
crm_xml_add_int(data, PCMK_XA_INDEX, level);
rc = stonith_send_command(st, STONITH_OP_LEVEL_DEL, data, NULL, options, 0);
pcmk__xml_free(data);
return rc;
}
static int
stonith_api_remove_level(stonith_t * st, int options, const char *node, int level)
{
return stonith_api_remove_level_full(st, options, node,
NULL, NULL, NULL, level);
}
/*!
* \internal
* \brief Create XML for fence topology level registration request
*
* \param[in] node If not NULL, target level by this node name
* \param[in] pattern If not NULL, target by node name using this regex
* \param[in] attr If not NULL, target by this node attribute
* \param[in] value If not NULL, target by this node attribute value
* \param[in] level Index number of level to register
* \param[in] device_list List of devices in level
*
* \return Newly allocated XML tree on success, NULL otherwise
*
* \note The caller should set only one of node, pattern or attr/value.
*/
xmlNode *
create_level_registration_xml(const char *node, const char *pattern,
const char *attr, const char *value,
int level, const stonith_key_value_t *device_list)
{
GString *list = NULL;
xmlNode *data;
CRM_CHECK(node || pattern || (attr && value), return NULL);
data = pcmk__xe_create(NULL, PCMK_XE_FENCING_LEVEL);
crm_xml_add(data, PCMK__XA_ST_ORIGIN, __func__);
crm_xml_add_int(data, PCMK_XA_ID, level);
crm_xml_add_int(data, PCMK_XA_INDEX, level);
if (node) {
crm_xml_add(data, PCMK_XA_TARGET, node);
} else if (pattern) {
crm_xml_add(data, PCMK_XA_TARGET_PATTERN, pattern);
} else {
crm_xml_add(data, PCMK_XA_TARGET_ATTRIBUTE, attr);
crm_xml_add(data, PCMK_XA_TARGET_VALUE, value);
}
for (; device_list; device_list = device_list->next) {
pcmk__add_separated_word(&list, 1024, device_list->value, ",");
}
if (list != NULL) {
crm_xml_add(data, PCMK_XA_DEVICES, (const char *) list->str);
g_string_free(list, TRUE);
}
return data;
}
static int
stonith_api_register_level_full(stonith_t *st, int options, const char *node,
const char *pattern, const char *attr,
const char *value, int level,
const stonith_key_value_t *device_list)
{
int rc = 0;
xmlNode *data = create_level_registration_xml(node, pattern, attr, value,
level, device_list);
CRM_CHECK(data != NULL, return -EINVAL);
rc = stonith_send_command(st, STONITH_OP_LEVEL_ADD, data, NULL, options, 0);
pcmk__xml_free(data);
return rc;
}
static int
stonith_api_register_level(stonith_t * st, int options, const char *node, int level,
const stonith_key_value_t * device_list)
{
return stonith_api_register_level_full(st, options, node, NULL, NULL, NULL,
level, device_list);
}
static int
stonith_api_device_list(stonith_t *stonith, int call_options,
const char *namespace_s, stonith_key_value_t **devices,
int timeout)
{
int count = 0;
enum stonith_namespace ns = parse_namespace(namespace_s);
if (devices == NULL) {
crm_err("Parameter error: stonith_api_device_list");
return -EFAULT;
}
#if HAVE_STONITH_STONITH_H
// Include Linux-HA agents if requested
if ((ns == st_namespace_any) || (ns == st_namespace_lha)) {
count += stonith__list_lha_agents(devices);
}
#endif
// Include Red Hat agents if requested
if ((ns == st_namespace_any) || (ns == st_namespace_rhcs)) {
count += stonith__list_rhcs_agents(devices);
}
return count;
}
// See stonith_api_operations_t:metadata() documentation
static int
stonith_api_device_metadata(stonith_t *stonith, int call_options,
const char *agent, const char *namespace_s,
char **output, int timeout_sec)
{
/* By executing meta-data directly, we can get it from stonith_admin when
* the cluster is not running, which is important for higher-level tools.
*/
enum stonith_namespace ns = get_namespace_from_agent(agent);
if (timeout_sec <= 0) {
timeout_sec = PCMK_DEFAULT_ACTION_TIMEOUT_MS;
}
crm_trace("Looking up metadata for %s agent %s", namespace_text(ns), agent);
switch (ns) {
case st_namespace_rhcs:
return stonith__rhcs_metadata(agent, timeout_sec, output);
#if HAVE_STONITH_STONITH_H
case st_namespace_lha:
return stonith__lha_metadata(agent, timeout_sec, output);
#endif
default:
crm_err("Can't get fence agent '%s' meta-data: No such agent",
agent);
break;
}
return -ENODEV;
}
static int
stonith_api_query(stonith_t * stonith, int call_options, const char *target,
stonith_key_value_t ** devices, int timeout)
{
int rc = 0, lpc = 0, max = 0;
xmlNode *data = NULL;
xmlNode *output = NULL;
xmlXPathObject *xpathObj = NULL;
CRM_CHECK(devices != NULL, return -EINVAL);
data = pcmk__xe_create(NULL, PCMK__XE_ST_DEVICE_ID);
crm_xml_add(data, PCMK__XA_ST_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_ST_TARGET, target);
crm_xml_add(data, PCMK__XA_ST_DEVICE_ACTION, PCMK_ACTION_OFF);
rc = stonith_send_command(stonith, STONITH_OP_QUERY, data, &output, call_options, timeout);
if (rc < 0) {
return rc;
}
xpathObj = pcmk__xpath_search(output->doc, "//*[@" PCMK_XA_AGENT "]");
if (xpathObj) {
max = pcmk__xpath_num_results(xpathObj);
for (lpc = 0; lpc < max; lpc++) {
xmlNode *match = pcmk__xpath_result(xpathObj, lpc);
CRM_LOG_ASSERT(match != NULL);
if(match != NULL) {
const char *match_id = crm_element_value(match, PCMK_XA_ID);
xmlChar *match_path = xmlGetNodePath(match);
crm_info("//*[@" PCMK_XA_AGENT "][%d] = %s", lpc, match_path);
free(match_path);
*devices = stonith__key_value_add(*devices, NULL, match_id);
}
}
xmlXPathFreeObject(xpathObj);
}
pcmk__xml_free(output);
pcmk__xml_free(data);
return max;
}
/*!
* \internal
* \brief Make a STONITH_OP_EXEC request
*
* \param[in,out] stonith Fencer connection
* \param[in] call_options Bitmask of \c stonith_call_options
* \param[in] id Fence device ID that request is for
* \param[in] action Agent action to request (list, status, monitor)
* \param[in] target Name of target node for requested action
* \param[in] timeout_sec Error if not completed within this many seconds
* \param[out] output Where to set agent output
*/
static int
stonith_api_call(stonith_t *stonith, int call_options, const char *id,
const char *action, const char *target, int timeout_sec,
xmlNode **output)
{
int rc = 0;
xmlNode *data = NULL;
data = pcmk__xe_create(NULL, PCMK__XE_ST_DEVICE_ID);
crm_xml_add(data, PCMK__XA_ST_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_ST_DEVICE_ID, id);
crm_xml_add(data, PCMK__XA_ST_DEVICE_ACTION, action);
crm_xml_add(data, PCMK__XA_ST_TARGET, target);
rc = stonith_send_command(stonith, STONITH_OP_EXEC, data, output,
call_options, timeout_sec);
pcmk__xml_free(data);
return rc;
}
static int
stonith_api_list(stonith_t * stonith, int call_options, const char *id, char **list_info,
int timeout)
{
int rc;
xmlNode *output = NULL;
rc = stonith_api_call(stonith, call_options, id, PCMK_ACTION_LIST, NULL,
timeout, &output);
if (output && list_info) {
const char *list_str;
list_str = crm_element_value(output, PCMK__XA_ST_OUTPUT);
if (list_str) {
*list_info = strdup(list_str);
}
}
if (output) {
pcmk__xml_free(output);
}
return rc;
}
static int
stonith_api_monitor(stonith_t * stonith, int call_options, const char *id, int timeout)
{
return stonith_api_call(stonith, call_options, id, PCMK_ACTION_MONITOR,
NULL, timeout, NULL);
}
static int
stonith_api_status(stonith_t * stonith, int call_options, const char *id, const char *port,
int timeout)
{
return stonith_api_call(stonith, call_options, id, PCMK_ACTION_STATUS, port,
timeout, NULL);
}
static int
stonith_api_fence_with_delay(stonith_t * stonith, int call_options, const char *node,
const char *action, int timeout, int tolerance, int delay)
{
int rc = 0;
xmlNode *data = NULL;
data = pcmk__xe_create(NULL, __func__);
crm_xml_add(data, PCMK__XA_ST_TARGET, node);
crm_xml_add(data, PCMK__XA_ST_DEVICE_ACTION, action);
crm_xml_add_int(data, PCMK__XA_ST_TIMEOUT, timeout);
crm_xml_add_int(data, PCMK__XA_ST_TOLERANCE, tolerance);
crm_xml_add_int(data, PCMK__XA_ST_DELAY, delay);
rc = stonith_send_command(stonith, STONITH_OP_FENCE, data, NULL, call_options, timeout);
pcmk__xml_free(data);
return rc;
}
static int
stonith_api_fence(stonith_t * stonith, int call_options, const char *node, const char *action,
int timeout, int tolerance)
{
return stonith_api_fence_with_delay(stonith, call_options, node, action,
timeout, tolerance, 0);
}
static int
stonith_api_confirm(stonith_t * stonith, int call_options, const char *target)
{
stonith__set_call_options(call_options, target, st_opt_manual_ack);
return stonith_api_fence(stonith, call_options, target, PCMK_ACTION_OFF, 0,
0);
}
static int
stonith_api_history(stonith_t * stonith, int call_options, const char *node,
stonith_history_t ** history, int timeout)
{
int rc = 0;
xmlNode *data = NULL;
xmlNode *output = NULL;
stonith_history_t *last = NULL;
*history = NULL;
if (node) {
data = pcmk__xe_create(NULL, __func__);
crm_xml_add(data, PCMK__XA_ST_TARGET, node);
}
stonith__set_call_options(call_options, node, st_opt_sync_call);
rc = stonith_send_command(stonith, STONITH_OP_FENCE_HISTORY, data, &output,
call_options, timeout);
pcmk__xml_free(data);
if (rc == 0) {
xmlNode *op = NULL;
xmlNode *reply = pcmk__xpath_find_one(output->doc,
"//" PCMK__XE_ST_HISTORY,
LOG_NEVER);
for (op = pcmk__xe_first_child(reply, NULL, NULL, NULL); op != NULL;
op = pcmk__xe_next(op, NULL)) {
stonith_history_t *kvp;
long long completed;
long long completed_nsec = 0L;
kvp = pcmk__assert_alloc(1, sizeof(stonith_history_t));
kvp->target = crm_element_value_copy(op, PCMK__XA_ST_TARGET);
kvp->action = crm_element_value_copy(op, PCMK__XA_ST_DEVICE_ACTION);
kvp->origin = crm_element_value_copy(op, PCMK__XA_ST_ORIGIN);
kvp->delegate = crm_element_value_copy(op, PCMK__XA_ST_DELEGATE);
kvp->client = crm_element_value_copy(op, PCMK__XA_ST_CLIENTNAME);
crm_element_value_ll(op, PCMK__XA_ST_DATE, &completed);
kvp->completed = (time_t) completed;
crm_element_value_ll(op, PCMK__XA_ST_DATE_NSEC, &completed_nsec);
kvp->completed_nsec = completed_nsec;
crm_element_value_int(op, PCMK__XA_ST_STATE, &kvp->state);
kvp->exit_reason = crm_element_value_copy(op, PCMK_XA_EXIT_REASON);
if (last) {
last->next = kvp;
} else {
*history = kvp;
}
last = kvp;
}
}
pcmk__xml_free(output);
return rc;
}
/*!
* \internal
* \brief Free a list of fencing history objects and all members of each object
*
* \param[in,out] head Head of fencing history object list
*/
void
stonith__history_free(stonith_history_t *head)
{
/* @COMPAT Drop "next" member of stonith_history_t, use a GList or GSList,
* and use the appropriate free function (while ensuring the members get
* freed)
*/
while (head != NULL) {
stonith_history_t *next = head->next;
free(head->target);
free(head->action);
free(head->origin);
free(head->delegate);
free(head->client);
free(head->exit_reason);
free(head);
head = next;
}
}
static gint
stonithlib_GCompareFunc(gconstpointer a, gconstpointer b)
{
int rc = 0;
const stonith_notify_client_t *a_client = a;
const stonith_notify_client_t *b_client = b;
if (a_client->delete || b_client->delete) {
/* make entries marked for deletion not findable */
return -1;
}
CRM_CHECK(a_client->event != NULL && b_client->event != NULL, return 0);
rc = strcmp(a_client->event, b_client->event);
if (rc == 0) {
if (a_client->notify == NULL || b_client->notify == NULL) {
return 0;
} else if (a_client->notify == b_client->notify) {
return 0;
} else if (((long)a_client->notify) < ((long)b_client->notify)) {
crm_err("callbacks for %s are not equal: %p vs. %p",
a_client->event, a_client->notify, b_client->notify);
return -1;
}
crm_err("callbacks for %s are not equal: %p vs. %p",
a_client->event, a_client->notify, b_client->notify);
return 1;
}
return rc;
}
xmlNode *
stonith_create_op(int call_id, const char *token, const char *op, xmlNode * data, int call_options)
{
xmlNode *op_msg = NULL;
CRM_CHECK(token != NULL, return NULL);
op_msg = pcmk__xe_create(NULL, PCMK__XE_STONITH_COMMAND);
crm_xml_add(op_msg, PCMK__XA_T, PCMK__VALUE_STONITH_NG);
crm_xml_add(op_msg, PCMK__XA_ST_OP, op);
crm_xml_add_int(op_msg, PCMK__XA_ST_CALLID, call_id);
crm_trace("Sending call options: %.8lx, %d", (long)call_options, call_options);
crm_xml_add_int(op_msg, PCMK__XA_ST_CALLOPT, call_options);
if (data != NULL) {
xmlNode *wrapper = pcmk__xe_create(op_msg, PCMK__XE_ST_CALLDATA);
pcmk__xml_copy(wrapper, data);
}
return op_msg;
}
static void
stonith_destroy_op_callback(gpointer data)
{
stonith_callback_client_t *blob = data;
if (blob->timer && blob->timer->ref > 0) {
g_source_remove(blob->timer->ref);
}
free(blob->timer);
free(blob);
}
static int
stonith_api_signoff(stonith_t * stonith)
{
stonith_private_t *native = stonith->st_private;
crm_debug("Disconnecting from the fencer");
if (native->source != NULL) {
/* Attached to mainloop */
mainloop_del_ipc_client(native->source);
native->source = NULL;
native->ipc = NULL;
} else if (native->ipc) {
/* Not attached to mainloop */
crm_ipc_t *ipc = native->ipc;
native->ipc = NULL;
crm_ipc_close(ipc);
crm_ipc_destroy(ipc);
}
free(native->token); native->token = NULL;
stonith->state = stonith_disconnected;
return pcmk_ok;
}
static int
stonith_api_del_callback(stonith_t * stonith, int call_id, bool all_callbacks)
{
stonith_private_t *private = stonith->st_private;
if (all_callbacks) {
private->op_callback = NULL;
g_hash_table_destroy(private->stonith_op_callback_table);
private->stonith_op_callback_table = pcmk__intkey_table(stonith_destroy_op_callback);
} else if (call_id == 0) {
private->op_callback = NULL;
} else {
pcmk__intkey_table_remove(private->stonith_op_callback_table, call_id);
}
return pcmk_ok;
}
/*!
* \internal
* \brief Invoke a (single) specified fence action callback
*
* \param[in,out] st Fencer API connection
* \param[in] call_id If positive, call ID of completed fence action,
* otherwise legacy return code for early failure
* \param[in,out] result Full result for action
* \param[in,out] userdata User data to pass to callback
* \param[in] callback Fence action callback to invoke
*/
static void
invoke_fence_action_callback(stonith_t *st, int call_id,
pcmk__action_result_t *result,
void *userdata,
void (*callback) (stonith_t *st,
stonith_callback_data_t *data))
{
stonith_callback_data_t data = { 0, };
data.call_id = call_id;
data.rc = pcmk_rc2legacy(stonith__result2rc(result));
data.userdata = userdata;
data.opaque = (void *) result;
callback(st, &data);
}
/*!
* \internal
* \brief Invoke any callbacks registered for a specified fence action result
*
* Given a fence action result from the fencer, invoke any callback registered
* for that action, as well as any global callback registered.
*
* \param[in,out] stonith Fencer API connection
* \param[in] msg If non-NULL, fencer reply
* \param[in] call_id If \p msg is NULL, call ID of action that timed out
*/
static void
invoke_registered_callbacks(stonith_t *stonith, const xmlNode *msg, int call_id)
{
stonith_private_t *private = NULL;
stonith_callback_client_t *cb_info = NULL;
pcmk__action_result_t result = PCMK__UNKNOWN_RESULT;
CRM_CHECK(stonith != NULL, return);
CRM_CHECK(stonith->st_private != NULL, return);
private = stonith->st_private;
if (msg == NULL) {
// Fencer didn't reply in time
pcmk__set_result(&result, CRM_EX_ERROR, PCMK_EXEC_TIMEOUT,
"Fencer accepted request but did not reply in time");
CRM_LOG_ASSERT(call_id > 0);
} else {
// We have the fencer reply
if ((crm_element_value_int(msg, PCMK__XA_ST_CALLID, &call_id) != 0)
|| (call_id <= 0)) {
crm_log_xml_warn(msg, "Bad fencer reply");
}
stonith__xe_get_result(msg, &result);
}
if (call_id > 0) {
cb_info = pcmk__intkey_table_lookup(private->stonith_op_callback_table,
call_id);
}
if ((cb_info != NULL) && (cb_info->callback != NULL)
&& (pcmk__result_ok(&result) || !(cb_info->only_success))) {
crm_trace("Invoking callback %s for call %d",
pcmk__s(cb_info->id, "without ID"), call_id);
invoke_fence_action_callback(stonith, call_id, &result,
cb_info->user_data, cb_info->callback);
} else if ((private->op_callback == NULL) && !pcmk__result_ok(&result)) {
crm_warn("Fencing action without registered callback failed: %d (%s%s%s)",
result.exit_status,
pcmk_exec_status_str(result.execution_status),
((result.exit_reason == NULL)? "" : ": "),
((result.exit_reason == NULL)? "" : result.exit_reason));
crm_log_xml_debug(msg, "Failed fence update");
}
if (private->op_callback != NULL) {
crm_trace("Invoking global callback for call %d", call_id);
invoke_fence_action_callback(stonith, call_id, &result, NULL,
private->op_callback);
}
if (cb_info != NULL) {
stonith_api_del_callback(stonith, call_id, FALSE);
}
pcmk__reset_result(&result);
}
static gboolean
stonith_async_timeout_handler(gpointer data)
{
struct timer_rec_s *timer = data;
crm_err("Async call %d timed out after %dms", timer->call_id, timer->timeout);
invoke_registered_callbacks(timer->stonith, NULL, timer->call_id);
/* Always return TRUE, never remove the handler
* We do that in stonith_del_callback()
*/
return TRUE;
}
static void
set_callback_timeout(stonith_callback_client_t * callback, stonith_t * stonith, int call_id,
int timeout)
{
struct timer_rec_s *async_timer = callback->timer;
if (timeout <= 0) {
return;
}
if (!async_timer) {
async_timer = pcmk__assert_alloc(1, sizeof(struct timer_rec_s));
callback->timer = async_timer;
}
async_timer->stonith = stonith;
async_timer->call_id = call_id;
/* Allow a fair bit of grace to allow the server to tell us of a timeout
* This is only a fallback
*/
async_timer->timeout = (timeout + 60) * 1000;
if (async_timer->ref) {
g_source_remove(async_timer->ref);
}
async_timer->ref =
pcmk__create_timer(async_timer->timeout, stonith_async_timeout_handler,
async_timer);
}
static void
update_callback_timeout(int call_id, int timeout, stonith_t * st)
{
stonith_callback_client_t *callback = NULL;
stonith_private_t *private = st->st_private;
callback = pcmk__intkey_table_lookup(private->stonith_op_callback_table,
call_id);
if (!callback || !callback->allow_timeout_updates) {
return;
}
set_callback_timeout(callback, st, call_id, timeout);
}
static int
stonith_dispatch_internal(const char *buffer, ssize_t length, gpointer userdata)
{
const char *type = NULL;
struct notify_blob_s blob;
stonith_t *st = userdata;
stonith_private_t *private = NULL;
pcmk__assert(st != NULL);
private = st->st_private;
blob.stonith = st;
blob.xml = pcmk__xml_parse(buffer);
if (blob.xml == NULL) {
crm_warn("Received malformed message from fencer: %s", buffer);
return 0;
}
/* do callbacks */
type = crm_element_value(blob.xml, PCMK__XA_T);
crm_trace("Activating %s callbacks...", type);
if (pcmk__str_eq(type, PCMK__VALUE_STONITH_NG, pcmk__str_none)) {
invoke_registered_callbacks(st, blob.xml, 0);
} else if (pcmk__str_eq(type, PCMK__VALUE_ST_NOTIFY, pcmk__str_none)) {
foreach_notify_entry(private, stonith_send_notification, &blob);
} else if (pcmk__str_eq(type, PCMK__VALUE_ST_ASYNC_TIMEOUT_VALUE,
pcmk__str_none)) {
int call_id = 0;
int timeout = 0;
crm_element_value_int(blob.xml, PCMK__XA_ST_TIMEOUT, &timeout);
crm_element_value_int(blob.xml, PCMK__XA_ST_CALLID, &call_id);
update_callback_timeout(call_id, timeout, st);
} else {
crm_err("Unknown message type: %s", type);
crm_log_xml_warn(blob.xml, "BadReply");
}
pcmk__xml_free(blob.xml);
return 1;
}
static int
stonith_api_signon(stonith_t * stonith, const char *name, int *stonith_fd)
{
int rc = pcmk_ok;
stonith_private_t *native = NULL;
const char *display_name = name? name : "client";
struct ipc_client_callbacks st_callbacks = {
.dispatch = stonith_dispatch_internal,
.destroy = stonith_connection_destroy
};
CRM_CHECK(stonith != NULL, return -EINVAL);
native = stonith->st_private;
pcmk__assert(native != NULL);
crm_debug("Attempting fencer connection by %s with%s mainloop",
display_name, (stonith_fd? "out" : ""));
stonith->state = stonith_connected_command;
if (stonith_fd) {
/* No mainloop */
native->ipc = crm_ipc_new("stonith-ng", 0);
if (native->ipc != NULL) {
rc = pcmk__connect_generic_ipc(native->ipc);
if (rc == pcmk_rc_ok) {
rc = pcmk__ipc_fd(native->ipc, stonith_fd);
if (rc != pcmk_rc_ok) {
crm_debug("Couldn't get file descriptor for IPC: %s",
pcmk_rc_str(rc));
}
}
if (rc != pcmk_rc_ok) {
crm_ipc_close(native->ipc);
crm_ipc_destroy(native->ipc);
native->ipc = NULL;
}
}
} else {
/* With mainloop */
native->source =
mainloop_add_ipc_client("stonith-ng", G_PRIORITY_MEDIUM, 0, stonith, &st_callbacks);
native->ipc = mainloop_get_ipc_client(native->source);
}
if (native->ipc == NULL) {
rc = -ENOTCONN;
} else {
xmlNode *reply = NULL;
xmlNode *hello = pcmk__xe_create(NULL, PCMK__XE_STONITH_COMMAND);
crm_xml_add(hello, PCMK__XA_T, PCMK__VALUE_STONITH_NG);
crm_xml_add(hello, PCMK__XA_ST_OP, CRM_OP_REGISTER);
crm_xml_add(hello, PCMK__XA_ST_CLIENTNAME, name);
rc = crm_ipc_send(native->ipc, hello, crm_ipc_client_response, -1, &reply);
if (rc < 0) {
crm_debug("Couldn't register with the fencer: %s "
QB_XS " rc=%d", pcmk_strerror(rc), rc);
rc = -ECOMM;
} else if (reply == NULL) {
crm_debug("Couldn't register with the fencer: no reply");
rc = -EPROTO;
} else {
const char *msg_type = crm_element_value(reply, PCMK__XA_ST_OP);
native->token = crm_element_value_copy(reply, PCMK__XA_ST_CLIENTID);
if (!pcmk__str_eq(msg_type, CRM_OP_REGISTER, pcmk__str_none)) {
crm_debug("Couldn't register with the fencer: invalid reply type '%s'",
(msg_type? msg_type : "(missing)"));
crm_log_xml_debug(reply, "Invalid fencer reply");
rc = -EPROTO;
} else if (native->token == NULL) {
crm_debug("Couldn't register with the fencer: no token in reply");
crm_log_xml_debug(reply, "Invalid fencer reply");
rc = -EPROTO;
} else {
crm_debug("Connection to fencer by %s succeeded (registration token: %s)",
display_name, native->token);
rc = pcmk_ok;
}
}
pcmk__xml_free(reply);
pcmk__xml_free(hello);
}
if (rc != pcmk_ok) {
crm_debug("Connection attempt to fencer by %s failed: %s "
QB_XS " rc=%d", display_name, pcmk_strerror(rc), rc);
stonith->cmds->disconnect(stonith);
}
return rc;
}
static int
stonith_set_notification(stonith_t * stonith, const char *callback, int enabled)
{
int rc = pcmk_ok;
xmlNode *notify_msg = pcmk__xe_create(NULL, __func__);
stonith_private_t *native = stonith->st_private;
if (stonith->state != stonith_disconnected) {
crm_xml_add(notify_msg, PCMK__XA_ST_OP, STONITH_OP_NOTIFY);
if (enabled) {
crm_xml_add(notify_msg, PCMK__XA_ST_NOTIFY_ACTIVATE, callback);
} else {
crm_xml_add(notify_msg, PCMK__XA_ST_NOTIFY_DEACTIVATE, callback);
}
rc = crm_ipc_send(native->ipc, notify_msg, crm_ipc_client_response, -1, NULL);
if (rc < 0) {
crm_perror(LOG_DEBUG, "Couldn't register for fencing notifications: %d", rc);
rc = -ECOMM;
} else {
rc = pcmk_ok;
}
}
pcmk__xml_free(notify_msg);
return rc;
}
static int
stonith_api_add_notification(stonith_t * stonith, const char *event,
void (*callback) (stonith_t * stonith, stonith_event_t * e))
{
GList *list_item = NULL;
stonith_notify_client_t *new_client = NULL;
stonith_private_t *private = NULL;
private = stonith->st_private;
crm_trace("Adding callback for %s events (%d)", event, g_list_length(private->notify_list));
new_client = pcmk__assert_alloc(1, sizeof(stonith_notify_client_t));
new_client->event = event;
new_client->notify = callback;
list_item = g_list_find_custom(private->notify_list, new_client, stonithlib_GCompareFunc);
if (list_item != NULL) {
crm_warn("Callback already present");
free(new_client);
return -ENOTUNIQ;
} else {
private->notify_list = g_list_append(private->notify_list, new_client);
stonith_set_notification(stonith, event, 1);
crm_trace("Callback added (%d)", g_list_length(private->notify_list));
}
return pcmk_ok;
}
static void
del_notify_entry(gpointer data, gpointer user_data)
{
stonith_notify_client_t *entry = data;
stonith_t * stonith = user_data;
if (!entry->delete) {
crm_debug("Removing callback for %s events", entry->event);
stonith_api_del_notification(stonith, entry->event);
}
}
static int
stonith_api_del_notification(stonith_t * stonith, const char *event)
{
GList *list_item = NULL;
stonith_notify_client_t *new_client = NULL;
stonith_private_t *private = stonith->st_private;
if (event == NULL) {
foreach_notify_entry(private, del_notify_entry, stonith);
crm_trace("Removed callback");
return pcmk_ok;
}
crm_debug("Removing callback for %s events", event);
new_client = pcmk__assert_alloc(1, sizeof(stonith_notify_client_t));
new_client->event = event;
new_client->notify = NULL;
list_item = g_list_find_custom(private->notify_list, new_client, stonithlib_GCompareFunc);
stonith_set_notification(stonith, event, 0);
if (list_item != NULL) {
stonith_notify_client_t *list_client = list_item->data;
if (private->notify_refcnt) {
list_client->delete = TRUE;
private->notify_deletes = TRUE;
} else {
private->notify_list = g_list_remove(private->notify_list, list_client);
free(list_client);
}
crm_trace("Removed callback");
} else {
crm_trace("Callback not present");
}
free(new_client);
return pcmk_ok;
}
static int
stonith_api_add_callback(stonith_t * stonith, int call_id, int timeout, int options,
void *user_data, const char *callback_name,
void (*callback) (stonith_t * st, stonith_callback_data_t * data))
{
stonith_callback_client_t *blob = NULL;
stonith_private_t *private = NULL;
CRM_CHECK(stonith != NULL, return -EINVAL);
CRM_CHECK(stonith->st_private != NULL, return -EINVAL);
private = stonith->st_private;
if (call_id == 0) { // Add global callback
private->op_callback = callback;
} else if (call_id < 0) { // Call failed immediately, so call callback now
if (!(options & st_opt_report_only_success)) {
pcmk__action_result_t result = PCMK__UNKNOWN_RESULT;
crm_trace("Call failed, calling %s: %s", callback_name, pcmk_strerror(call_id));
pcmk__set_result(&result, CRM_EX_ERROR,
stonith__legacy2status(call_id), NULL);
invoke_fence_action_callback(stonith, call_id, &result,
user_data, callback);
} else {
crm_warn("Fencer call failed: %s", pcmk_strerror(call_id));
}
return FALSE;
}
blob = pcmk__assert_alloc(1, sizeof(stonith_callback_client_t));
blob->id = callback_name;
blob->only_success = (options & st_opt_report_only_success) ? TRUE : FALSE;
blob->user_data = user_data;
blob->callback = callback;
blob->allow_timeout_updates = (options & st_opt_timeout_updates) ? TRUE : FALSE;
if (timeout > 0) {
set_callback_timeout(blob, stonith, call_id, timeout);
}
pcmk__intkey_table_insert(private->stonith_op_callback_table, call_id,
blob);
crm_trace("Added callback to %s for call %d", callback_name, call_id);
return TRUE;
}
/*!
* \internal
* \brief Get the data section of a fencer notification
*
* \param[in] msg Notification XML
* \param[in] ntype Notification type
*/
static xmlNode *
get_event_data_xml(xmlNode *msg, const char *ntype)
{
char *data_addr = crm_strdup_printf("//%s", ntype);
xmlNode *data = pcmk__xpath_find_one(msg->doc, data_addr, LOG_DEBUG);
free(data_addr);
return data;
}
/*
*/
static stonith_event_t *
xml_to_event(xmlNode *msg)
{
stonith_event_t *event = pcmk__assert_alloc(1, sizeof(stonith_event_t));
struct event_private *event_private = NULL;
event->opaque = pcmk__assert_alloc(1, sizeof(struct event_private));
event_private = (struct event_private *) event->opaque;
crm_log_xml_trace(msg, "stonith_notify");
// All notification types have the operation result and notification subtype
stonith__xe_get_result(msg, &event_private->result);
event->operation = crm_element_value_copy(msg, PCMK__XA_ST_OP);
// @COMPAT The API originally provided the result as a legacy return code
event->result = pcmk_rc2legacy(stonith__result2rc(&event_private->result));
// Some notification subtypes have additional information
if (pcmk__str_eq(event->operation, PCMK__VALUE_ST_NOTIFY_FENCE,
pcmk__str_none)) {
xmlNode *data = get_event_data_xml(msg, event->operation);
if (data == NULL) {
crm_err("No data for %s event", event->operation);
crm_log_xml_notice(msg, "BadEvent");
} else {
event->origin = crm_element_value_copy(data, PCMK__XA_ST_ORIGIN);
event->action = crm_element_value_copy(data,
PCMK__XA_ST_DEVICE_ACTION);
event->target = crm_element_value_copy(data, PCMK__XA_ST_TARGET);
event->executioner = crm_element_value_copy(data,
PCMK__XA_ST_DELEGATE);
event->id = crm_element_value_copy(data, PCMK__XA_ST_REMOTE_OP);
event->client_origin =
crm_element_value_copy(data, PCMK__XA_ST_CLIENTNAME);
event->device = crm_element_value_copy(data, PCMK__XA_ST_DEVICE_ID);
}
} else if (pcmk__str_any_of(event->operation,
STONITH_OP_DEVICE_ADD, STONITH_OP_DEVICE_DEL,
STONITH_OP_LEVEL_ADD, STONITH_OP_LEVEL_DEL,
NULL)) {
xmlNode *data = get_event_data_xml(msg, event->operation);
if (data == NULL) {
crm_err("No data for %s event", event->operation);
crm_log_xml_notice(msg, "BadEvent");
} else {
event->device = crm_element_value_copy(data, PCMK__XA_ST_DEVICE_ID);
}
}
return event;
}
static void
event_free(stonith_event_t * event)
{
struct event_private *event_private = event->opaque;
free(event->id);
free(event->operation);
free(event->origin);
free(event->action);
free(event->target);
free(event->executioner);
free(event->device);
free(event->client_origin);
pcmk__reset_result(&event_private->result);
free(event->opaque);
free(event);
}
static void
stonith_send_notification(gpointer data, gpointer user_data)
{
struct notify_blob_s *blob = user_data;
stonith_notify_client_t *entry = data;
stonith_event_t *st_event = NULL;
const char *event = NULL;
if (blob->xml == NULL) {
crm_warn("Skipping callback - NULL message");
return;
}
event = crm_element_value(blob->xml, PCMK__XA_SUBT);
if (entry == NULL) {
crm_warn("Skipping callback - NULL callback client");
return;
} else if (entry->delete) {
crm_trace("Skipping callback - marked for deletion");
return;
} else if (entry->notify == NULL) {
crm_warn("Skipping callback - NULL callback");
return;
} else if (!pcmk__str_eq(entry->event, event, pcmk__str_none)) {
crm_trace("Skipping callback - event mismatch %p/%s vs. %s", entry, entry->event, event);
return;
}
st_event = xml_to_event(blob->xml);
crm_trace("Invoking callback for %p/%s event...", entry, event);
entry->notify(blob->stonith, st_event);
crm_trace("Callback invoked...");
event_free(st_event);
}
/*!
* \internal
* \brief Create and send an API request
*
* \param[in,out] stonith Stonith connection
* \param[in] op API operation to request
* \param[in] data Data to attach to request
* \param[out] output_data If not NULL, will be set to reply if synchronous
* \param[in] call_options Bitmask of stonith_call_options to use
* \param[in] timeout Error if not completed within this many seconds
*
* \return pcmk_ok (for synchronous requests) or positive call ID
* (for asynchronous requests) on success, -errno otherwise
*/
static int
stonith_send_command(stonith_t * stonith, const char *op, xmlNode * data, xmlNode ** output_data,
int call_options, int timeout)
{
int rc = 0;
int reply_id = -1;
xmlNode *op_msg = NULL;
xmlNode *op_reply = NULL;
stonith_private_t *native = NULL;
pcmk__assert((stonith != NULL) && (stonith->st_private != NULL)
&& (op != NULL));
native = stonith->st_private;
if (output_data != NULL) {
*output_data = NULL;
}
if ((stonith->state == stonith_disconnected) || (native->token == NULL)) {
return -ENOTCONN;
}
/* Increment the call ID, which must be positive to avoid conflicting with
* error codes. This shouldn't be a problem unless the client mucked with
* it or the counter wrapped around.
*/
stonith->call_id++;
if (stonith->call_id < 1) {
stonith->call_id = 1;
}
op_msg = stonith_create_op(stonith->call_id, native->token, op, data, call_options);
if (op_msg == NULL) {
return -EINVAL;
}
crm_xml_add_int(op_msg, PCMK__XA_ST_TIMEOUT, timeout);
crm_trace("Sending %s message to fencer with timeout %ds", op, timeout);
if (data) {
const char *delay_s = crm_element_value(data, PCMK__XA_ST_DELAY);
if (delay_s) {
crm_xml_add(op_msg, PCMK__XA_ST_DELAY, delay_s);
}
}
{
enum crm_ipc_flags ipc_flags = crm_ipc_flags_none;
if (call_options & st_opt_sync_call) {
pcmk__set_ipc_flags(ipc_flags, "stonith command",
crm_ipc_client_response);
}
rc = crm_ipc_send(native->ipc, op_msg, ipc_flags,
1000 * (timeout + 60), &op_reply);
}
pcmk__xml_free(op_msg);
if (rc < 0) {
crm_perror(LOG_ERR, "Couldn't perform %s operation (timeout=%ds): %d", op, timeout, rc);
rc = -ECOMM;
goto done;
}
crm_log_xml_trace(op_reply, "Reply");
if (!(call_options & st_opt_sync_call)) {
crm_trace("Async call %d, returning", stonith->call_id);
pcmk__xml_free(op_reply);
return stonith->call_id;
}
crm_element_value_int(op_reply, PCMK__XA_ST_CALLID, &reply_id);
if (reply_id == stonith->call_id) {
pcmk__action_result_t result = PCMK__UNKNOWN_RESULT;
crm_trace("Synchronous reply %d received", reply_id);
stonith__xe_get_result(op_reply, &result);
rc = pcmk_rc2legacy(stonith__result2rc(&result));
pcmk__reset_result(&result);
if ((call_options & st_opt_discard_reply) || output_data == NULL) {
crm_trace("Discarding reply");
} else {
*output_data = op_reply;
op_reply = NULL; /* Prevent subsequent free */
}
} else if (reply_id <= 0) {
crm_err("Received bad reply: No id set");
crm_log_xml_err(op_reply, "Bad reply");
pcmk__xml_free(op_reply);
op_reply = NULL;
rc = -ENOMSG;
} else {
crm_err("Received bad reply: %d (wanted %d)", reply_id, stonith->call_id);
crm_log_xml_err(op_reply, "Old reply");
pcmk__xml_free(op_reply);
op_reply = NULL;
rc = -ENOMSG;
}
done:
if (!crm_ipc_connected(native->ipc)) {
crm_err("Fencer disconnected");
free(native->token); native->token = NULL;
stonith->state = stonith_disconnected;
}
pcmk__xml_free(op_reply);
return rc;
}
/*!
* \internal
* \brief Process IPC messages for a fencer API connection
*
* This is used for testing purposes in scenarios that don't use a mainloop to
* dispatch messages automatically.
*
* \param[in,out] stonith_api Fencer API connetion object
*
* \return Standard Pacemaker return code
*/
int
stonith__api_dispatch(stonith_t *stonith_api)
{
stonith_private_t *private = NULL;
pcmk__assert(stonith_api != NULL);
private = stonith_api->st_private;
while (crm_ipc_ready(private->ipc)) {
if (crm_ipc_read(private->ipc) > 0) {
const char *msg = crm_ipc_buffer(private->ipc);
stonith_dispatch_internal(msg, strlen(msg), stonith_api);
+ pcmk__ipc_free_client_buffer(private->ipc);
}
if (!crm_ipc_connected(private->ipc)) {
crm_err("Connection closed");
return ENOTCONN;
}
}
return pcmk_rc_ok;
}
static int
free_stonith_api(stonith_t *stonith)
{
int rc = pcmk_ok;
crm_trace("Destroying %p", stonith);
if (stonith->state != stonith_disconnected) {
crm_trace("Unregistering notifications and disconnecting %p first",
stonith);
stonith->cmds->remove_notification(stonith, NULL);
rc = stonith->cmds->disconnect(stonith);
}
if (stonith->state == stonith_disconnected) {
stonith_private_t *private = stonith->st_private;
crm_trace("Removing %d callbacks", g_hash_table_size(private->stonith_op_callback_table));
g_hash_table_destroy(private->stonith_op_callback_table);
crm_trace("Destroying %d notification clients", g_list_length(private->notify_list));
g_list_free_full(private->notify_list, free);
free(stonith->st_private);
free(stonith->cmds);
free(stonith);
} else {
crm_err("Not free'ing active connection: %s (%d)", pcmk_strerror(rc), rc);
}
return rc;
}
static gboolean
is_stonith_param(gpointer key, gpointer value, gpointer user_data)
{
return pcmk_stonith_param(key);
}
int
stonith__validate(stonith_t *st, int call_options, const char *rsc_id,
const char *agent, GHashTable *params, int timeout_sec,
char **output, char **error_output)
{
int rc = pcmk_rc_ok;
/* Use a dummy node name in case the agent requires a target. We assume the
* actual target doesn't matter for validation purposes (if in practice,
* that is incorrect, we will need to allow the caller to pass the target).
*/
const char *target = "node1";
char *host_arg = NULL;
if (params != NULL) {
host_arg = pcmk__str_copy(g_hash_table_lookup(params, PCMK_STONITH_HOST_ARGUMENT));
/* Remove special stonith params from the table before doing anything else */
g_hash_table_foreach_remove(params, is_stonith_param, NULL);
}
#if PCMK__ENABLE_CIBSECRETS
rc = pcmk__substitute_secrets(rsc_id, params);
if (rc != pcmk_rc_ok) {
crm_warn("Could not replace secret parameters for validation of %s: %s",
agent, pcmk_rc_str(rc));
// rc is standard return value, don't return it in this function
}
#endif
if (output) {
*output = NULL;
}
if (error_output) {
*error_output = NULL;
}
if (timeout_sec <= 0) {
timeout_sec = PCMK_DEFAULT_ACTION_TIMEOUT_MS;
}
switch (get_namespace_from_agent(agent)) {
case st_namespace_rhcs:
rc = stonith__rhcs_validate(st, call_options, target, agent,
params, host_arg, timeout_sec,
output, error_output);
rc = pcmk_legacy2rc(rc);
break;
#if HAVE_STONITH_STONITH_H
case st_namespace_lha:
rc = stonith__lha_validate(st, call_options, target, agent,
params, timeout_sec, output,
error_output);
rc = pcmk_legacy2rc(rc);
break;
#endif
case st_namespace_invalid:
errno = ENOENT;
rc = errno;
if (error_output) {
*error_output = crm_strdup_printf("Agent %s not found", agent);
} else {
crm_err("Agent %s not found", agent);
}
break;
default:
errno = EOPNOTSUPP;
rc = errno;
if (error_output) {
*error_output = crm_strdup_printf("Agent %s does not support validation",
agent);
} else {
crm_err("Agent %s does not support validation", agent);
}
break;
}
free(host_arg);
return rc;
}
static int
stonith_api_validate(stonith_t *st, int call_options, const char *rsc_id,
const char *namespace_s, const char *agent,
const stonith_key_value_t *params, int timeout_sec,
char **output, char **error_output)
{
/* Validation should be done directly via the agent, so we can get it from
* stonith_admin when the cluster is not running, which is important for
* higher-level tools.
*/
int rc = pcmk_ok;
GHashTable *params_table = pcmk__strkey_table(free, free);
// Convert parameter list to a hash table
for (; params; params = params->next) {
if (!pcmk_stonith_param(params->key)) {
pcmk__insert_dup(params_table, params->key, params->value);
}
}
rc = stonith__validate(st, call_options, rsc_id, agent, params_table,
timeout_sec, output, error_output);
g_hash_table_destroy(params_table);
return rc;
}
/*!
* \internal
* \brief Create a new fencer API connection object
*
* \return Newly allocated fencer API connection object, or \c NULL on
* allocation failure
*/
stonith_t *
stonith__api_new(void)
{
stonith_t *new_stonith = NULL;
stonith_private_t *private = NULL;
new_stonith = calloc(1, sizeof(stonith_t));
if (new_stonith == NULL) {
return NULL;
}
private = calloc(1, sizeof(stonith_private_t));
if (private == NULL) {
free(new_stonith);
return NULL;
}
new_stonith->st_private = private;
private->stonith_op_callback_table = pcmk__intkey_table(stonith_destroy_op_callback);
private->notify_list = NULL;
private->notify_refcnt = 0;
private->notify_deletes = FALSE;
new_stonith->call_id = 1;
new_stonith->state = stonith_disconnected;
new_stonith->cmds = calloc(1, sizeof(stonith_api_operations_t));
if (new_stonith->cmds == NULL) {
free(new_stonith->st_private);
free(new_stonith);
return NULL;
}
new_stonith->cmds->free = free_stonith_api;
new_stonith->cmds->connect = stonith_api_signon;
new_stonith->cmds->disconnect = stonith_api_signoff;
new_stonith->cmds->list = stonith_api_list;
new_stonith->cmds->monitor = stonith_api_monitor;
new_stonith->cmds->status = stonith_api_status;
new_stonith->cmds->fence = stonith_api_fence;
new_stonith->cmds->fence_with_delay = stonith_api_fence_with_delay;
new_stonith->cmds->confirm = stonith_api_confirm;
new_stonith->cmds->history = stonith_api_history;
new_stonith->cmds->list_agents = stonith_api_device_list;
new_stonith->cmds->metadata = stonith_api_device_metadata;
new_stonith->cmds->query = stonith_api_query;
new_stonith->cmds->remove_device = stonith_api_remove_device;
new_stonith->cmds->register_device = stonith_api_register_device;
new_stonith->cmds->remove_level = stonith_api_remove_level;
new_stonith->cmds->remove_level_full = stonith_api_remove_level_full;
new_stonith->cmds->register_level = stonith_api_register_level;
new_stonith->cmds->register_level_full = stonith_api_register_level_full;
new_stonith->cmds->remove_callback = stonith_api_del_callback;
new_stonith->cmds->register_callback = stonith_api_add_callback;
new_stonith->cmds->remove_notification = stonith_api_del_notification;
new_stonith->cmds->register_notification = stonith_api_add_notification;
new_stonith->cmds->validate = stonith_api_validate;
return new_stonith;
}
/*!
* \internal
* \brief Free a fencer API connection object
*
* \param[in,out] stonith_api Fencer API connection object
*/
void
stonith__api_free(stonith_t *stonith_api)
{
crm_trace("Destroying %p", stonith_api);
if (stonith_api != NULL) {
stonith_api->cmds->free(stonith_api);
}
}
/*!
* \internal
* \brief Connect to the fencer, retrying on failure
*
* \param[in,out] stonith Fencer API connection object
* \param[in] name Client name to use with fencer
* \param[in] max_attempts Maximum number of attempts
*
* \return \c pcmk_rc_ok on success, or result of last attempt otherwise
*/
int
stonith__api_connect_retry(stonith_t *stonith_api, const char *name,
int max_attempts)
{
int rc = EINVAL; // if max_attempts is not positive
for (int attempt = 1; attempt <= max_attempts; attempt++) {
rc = stonith_api->cmds->connect(stonith_api, name, NULL);
rc = pcmk_legacy2rc(rc);
if (rc == pcmk_rc_ok) {
return rc;
}
if (attempt < max_attempts) {
crm_notice("Fencer connection attempt %d of %d failed "
"(retrying in 2s): %s " QB_XS " rc=%d",
attempt, max_attempts, pcmk_rc_str(rc), rc);
sleep(2);
}
}
crm_notice("Could not connect to fencer: %s " QB_XS " rc=%d",
pcmk_rc_str(rc), rc);
return rc;
}
/*!
* \internal
* \brief Append a newly allocated STONITH key-value pair to a list
*
* \param[in,out] head Head of key-value pair list (\c NULL for new list)
* \param[in] key Key to add
* \param[in] value Value to add
*
* \return Head of appended-to list (equal to \p head if \p head is not \c NULL)
* \note The caller is responsible for freeing the return value using
* \c stonith__key_value_freeall().
*/
stonith_key_value_t *
stonith__key_value_add(stonith_key_value_t *head, const char *key,
const char *value)
{
/* @COMPAT Replace this function with pcmk_prepend_nvpair(), and reverse the
* list when finished adding to it; or with a hash table where order does
* not matter
*/
stonith_key_value_t *pair = pcmk__assert_alloc(1,
sizeof(stonith_key_value_t));
pair->key = pcmk__str_copy(key);
pair->value = pcmk__str_copy(value);
if (head != NULL) {
stonith_key_value_t *end = head;
for (; end->next != NULL; end = end->next);
end->next = pair;
} else {
head = pair;
}
return head;
}
/*!
* \internal
* \brief Free all items in a \c stonith_key_value_t list
*
* This means freeing the list itself with all of its nodes. Keys and values may
* be freed depending on arguments.
*
* \param[in,out] head Head of list
* \param[in] keys If \c true, free all keys
* \param[in] values If \c true, free all values
*/
void
stonith__key_value_freeall(stonith_key_value_t *head, bool keys, bool values)
{
while (head != NULL) {
stonith_key_value_t *next = head->next;
if (keys) {
free(head->key);
}
if (values) {
free(head->value);
}
free(head);
head = next;
}
}
#define api_log_open() openlog("stonith-api", LOG_CONS | LOG_NDELAY | LOG_PID, LOG_DAEMON)
#define api_log(level, fmt, args...) syslog(level, "%s: "fmt, __func__, args)
int
stonith_api_kick(uint32_t nodeid, const char *uname, int timeout, bool off)
{
int rc = pcmk_ok;
stonith_t *st = stonith__api_new();
const char *action = off? PCMK_ACTION_OFF : PCMK_ACTION_REBOOT;
api_log_open();
if (st == NULL) {
api_log(LOG_ERR, "API initialization failed, could not kick (%s) node %u/%s",
action, nodeid, uname);
return -EPROTO;
}
rc = st->cmds->connect(st, "stonith-api", NULL);
if (rc != pcmk_ok) {
api_log(LOG_ERR, "Connection failed, could not kick (%s) node %u/%s : %s (%d)",
action, nodeid, uname, pcmk_strerror(rc), rc);
} else {
char *name = (uname == NULL)? pcmk__itoa(nodeid) : strdup(uname);
int opts = 0;
stonith__set_call_options(opts, name,
st_opt_sync_call|st_opt_allow_self_fencing);
if ((uname == NULL) && (nodeid > 0)) {
stonith__set_call_options(opts, name, st_opt_cs_nodeid);
}
rc = st->cmds->fence(st, opts, name, action, timeout, 0);
free(name);
if (rc != pcmk_ok) {
api_log(LOG_ERR, "Could not kick (%s) node %u/%s : %s (%d)",
action, nodeid, uname, pcmk_strerror(rc), rc);
} else {
api_log(LOG_NOTICE, "Node %u/%s kicked: %s", nodeid, uname, action);
}
}
stonith__api_free(st);
return rc;
}
time_t
stonith_api_time(uint32_t nodeid, const char *uname, bool in_progress)
{
int rc = pcmk_ok;
time_t when = 0;
stonith_t *st = stonith__api_new();
stonith_history_t *history = NULL, *hp = NULL;
if (st == NULL) {
api_log(LOG_ERR, "Could not retrieve fence history for %u/%s: "
"API initialization failed", nodeid, uname);
return when;
}
rc = st->cmds->connect(st, "stonith-api", NULL);
if (rc != pcmk_ok) {
api_log(LOG_NOTICE, "Connection failed: %s (%d)", pcmk_strerror(rc), rc);
} else {
int entries = 0;
int progress = 0;
int completed = 0;
int opts = 0;
char *name = (uname == NULL)? pcmk__itoa(nodeid) : strdup(uname);
stonith__set_call_options(opts, name, st_opt_sync_call);
if ((uname == NULL) && (nodeid > 0)) {
stonith__set_call_options(opts, name, st_opt_cs_nodeid);
}
rc = st->cmds->history(st, opts, name, &history, 120);
free(name);
for (hp = history; hp; hp = hp->next) {
entries++;
if (in_progress) {
progress++;
if (hp->state != st_done && hp->state != st_failed) {
when = time(NULL);
}
} else if (hp->state == st_done) {
completed++;
if (hp->completed > when) {
when = hp->completed;
}
}
}
stonith__history_free(history);
if(rc == pcmk_ok) {
api_log(LOG_INFO, "Found %d entries for %u/%s: %d in progress, %d completed", entries, nodeid, uname, progress, completed);
} else {
api_log(LOG_ERR, "Could not retrieve fence history for %u/%s: %s (%d)", nodeid, uname, pcmk_strerror(rc), rc);
}
}
stonith__api_free(st);
if(when) {
api_log(LOG_INFO, "Node %u/%s last kicked at: %ld", nodeid, uname, (long int)when);
}
return when;
}
/*!
* \internal
* \brief Check whether a fence agent with a given name exists
*
* \param[in] name Agent name
*
* \retval \c true If a fence agent named \p name exists
* \retval \c false Otherwise
*/
bool
stonith__agent_exists(const char *name)
{
stonith_t *stonith_api = NULL;
stonith_key_value_t *agents = NULL;
bool rc = false;
if (name == NULL) {
return false;
}
stonith_api = stonith__api_new();
if (stonith_api == NULL) {
crm_err("Could not list fence agents: API memory allocation failed");
return false;
}
// The list_agents method ignores its timeout argument
stonith_api->cmds->list_agents(stonith_api, st_opt_sync_call, NULL, &agents,
0);
for (const stonith_key_value_t *iter = agents; iter != NULL;
iter = iter->next) {
if (pcmk__str_eq(iter->value, name, pcmk__str_none)) {
rc = true;
break;
}
}
stonith__key_value_freeall(agents, true, true);
stonith__api_free(stonith_api);
return rc;
}
/*!
* \internal
* \brief Parse a target name from one line of a target list string
*
* \param[in] line One line of a target list string
* \param[in] len String length of line
* \param[in,out] output List to add newly allocated target name to
*/
static void
parse_list_line(const char *line, int len, GList **output)
{
size_t i = 0;
size_t entry_start = 0;
/* Skip complaints about additional parameters device doesn't understand
*
* @TODO Document or eliminate the implied restriction of target names
*/
if (strstr(line, "invalid") || strstr(line, "variable")) {
crm_debug("Skipping list output line: %s", line);
return;
}
// Process line content, character by character
for (i = 0; i <= len; i++) {
if (isspace(line[i]) || (line[i] == ',') || (line[i] == ';')
|| (line[i] == '\0')) {
// We've found a separator (i.e. the end of an entry)
int rc = 0;
char *entry = NULL;
if (i == entry_start) {
// Skip leading and sequential separators
entry_start = i + 1;
continue;
}
entry = pcmk__assert_alloc(i - entry_start + 1, sizeof(char));
/* Read entry, stopping at first separator
*
* @TODO Document or eliminate these character restrictions
*/
rc = sscanf(line + entry_start, "%[a-zA-Z0-9_-.]", entry);
if (rc != 1) {
crm_warn("Could not parse list output entry: %s "
QB_XS " entry_start=%d position=%d",
line + entry_start, entry_start, i);
free(entry);
} else if (pcmk__strcase_any_of(entry, PCMK_ACTION_ON,
PCMK_ACTION_OFF, NULL)) {
/* Some agents print the target status in the list output,
* though none are known now (the separate list-status command
* is used for this, but it can also print "UNKNOWN"). To handle
* this possibility, skip such entries.
*
* @TODO Document or eliminate the implied restriction of target
* names.
*/
free(entry);
} else {
// We have a valid entry
*output = g_list_append(*output, entry);
}
entry_start = i + 1;
}
}
}
/*!
* \internal
* \brief Parse a list of targets from a string
*
* \param[in] list_output Target list as a string
*
* \return List of target names
* \note The target list string format is flexible, to allow for user-specified
* lists such pcmk_host_list and the output of an agent's list action
* (whether direct or via the API, which escapes newlines). There may be
* multiple lines, separated by either a newline or an escaped newline
* (backslash n). Each line may have one or more target names, separated
* by any combination of whitespace, commas, and semi-colons. Lines
* containing "invalid" or "variable" will be ignored entirely. Target
* names "on" or "off" (case-insensitive) will be ignored. Target names
* may contain only alphanumeric characters, underbars (_), dashes (-),
* and dots (.) (if any other character occurs in the name, it and all
* subsequent characters in the name will be ignored).
* \note The caller is responsible for freeing the result with
* g_list_free_full(result, free).
*/
GList *
stonith__parse_targets(const char *target_spec)
{
GList *targets = NULL;
if (target_spec != NULL) {
size_t out_len = strlen(target_spec);
size_t line_start = 0; // Starting index of line being processed
for (size_t i = 0; i <= out_len; ++i) {
if ((target_spec[i] == '\n') || (target_spec[i] == '\0')
|| ((target_spec[i] == '\\') && (target_spec[i + 1] == 'n'))) {
// We've reached the end of one line of output
int len = i - line_start;
if (len > 0) {
char *line = strndup(target_spec + line_start, len);
line[len] = '\0'; // Because it might be a newline
parse_list_line(line, len, &targets);
free(line);
}
if (target_spec[i] == '\\') {
++i; // backslash-n takes up two positions
}
line_start = i + 1;
}
}
}
return targets;
}
/*!
* \internal
* \brief Check whether a fencing failure was followed by an equivalent success
*
* \param[in] event Fencing failure
* \param[in] top_history Complete fencing history (must be sorted by
* stonith__sort_history() beforehand)
*
* \return The name of the node that executed the fencing if a later successful
* event exists, or NULL if no such event exists
*/
const char *
stonith__later_succeeded(const stonith_history_t *event,
const stonith_history_t *top_history)
{
const char *other = NULL;
for (const stonith_history_t *prev_hp = top_history;
prev_hp != NULL; prev_hp = prev_hp->next) {
if (prev_hp == event) {
break;
}
if ((prev_hp->state == st_done) &&
pcmk__str_eq(event->target, prev_hp->target, pcmk__str_casei) &&
pcmk__str_eq(event->action, prev_hp->action, pcmk__str_none) &&
((event->completed < prev_hp->completed) ||
((event->completed == prev_hp->completed) && (event->completed_nsec < prev_hp->completed_nsec)))) {
if ((event->delegate == NULL)
|| pcmk__str_eq(event->delegate, prev_hp->delegate,
pcmk__str_casei)) {
// Prefer equivalent fencing by same executioner
return prev_hp->delegate;
} else if (other == NULL) {
// Otherwise remember first successful executioner
other = (prev_hp->delegate == NULL)? "some node" : prev_hp->delegate;
}
}
}
return other;
}
/*!
* \internal
* \brief Sort fencing history, pending first then by most recently completed
*
* \param[in,out] history List of stonith actions
*
* \return New head of sorted \p history
*/
stonith_history_t *
stonith__sort_history(stonith_history_t *history)
{
stonith_history_t *new = NULL, *pending = NULL, *hp, *np, *tmp;
for (hp = history; hp; ) {
tmp = hp->next;
if ((hp->state == st_done) || (hp->state == st_failed)) {
/* sort into new */
if ((!new) || (hp->completed > new->completed) ||
((hp->completed == new->completed) && (hp->completed_nsec > new->completed_nsec))) {
hp->next = new;
new = hp;
} else {
np = new;
do {
if ((!np->next) || (hp->completed > np->next->completed) ||
((hp->completed == np->next->completed) && (hp->completed_nsec > np->next->completed_nsec))) {
hp->next = np->next;
np->next = hp;
break;
}
np = np->next;
} while (1);
}
} else {
/* put into pending */
hp->next = pending;
pending = hp;
}
hp = tmp;
}
/* pending actions don't have a completed-stamp so make them go front */
if (pending) {
stonith_history_t *last_pending = pending;
while (last_pending->next) {
last_pending = last_pending->next;
}
last_pending->next = new;
new = pending;
}
return new;
}
/*!
* \internal
* \brief Return string equivalent of a fencing operation state value
*
* \param[in] state Fencing operation state value
*
* \return Human-friendly string equivalent of \p state
*/
const char *
stonith__op_state_text(enum op_state state)
{
// @COMPAT Move this to the fencer after dropping stonith_op_state_str()
switch (state) {
case st_query:
return "querying";
case st_exec:
return "executing";
case st_done:
return "completed";
case st_duplicate:
return "duplicate";
case st_failed:
return "failed";
default:
return "unknown";
}
}
stonith_history_t *
stonith__first_matching_event(stonith_history_t *history,
bool (*matching_fn)(stonith_history_t *, void *),
void *user_data)
{
for (stonith_history_t *hp = history; hp; hp = hp->next) {
if (matching_fn(hp, user_data)) {
return hp;
}
}
return NULL;
}
bool
stonith__event_state_pending(stonith_history_t *history, void *user_data)
{
return history->state != st_failed && history->state != st_done;
}
bool
stonith__event_state_eq(stonith_history_t *history, void *user_data)
{
return history->state == GPOINTER_TO_INT(user_data);
}
bool
stonith__event_state_neq(stonith_history_t *history, void *user_data)
{
return history->state != GPOINTER_TO_INT(user_data);
}
/*!
* \internal
* \brief Check whether a given parameter exists in a fence agent's metadata
*
* \param[in] metadata Agent metadata
* \param[in] name Parameter name
*
* \retval \c true If \p name exists as a parameter in \p metadata
* \retval \c false Otherwise
*/
static bool
param_is_supported(xmlNode *metadata, const char *name)
{
char *xpath_s = crm_strdup_printf("//" PCMK_XE_PARAMETER
"[@" PCMK_XA_NAME "='%s']",
name);
xmlXPathObject *xpath = pcmk__xpath_search(metadata->doc, xpath_s);
bool supported = (pcmk__xpath_num_results(xpath) > 0);
free(xpath_s);
xmlXPathFreeObject(xpath);
return supported;
}
/*!
* \internal
* \brief Get the default host argument based on a device's agent metadata
*
* If an agent supports the "plug" parameter, default to that. Otherwise default
* to the "port" parameter if supported. Otherwise return \c NULL.
*
* \param[in] metadata Agent metadata
*
* \return Parameter name for default host argument
*/
const char *
stonith__default_host_arg(xmlNode *metadata)
{
CRM_CHECK(metadata != NULL, return NULL);
if (param_is_supported(metadata, "plug")) {
return "plug";
}
if (param_is_supported(metadata, "port")) {
return "port";
}
return NULL;
}
/*!
* \internal
* \brief Retrieve fence agent meta-data asynchronously
*
* \param[in] agent Agent to execute
* \param[in] timeout_sec Error if not complete within this time
* \param[in] callback Function to call with result (this will always be
* called, whether by this function directly or
* later via the main loop, and on success the
* metadata will be in its result argument's
* action_stdout)
* \param[in,out] user_data User data to pass to callback
*
* \return Standard Pacemaker return code
* \note The caller must use a main loop. This function is not a
* stonith_api_operations_t method because it does not need a stonith_t
* object and does not go through the fencer, but executes the agent
* directly.
*/
int
stonith__metadata_async(const char *agent, int timeout_sec,
void (*callback)(int pid,
const pcmk__action_result_t *result,
void *user_data),
void *user_data)
{
switch (get_namespace_from_agent(agent)) {
case st_namespace_rhcs:
{
stonith_action_t *action = NULL;
int rc = pcmk_ok;
action = stonith__action_create(agent, PCMK_ACTION_METADATA,
NULL, timeout_sec, NULL, NULL,
NULL);
rc = stonith__execute_async(action, user_data, callback, NULL);
if (rc != pcmk_ok) {
callback(0, stonith__action_result(action), user_data);
stonith__destroy_action(action);
}
return pcmk_legacy2rc(rc);
}
#if HAVE_STONITH_STONITH_H
case st_namespace_lha:
// LHA metadata is simply synthesized, so simulate async
{
pcmk__action_result_t result = {
.exit_status = CRM_EX_OK,
.execution_status = PCMK_EXEC_DONE,
.exit_reason = NULL,
.action_stdout = NULL,
.action_stderr = NULL,
};
stonith__lha_metadata(agent, timeout_sec,
&result.action_stdout);
callback(0, &result, user_data);
pcmk__reset_result(&result);
return pcmk_rc_ok;
}
#endif
default:
{
pcmk__action_result_t result = {
.exit_status = CRM_EX_NOSUCH,
.execution_status = PCMK_EXEC_ERROR_HARD,
.exit_reason = crm_strdup_printf("No such agent '%s'",
agent),
.action_stdout = NULL,
.action_stderr = NULL,
};
callback(0, &result, user_data);
pcmk__reset_result(&result);
return ENOENT;
}
}
}
/*!
* \internal
* \brief Return the exit status from an async action callback
*
* \param[in] data Callback data
*
* \return Exit status from callback data
*/
int
stonith__exit_status(const stonith_callback_data_t *data)
{
if ((data == NULL) || (data->opaque == NULL)) {
return CRM_EX_ERROR;
}
return ((pcmk__action_result_t *) data->opaque)->exit_status;
}
/*!
* \internal
* \brief Return the execution status from an async action callback
*
* \param[in] data Callback data
*
* \return Execution status from callback data
*/
int
stonith__execution_status(const stonith_callback_data_t *data)
{
if ((data == NULL) || (data->opaque == NULL)) {
return PCMK_EXEC_UNKNOWN;
}
return ((pcmk__action_result_t *) data->opaque)->execution_status;
}
/*!
* \internal
* \brief Return the exit reason from an async action callback
*
* \param[in] data Callback data
*
* \return Exit reason from callback data
*/
const char *
stonith__exit_reason(const stonith_callback_data_t *data)
{
if ((data == NULL) || (data->opaque == NULL)) {
return NULL;
}
return ((pcmk__action_result_t *) data->opaque)->exit_reason;
}
/*!
* \internal
* \brief Return the exit status from an event notification
*
* \param[in] event Event
*
* \return Exit status from event
*/
int
stonith__event_exit_status(const stonith_event_t *event)
{
if ((event == NULL) || (event->opaque == NULL)) {
return CRM_EX_ERROR;
} else {
struct event_private *event_private = event->opaque;
return event_private->result.exit_status;
}
}
/*!
* \internal
* \brief Return the execution status from an event notification
*
* \param[in] event Event
*
* \return Execution status from event
*/
int
stonith__event_execution_status(const stonith_event_t *event)
{
if ((event == NULL) || (event->opaque == NULL)) {
return PCMK_EXEC_UNKNOWN;
} else {
struct event_private *event_private = event->opaque;
return event_private->result.execution_status;
}
}
/*!
* \internal
* \brief Return the exit reason from an event notification
*
* \param[in] event Event
*
* \return Exit reason from event
*/
const char *
stonith__event_exit_reason(const stonith_event_t *event)
{
if ((event == NULL) || (event->opaque == NULL)) {
return NULL;
} else {
struct event_private *event_private = event->opaque;
return event_private->result.exit_reason;
}
}
/*!
* \internal
* \brief Return a human-friendly description of a fencing event
*
* \param[in] event Event to describe
*
* \return Newly allocated string with description of \p event
* \note The caller is responsible for freeing the return value.
* This function asserts on memory errors and never returns NULL.
*/
char *
stonith__event_description(const stonith_event_t *event)
{
// Use somewhat readable defaults
const char *origin = pcmk__s(event->client_origin, "a client");
const char *origin_node = pcmk__s(event->origin, "a node");
const char *executioner = pcmk__s(event->executioner, "the cluster");
const char *device = pcmk__s(event->device, "unknown");
const char *action = pcmk__s(event->action, event->operation);
const char *target = pcmk__s(event->target, "no node");
const char *reason = stonith__event_exit_reason(event);
const char *status;
if (action == NULL) {
action = "(unknown)";
}
if (stonith__event_execution_status(event) != PCMK_EXEC_DONE) {
status = pcmk_exec_status_str(stonith__event_execution_status(event));
} else if (stonith__event_exit_status(event) != CRM_EX_OK) {
status = pcmk_exec_status_str(PCMK_EXEC_ERROR);
} else {
status = crm_exit_str(CRM_EX_OK);
}
if (pcmk__str_eq(event->operation, PCMK__VALUE_ST_NOTIFY_HISTORY,
pcmk__str_none)) {
return crm_strdup_printf("Fencing history may have changed");
} else if (pcmk__str_eq(event->operation, STONITH_OP_DEVICE_ADD,
pcmk__str_none)) {
return crm_strdup_printf("A fencing device (%s) was added", device);
} else if (pcmk__str_eq(event->operation, STONITH_OP_DEVICE_DEL,
pcmk__str_none)) {
return crm_strdup_printf("A fencing device (%s) was removed", device);
} else if (pcmk__str_eq(event->operation, STONITH_OP_LEVEL_ADD,
pcmk__str_none)) {
return crm_strdup_printf("A fencing topology level (%s) was added",
device);
} else if (pcmk__str_eq(event->operation, STONITH_OP_LEVEL_DEL,
pcmk__str_none)) {
return crm_strdup_printf("A fencing topology level (%s) was removed",
device);
}
// event->operation should be PCMK__VALUE_ST_NOTIFY_FENCE at this point
return crm_strdup_printf("Operation %s of %s by %s for %s@%s: %s%s%s%s (ref=%s)",
action, target, executioner, origin, origin_node,
status,
((reason == NULL)? "" : " ("), pcmk__s(reason, ""),
((reason == NULL)? "" : ")"),
pcmk__s(event->id, "(none)"));
}
// Deprecated functions kept only for backward API compatibility
// LCOV_EXCL_START
// See comments in stonith-ng.h for why we re-declare before defining
stonith_t *stonith_api_new(void);
stonith_t *
stonith_api_new(void)
{
return stonith__api_new();
}
void stonith_api_delete(stonith_t *stonith);
void
stonith_api_delete(stonith_t *stonith)
{
stonith__api_free(stonith);
}
static void
stonith_dump_pending_op(gpointer key, gpointer value, gpointer user_data)
{
int call = GPOINTER_TO_INT(key);
stonith_callback_client_t *blob = value;
crm_debug("Call %d (%s): pending", call, pcmk__s(blob->id, "no ID"));
}
void stonith_dump_pending_callbacks(stonith_t *stonith);
void
stonith_dump_pending_callbacks(stonith_t *stonith)
{
stonith_private_t *private = stonith->st_private;
if (private->stonith_op_callback_table == NULL) {
return;
}
return g_hash_table_foreach(private->stonith_op_callback_table,
stonith_dump_pending_op, NULL);
}
bool stonith_dispatch(stonith_t *stonith_api);
bool
stonith_dispatch(stonith_t *stonith_api)
{
return (stonith__api_dispatch(stonith_api) == pcmk_rc_ok);
}
stonith_key_value_t *stonith_key_value_add(stonith_key_value_t *head,
const char *key, const char *value);
stonith_key_value_t *
stonith_key_value_add(stonith_key_value_t *head, const char *key,
const char *value)
{
return stonith__key_value_add(head, key, value);
}
void stonith_key_value_freeall(stonith_key_value_t *head, int keys, int values);
void
stonith_key_value_freeall(stonith_key_value_t *head, int keys, int values)
{
stonith__key_value_freeall(head, (keys != 0), (values != 0));
}
void stonith_history_free(stonith_history_t *head);
void
stonith_history_free(stonith_history_t *head)
{
stonith__history_free(head);
}
int stonith_api_connect_retry(stonith_t *st, const char *name,
int max_attempts);
int
stonith_api_connect_retry(stonith_t *st, const char *name, int max_attempts)
{
return pcmk_rc2legacy(stonith__api_connect_retry(st, name, max_attempts));
}
const char *stonith_op_state_str(enum op_state state);
const char *
stonith_op_state_str(enum op_state state)
{
return stonith__op_state_text(state);
}
bool stonith_agent_exists(const char *agent, int timeout);
bool
stonith_agent_exists(const char *agent, int timeout)
{
return stonith__agent_exists(agent);
}
const char *stonith_action_str(const char *action);
const char *
stonith_action_str(const char *action)
{
if (action == NULL) {
return "fencing";
} else if (strcmp(action, PCMK_ACTION_ON) == 0) {
return "unfencing";
} else if (strcmp(action, PCMK_ACTION_OFF) == 0) {
return "turning off";
} else {
return action;
}
}
enum stonith_namespace stonith_text2namespace(const char *namespace_s);
enum stonith_namespace
stonith_text2namespace(const char *namespace_s)
{
return parse_namespace(namespace_s);
}
const char *stonith_namespace2text(enum stonith_namespace st_namespace);
const char *
stonith_namespace2text(enum stonith_namespace st_namespace)
{
return namespace_text(st_namespace);
}
enum stonith_namespace stonith_get_namespace(const char *agent,
const char *namespace_s);
enum stonith_namespace
stonith_get_namespace(const char *agent, const char *namespace_s)
{
if (pcmk__str_eq(namespace_s, "internal", pcmk__str_none)) {
return st_namespace_internal;
}
return get_namespace_from_agent(agent);
}
// LCOV_EXCL_STOP
// End deprecated API
diff --git a/lib/lrmd/lrmd_client.c b/lib/lrmd/lrmd_client.c
index 6a37ab5bf4..c86d735e4b 100644
--- a/lib/lrmd/lrmd_client.c
+++ b/lib/lrmd/lrmd_client.c
@@ -1,2683 +1,2684 @@
/*
* Copyright 2012-2025 the Pacemaker project contributors
*
* The version control history for this file may have further details.
*
* This source code is licensed under the GNU Lesser General Public License
* version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY.
*/
#include
#include
#include
#include
#include // uint32_t, uint64_t
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include
#include // stonith__*
#include
#include
#include
#include
#include
#include
#define MAX_TLS_RECV_WAIT 10000
CRM_TRACE_INIT_DATA(lrmd);
static int lrmd_api_disconnect(lrmd_t * lrmd);
static int lrmd_api_is_connected(lrmd_t * lrmd);
/* IPC proxy functions */
int lrmd_internal_proxy_send(lrmd_t * lrmd, xmlNode *msg);
static void lrmd_internal_proxy_dispatch(lrmd_t *lrmd, xmlNode *msg);
void lrmd_internal_set_proxy_callback(lrmd_t * lrmd, void *userdata, void (*callback)(lrmd_t *lrmd, void *userdata, xmlNode *msg));
// GnuTLS client handshake timeout in seconds
#define TLS_HANDSHAKE_TIMEOUT 5
static void lrmd_tls_disconnect(lrmd_t * lrmd);
static int global_remote_msg_id = 0;
static void lrmd_tls_connection_destroy(gpointer userdata);
static int add_tls_to_mainloop(lrmd_t *lrmd, bool do_api_handshake);
typedef struct lrmd_private_s {
uint64_t type;
char *token;
mainloop_io_t *source;
/* IPC parameters */
crm_ipc_t *ipc;
pcmk__remote_t *remote;
/* Extra TLS parameters */
char *remote_nodename;
char *server;
int port;
pcmk__tls_t *tls;
/* while the async connection is occurring, this is the id
* of the connection timeout timer. */
int async_timer;
int sock;
/* since tls requires a round trip across the network for a
* request/reply, there are times where we just want to be able
* to send a request from the client and not wait around (or even care
* about) what the reply is. */
int expected_late_replies;
GList *pending_notify;
crm_trigger_t *process_notify;
crm_trigger_t *handshake_trigger;
lrmd_event_callback callback;
/* Internal IPC proxy msg passing for remote guests */
void (*proxy_callback)(lrmd_t *lrmd, void *userdata, xmlNode *msg);
void *proxy_callback_userdata;
char *peer_version;
} lrmd_private_t;
static int process_lrmd_handshake_reply(xmlNode *reply, lrmd_private_t *native);
static void report_async_connection_result(lrmd_t * lrmd, int rc);
static lrmd_list_t *
lrmd_list_add(lrmd_list_t * head, const char *value)
{
lrmd_list_t *p, *end;
p = pcmk__assert_alloc(1, sizeof(lrmd_list_t));
p->val = strdup(value);
end = head;
while (end && end->next) {
end = end->next;
}
if (end) {
end->next = p;
} else {
head = p;
}
return head;
}
void
lrmd_list_freeall(lrmd_list_t * head)
{
lrmd_list_t *p;
while (head) {
char *val = (char *)head->val;
p = head->next;
free(val);
free(head);
head = p;
}
}
lrmd_key_value_t *
lrmd_key_value_add(lrmd_key_value_t * head, const char *key, const char *value)
{
lrmd_key_value_t *p, *end;
p = pcmk__assert_alloc(1, sizeof(lrmd_key_value_t));
p->key = strdup(key);
p->value = strdup(value);
end = head;
while (end && end->next) {
end = end->next;
}
if (end) {
end->next = p;
} else {
head = p;
}
return head;
}
void
lrmd_key_value_freeall(lrmd_key_value_t * head)
{
lrmd_key_value_t *p;
while (head) {
p = head->next;
free(head->key);
free(head->value);
free(head);
head = p;
}
}
/*!
* \brief Create a new lrmd_event_data_t object
*
* \param[in] rsc_id ID of resource involved in event
* \param[in] task Action name
* \param[in] interval_ms Action interval
*
* \return Newly allocated and initialized lrmd_event_data_t
* \note This functions asserts on memory errors, so the return value is
* guaranteed to be non-NULL. The caller is responsible for freeing the
* result with lrmd_free_event().
*/
lrmd_event_data_t *
lrmd_new_event(const char *rsc_id, const char *task, guint interval_ms)
{
lrmd_event_data_t *event = pcmk__assert_alloc(1, sizeof(lrmd_event_data_t));
// lrmd_event_data_t has (const char *) members that lrmd_free_event() frees
event->rsc_id = pcmk__str_copy(rsc_id);
event->op_type = pcmk__str_copy(task);
event->interval_ms = interval_ms;
return event;
}
lrmd_event_data_t *
lrmd_copy_event(lrmd_event_data_t * event)
{
lrmd_event_data_t *copy = NULL;
copy = pcmk__assert_alloc(1, sizeof(lrmd_event_data_t));
copy->type = event->type;
// lrmd_event_data_t has (const char *) members that lrmd_free_event() frees
copy->rsc_id = pcmk__str_copy(event->rsc_id);
copy->op_type = pcmk__str_copy(event->op_type);
copy->user_data = pcmk__str_copy(event->user_data);
copy->output = pcmk__str_copy(event->output);
copy->remote_nodename = pcmk__str_copy(event->remote_nodename);
copy->exit_reason = pcmk__str_copy(event->exit_reason);
copy->call_id = event->call_id;
copy->timeout = event->timeout;
copy->interval_ms = event->interval_ms;
copy->start_delay = event->start_delay;
copy->rsc_deleted = event->rsc_deleted;
copy->rc = event->rc;
copy->op_status = event->op_status;
copy->t_run = event->t_run;
copy->t_rcchange = event->t_rcchange;
copy->exec_time = event->exec_time;
copy->queue_time = event->queue_time;
copy->connection_rc = event->connection_rc;
copy->params = pcmk__str_table_dup(event->params);
return copy;
}
/*!
* \brief Free an executor event
*
* \param[in,out] Executor event object to free
*/
void
lrmd_free_event(lrmd_event_data_t *event)
{
if (event == NULL) {
return;
}
// @TODO Why are these const char *?
free((void *) event->rsc_id);
free((void *) event->op_type);
free((void *) event->user_data);
free((void *) event->remote_nodename);
lrmd__reset_result(event);
if (event->params != NULL) {
g_hash_table_destroy(event->params);
}
free(event);
}
static void
lrmd_dispatch_internal(gpointer data, gpointer user_data)
{
xmlNode *msg = data;
lrmd_t *lrmd = user_data;
const char *type;
const char *proxy_session = crm_element_value(msg,
PCMK__XA_LRMD_IPC_SESSION);
lrmd_private_t *native = lrmd->lrmd_private;
lrmd_event_data_t event = { 0, };
if (proxy_session != NULL) {
/* this is proxy business */
lrmd_internal_proxy_dispatch(lrmd, msg);
return;
} else if (!native->callback) {
/* no callback set */
crm_trace("notify event received but client has not set callback");
return;
}
event.remote_nodename = native->remote_nodename;
type = crm_element_value(msg, PCMK__XA_LRMD_OP);
crm_element_value_int(msg, PCMK__XA_LRMD_CALLID, &event.call_id);
event.rsc_id = crm_element_value(msg, PCMK__XA_LRMD_RSC_ID);
if (pcmk__str_eq(type, LRMD_OP_RSC_REG, pcmk__str_none)) {
event.type = lrmd_event_register;
} else if (pcmk__str_eq(type, LRMD_OP_RSC_UNREG, pcmk__str_none)) {
event.type = lrmd_event_unregister;
} else if (pcmk__str_eq(type, LRMD_OP_RSC_EXEC, pcmk__str_none)) {
int rc = 0;
int exec_time = 0;
int queue_time = 0;
time_t epoch = 0;
crm_element_value_int(msg, PCMK__XA_LRMD_TIMEOUT, &event.timeout);
crm_element_value_ms(msg, PCMK__XA_LRMD_RSC_INTERVAL,
&event.interval_ms);
crm_element_value_int(msg, PCMK__XA_LRMD_RSC_START_DELAY,
&event.start_delay);
crm_element_value_int(msg, PCMK__XA_LRMD_EXEC_RC, &rc);
event.rc = (enum ocf_exitcode) rc;
crm_element_value_int(msg, PCMK__XA_LRMD_EXEC_OP_STATUS,
&event.op_status);
crm_element_value_int(msg, PCMK__XA_LRMD_RSC_DELETED,
&event.rsc_deleted);
crm_element_value_epoch(msg, PCMK__XA_LRMD_RUN_TIME, &epoch);
event.t_run = epoch;
crm_element_value_epoch(msg, PCMK__XA_LRMD_RCCHANGE_TIME, &epoch);
event.t_rcchange = epoch;
crm_element_value_int(msg, PCMK__XA_LRMD_EXEC_TIME, &exec_time);
CRM_LOG_ASSERT(exec_time >= 0);
event.exec_time = QB_MAX(0, exec_time);
crm_element_value_int(msg, PCMK__XA_LRMD_QUEUE_TIME, &queue_time);
CRM_LOG_ASSERT(queue_time >= 0);
event.queue_time = QB_MAX(0, queue_time);
event.op_type = crm_element_value(msg, PCMK__XA_LRMD_RSC_ACTION);
event.user_data = crm_element_value(msg,
PCMK__XA_LRMD_RSC_USERDATA_STR);
event.type = lrmd_event_exec_complete;
/* output and exit_reason may be freed by a callback */
event.output = crm_element_value_copy(msg, PCMK__XA_LRMD_RSC_OUTPUT);
lrmd__set_result(&event, event.rc, event.op_status,
crm_element_value(msg, PCMK__XA_LRMD_RSC_EXIT_REASON));
event.params = xml2list(msg);
} else if (pcmk__str_eq(type, LRMD_OP_NEW_CLIENT, pcmk__str_none)) {
event.type = lrmd_event_new_client;
} else if (pcmk__str_eq(type, LRMD_OP_POKE, pcmk__str_none)) {
event.type = lrmd_event_poke;
} else {
return;
}
crm_trace("op %s notify event received", type);
native->callback(&event);
if (event.params) {
g_hash_table_destroy(event.params);
}
lrmd__reset_result(&event);
}
// \return Always 0, to indicate that IPC mainloop source should be kept
static int
lrmd_ipc_dispatch(const char *buffer, ssize_t length, gpointer userdata)
{
lrmd_t *lrmd = userdata;
lrmd_private_t *native = lrmd->lrmd_private;
if (native->callback != NULL) {
xmlNode *msg = pcmk__xml_parse(buffer);
lrmd_dispatch_internal(msg, lrmd);
pcmk__xml_free(msg);
}
return 0;
}
static void
lrmd_free_xml(gpointer userdata)
{
pcmk__xml_free((xmlNode *) userdata);
}
static bool
remote_executor_connected(lrmd_t * lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
return (native->remote->tls_session != NULL);
}
static void
handle_remote_msg(xmlNode *xml, lrmd_t *lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
const char *msg_type = NULL;
msg_type = crm_element_value(xml, PCMK__XA_LRMD_REMOTE_MSG_TYPE);
if (pcmk__str_eq(msg_type, "notify", pcmk__str_casei)) {
lrmd_dispatch_internal(xml, lrmd);
} else if (pcmk__str_eq(msg_type, "reply", pcmk__str_casei)) {
const char *op = crm_element_value(xml, PCMK__XA_LRMD_OP);
if (native->expected_late_replies > 0) {
native->expected_late_replies--;
/* The register op message we get as a response to lrmd_handshake_async
* is a reply, so we have to handle that here.
*/
if (pcmk__str_eq(op, "register", pcmk__str_casei)) {
int rc = process_lrmd_handshake_reply(xml, native);
report_async_connection_result(lrmd, pcmk_rc2legacy(rc));
}
} else {
int reply_id = 0;
crm_element_value_int(xml, PCMK__XA_LRMD_CALLID, &reply_id);
/* if this happens, we want to know about it */
crm_err("Got outdated Pacemaker Remote reply %d", reply_id);
}
}
}
/*!
* \internal
* \brief Notify trigger handler
*
* \param[in,out] userdata API connection
*
* \return Always return G_SOURCE_CONTINUE to leave this trigger handler in the
* mainloop
*/
static int
process_pending_notifies(gpointer userdata)
{
lrmd_t *lrmd = userdata;
lrmd_private_t *native = lrmd->lrmd_private;
if (native->pending_notify == NULL) {
return G_SOURCE_CONTINUE;
}
crm_trace("Processing pending notifies");
g_list_foreach(native->pending_notify, lrmd_dispatch_internal, lrmd);
g_list_free_full(native->pending_notify, lrmd_free_xml);
native->pending_notify = NULL;
return G_SOURCE_CONTINUE;
}
/*!
* \internal
* \brief TLS dispatch function for file descriptor sources
*
* \param[in,out] userdata API connection
*
* \return -1 on error to remove the source from the mainloop, or 0 otherwise
* to leave it in the mainloop
*/
static int
lrmd_tls_dispatch(gpointer userdata)
{
lrmd_t *lrmd = userdata;
lrmd_private_t *native = lrmd->lrmd_private;
xmlNode *xml = NULL;
int rc = pcmk_rc_ok;
if (!remote_executor_connected(lrmd)) {
crm_trace("TLS dispatch triggered after disconnect");
return -1;
}
crm_trace("TLS dispatch triggered");
rc = pcmk__remote_ready(native->remote, 0);
if (rc == pcmk_rc_ok) {
rc = pcmk__read_remote_message(native->remote, -1);
}
if (rc != pcmk_rc_ok && rc != ETIME) {
crm_info("Lost %s executor connection while reading data",
(native->remote_nodename? native->remote_nodename : "local"));
lrmd_tls_disconnect(lrmd);
return -1;
}
/* If rc is ETIME, there was nothing to read but we may already have a
* full message in the buffer
*/
xml = pcmk__remote_message_xml(native->remote);
if (xml == NULL) {
return 0;
}
handle_remote_msg(xml, lrmd);
pcmk__xml_free(xml);
return 0;
}
/* Not used with mainloop */
int
lrmd_poll(lrmd_t * lrmd, int timeout)
{
lrmd_private_t *native = lrmd->lrmd_private;
switch (native->type) {
case pcmk__client_ipc:
return crm_ipc_ready(native->ipc);
case pcmk__client_tls:
if (native->pending_notify) {
return 1;
} else {
int rc = pcmk__remote_ready(native->remote, 0);
switch (rc) {
case pcmk_rc_ok:
return 1;
case ETIME:
return 0;
default:
return pcmk_rc2legacy(rc);
}
}
default:
crm_err("Unsupported executor connection type (bug?): %d",
native->type);
return -EPROTONOSUPPORT;
}
}
/* Not used with mainloop */
bool
lrmd_dispatch(lrmd_t * lrmd)
{
lrmd_private_t *private = NULL;
pcmk__assert(lrmd != NULL);
private = lrmd->lrmd_private;
switch (private->type) {
case pcmk__client_ipc:
while (crm_ipc_ready(private->ipc)) {
if (crm_ipc_read(private->ipc) > 0) {
const char *msg = crm_ipc_buffer(private->ipc);
lrmd_ipc_dispatch(msg, strlen(msg), lrmd);
+ pcmk__ipc_free_client_buffer(private->ipc);
}
}
break;
case pcmk__client_tls:
lrmd_tls_dispatch(lrmd);
break;
default:
crm_err("Unsupported executor connection type (bug?): %d",
private->type);
}
if (lrmd_api_is_connected(lrmd) == FALSE) {
crm_err("Connection closed");
return FALSE;
}
return TRUE;
}
static xmlNode *
lrmd_create_op(const char *token, const char *op, xmlNode *data, int timeout,
enum lrmd_call_options options)
{
xmlNode *op_msg = NULL;
CRM_CHECK(token != NULL, return NULL);
op_msg = pcmk__xe_create(NULL, PCMK__XE_LRMD_COMMAND);
crm_xml_add(op_msg, PCMK__XA_T, PCMK__VALUE_LRMD);
crm_xml_add(op_msg, PCMK__XA_LRMD_OP, op);
crm_xml_add_int(op_msg, PCMK__XA_LRMD_TIMEOUT, timeout);
crm_xml_add_int(op_msg, PCMK__XA_LRMD_CALLOPT, options);
if (data != NULL) {
xmlNode *wrapper = pcmk__xe_create(op_msg, PCMK__XE_LRMD_CALLDATA);
pcmk__xml_copy(wrapper, data);
}
crm_trace("Created executor %s command with call options %.8lx (%d)",
op, (long)options, options);
return op_msg;
}
static void
lrmd_ipc_connection_destroy(gpointer userdata)
{
lrmd_t *lrmd = userdata;
lrmd_private_t *native = lrmd->lrmd_private;
switch (native->type) {
case pcmk__client_ipc:
crm_info("Disconnected from local executor");
break;
case pcmk__client_tls:
crm_info("Disconnected from remote executor on %s",
native->remote_nodename);
break;
default:
crm_err("Unsupported executor connection type %d (bug?)",
native->type);
}
/* Prevent these from being cleaned up in lrmd_api_disconnect() */
native->ipc = NULL;
native->source = NULL;
if (native->callback) {
lrmd_event_data_t event = { 0, };
event.type = lrmd_event_disconnect;
event.remote_nodename = native->remote_nodename;
native->callback(&event);
}
}
static void
lrmd_tls_connection_destroy(gpointer userdata)
{
lrmd_t *lrmd = userdata;
lrmd_private_t *native = lrmd->lrmd_private;
crm_info("TLS connection destroyed");
if (native->remote->tls_session) {
gnutls_bye(native->remote->tls_session, GNUTLS_SHUT_RDWR);
gnutls_deinit(native->remote->tls_session);
native->remote->tls_session = NULL;
}
if (native->tls) {
pcmk__free_tls(native->tls);
native->tls = NULL;
}
if (native->sock >= 0) {
close(native->sock);
}
if (native->process_notify) {
mainloop_destroy_trigger(native->process_notify);
native->process_notify = NULL;
}
if (native->pending_notify) {
g_list_free_full(native->pending_notify, lrmd_free_xml);
native->pending_notify = NULL;
}
if (native->handshake_trigger != NULL) {
mainloop_destroy_trigger(native->handshake_trigger);
native->handshake_trigger = NULL;
}
free(native->remote->buffer);
free(native->remote->start_state);
native->remote->buffer = NULL;
native->remote->start_state = NULL;
native->source = 0;
native->sock = -1;
if (native->callback) {
lrmd_event_data_t event = { 0, };
event.remote_nodename = native->remote_nodename;
event.type = lrmd_event_disconnect;
native->callback(&event);
}
return;
}
// \return Standard Pacemaker return code
int
lrmd__remote_send_xml(pcmk__remote_t *session, xmlNode *msg, uint32_t id,
const char *msg_type)
{
crm_xml_add_int(msg, PCMK__XA_LRMD_REMOTE_MSG_ID, id);
crm_xml_add(msg, PCMK__XA_LRMD_REMOTE_MSG_TYPE, msg_type);
return pcmk__remote_send_xml(session, msg);
}
// \return Standard Pacemaker return code
static int
read_remote_reply(lrmd_t *lrmd, int total_timeout, int expected_reply_id,
xmlNode **reply)
{
lrmd_private_t *native = lrmd->lrmd_private;
time_t start = time(NULL);
const char *msg_type = NULL;
int reply_id = 0;
int remaining_timeout = 0;
int rc = pcmk_rc_ok;
/* A timeout of 0 here makes no sense. We have to wait a period of time
* for the response to come back. If -1 or 0, default to 10 seconds. */
if (total_timeout <= 0 || total_timeout > MAX_TLS_RECV_WAIT) {
total_timeout = MAX_TLS_RECV_WAIT;
}
for (*reply = NULL; *reply == NULL; ) {
*reply = pcmk__remote_message_xml(native->remote);
if (*reply == NULL) {
/* read some more off the tls buffer if we still have time left. */
if (remaining_timeout) {
remaining_timeout = total_timeout - ((time(NULL) - start) * 1000);
} else {
remaining_timeout = total_timeout;
}
if (remaining_timeout <= 0) {
return ETIME;
}
rc = pcmk__read_remote_message(native->remote, remaining_timeout);
if (rc != pcmk_rc_ok) {
return rc;
}
*reply = pcmk__remote_message_xml(native->remote);
if (*reply == NULL) {
return ENOMSG;
}
}
crm_element_value_int(*reply, PCMK__XA_LRMD_REMOTE_MSG_ID, &reply_id);
msg_type = crm_element_value(*reply, PCMK__XA_LRMD_REMOTE_MSG_TYPE);
if (!msg_type) {
crm_err("Empty msg type received while waiting for reply");
pcmk__xml_free(*reply);
*reply = NULL;
} else if (pcmk__str_eq(msg_type, "notify", pcmk__str_casei)) {
/* got a notify while waiting for reply, trigger the notify to be processed later */
crm_info("queueing notify");
native->pending_notify = g_list_append(native->pending_notify, *reply);
if (native->process_notify) {
crm_info("notify trigger set.");
mainloop_set_trigger(native->process_notify);
}
*reply = NULL;
} else if (!pcmk__str_eq(msg_type, "reply", pcmk__str_casei)) {
/* msg isn't a reply, make some noise */
crm_err("Expected a reply, got %s", msg_type);
pcmk__xml_free(*reply);
*reply = NULL;
} else if (reply_id != expected_reply_id) {
if (native->expected_late_replies > 0) {
native->expected_late_replies--;
} else {
crm_err("Got outdated reply, expected id %d got id %d", expected_reply_id, reply_id);
}
pcmk__xml_free(*reply);
*reply = NULL;
}
}
if (native->remote->buffer && native->process_notify) {
mainloop_set_trigger(native->process_notify);
}
return rc;
}
// \return Standard Pacemaker return code
static int
send_remote_message(lrmd_t *lrmd, xmlNode *msg)
{
int rc = pcmk_rc_ok;
lrmd_private_t *native = lrmd->lrmd_private;
global_remote_msg_id++;
if (global_remote_msg_id <= 0) {
global_remote_msg_id = 1;
}
rc = lrmd__remote_send_xml(native->remote, msg, global_remote_msg_id,
"request");
if (rc != pcmk_rc_ok) {
crm_err("Disconnecting because TLS message could not be sent to "
"Pacemaker Remote: %s", pcmk_rc_str(rc));
lrmd_tls_disconnect(lrmd);
}
return rc;
}
static int
lrmd_tls_send_recv(lrmd_t * lrmd, xmlNode * msg, int timeout, xmlNode ** reply)
{
int rc = 0;
xmlNode *xml = NULL;
if (!remote_executor_connected(lrmd)) {
return -ENOTCONN;
}
rc = send_remote_message(lrmd, msg);
if (rc != pcmk_rc_ok) {
return pcmk_rc2legacy(rc);
}
rc = read_remote_reply(lrmd, timeout, global_remote_msg_id, &xml);
if (rc != pcmk_rc_ok) {
crm_err("Disconnecting remote after request %d reply not received: %s "
QB_XS " rc=%d timeout=%dms",
global_remote_msg_id, pcmk_rc_str(rc), rc, timeout);
lrmd_tls_disconnect(lrmd);
}
if (reply) {
*reply = xml;
} else {
pcmk__xml_free(xml);
}
return pcmk_rc2legacy(rc);
}
static int
lrmd_send_xml(lrmd_t * lrmd, xmlNode * msg, int timeout, xmlNode ** reply)
{
int rc = pcmk_ok;
lrmd_private_t *native = lrmd->lrmd_private;
switch (native->type) {
case pcmk__client_ipc:
rc = crm_ipc_send(native->ipc, msg, crm_ipc_client_response, timeout, reply);
break;
case pcmk__client_tls:
rc = lrmd_tls_send_recv(lrmd, msg, timeout, reply);
break;
default:
crm_err("Unsupported executor connection type (bug?): %d",
native->type);
rc = -EPROTONOSUPPORT;
}
return rc;
}
static int
lrmd_send_xml_no_reply(lrmd_t * lrmd, xmlNode * msg)
{
int rc = pcmk_ok;
lrmd_private_t *native = lrmd->lrmd_private;
switch (native->type) {
case pcmk__client_ipc:
rc = crm_ipc_send(native->ipc, msg, crm_ipc_flags_none, 0, NULL);
break;
case pcmk__client_tls:
rc = send_remote_message(lrmd, msg);
if (rc == pcmk_rc_ok) {
/* we don't want to wait around for the reply, but
* since the request/reply protocol needs to behave the same
* as libqb, a reply will eventually come later anyway. */
native->expected_late_replies++;
}
rc = pcmk_rc2legacy(rc);
break;
default:
crm_err("Unsupported executor connection type (bug?): %d",
native->type);
rc = -EPROTONOSUPPORT;
}
return rc;
}
static int
lrmd_api_is_connected(lrmd_t * lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
switch (native->type) {
case pcmk__client_ipc:
return crm_ipc_connected(native->ipc);
case pcmk__client_tls:
return remote_executor_connected(lrmd);
default:
crm_err("Unsupported executor connection type (bug?): %d",
native->type);
return 0;
}
}
/*!
* \internal
* \brief Send a prepared API command to the executor
*
* \param[in,out] lrmd Existing connection to the executor
* \param[in] op Name of API command to send
* \param[in] data Command data XML to add to the sent command
* \param[out] output_data If expecting a reply, it will be stored here
* \param[in] timeout Timeout in milliseconds (if 0, defaults to
* a sensible value per the type of connection,
* standard vs. pacemaker remote);
* also propagated to the command XML
* \param[in] call_options Call options to pass to server when sending
* \param[in] expect_reply If true, wait for a reply from the server;
* must be true for IPC (as opposed to TLS) clients
*
* \return pcmk_ok on success, -errno on error
*/
static int
lrmd_send_command(lrmd_t *lrmd, const char *op, xmlNode *data,
xmlNode **output_data, int timeout,
enum lrmd_call_options options, bool expect_reply)
{
int rc = pcmk_ok;
lrmd_private_t *native = lrmd->lrmd_private;
xmlNode *op_msg = NULL;
xmlNode *op_reply = NULL;
if (!lrmd_api_is_connected(lrmd)) {
return -ENOTCONN;
}
if (op == NULL) {
crm_err("No operation specified");
return -EINVAL;
}
CRM_LOG_ASSERT(native->token != NULL);
crm_trace("Sending %s op to executor", op);
op_msg = lrmd_create_op(native->token, op, data, timeout, options);
if (op_msg == NULL) {
return -EINVAL;
}
if (expect_reply) {
rc = lrmd_send_xml(lrmd, op_msg, timeout, &op_reply);
} else {
rc = lrmd_send_xml_no_reply(lrmd, op_msg);
goto done;
}
if (rc < 0) {
crm_perror(LOG_ERR, "Couldn't perform %s operation (timeout=%d): %d", op, timeout, rc);
goto done;
} else if (op_reply == NULL) {
rc = -ENOMSG;
goto done;
}
rc = pcmk_ok;
crm_trace("%s op reply received", op);
if (crm_element_value_int(op_reply, PCMK__XA_LRMD_RC, &rc) != 0) {
rc = -ENOMSG;
goto done;
}
crm_log_xml_trace(op_reply, "Reply");
if (output_data) {
*output_data = op_reply;
op_reply = NULL; /* Prevent subsequent free */
}
done:
if (lrmd_api_is_connected(lrmd) == FALSE) {
crm_err("Executor disconnected");
}
pcmk__xml_free(op_msg);
pcmk__xml_free(op_reply);
return rc;
}
static int
lrmd_api_poke_connection(lrmd_t * lrmd)
{
int rc;
lrmd_private_t *native = lrmd->lrmd_private;
xmlNode *data = pcmk__xe_create(NULL, PCMK__XE_LRMD_RSC);
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
rc = lrmd_send_command(lrmd, LRMD_OP_POKE, data, NULL, 0, 0,
(native->type == pcmk__client_ipc));
pcmk__xml_free(data);
return rc < 0 ? rc : pcmk_ok;
}
// \return Standard Pacemaker return code
int
lrmd__validate_remote_settings(lrmd_t *lrmd, GHashTable *hash)
{
int rc = pcmk_rc_ok;
const char *value;
lrmd_private_t *native = lrmd->lrmd_private;
xmlNode *data = pcmk__xe_create(NULL, PCMK__XA_LRMD_OP);
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
value = g_hash_table_lookup(hash, PCMK_OPT_STONITH_WATCHDOG_TIMEOUT);
if ((value) &&
(stonith__watchdog_fencing_enabled_for_node(native->remote_nodename))) {
crm_xml_add(data, PCMK__XA_LRMD_WATCHDOG, value);
}
rc = lrmd_send_command(lrmd, LRMD_OP_CHECK, data, NULL, 0, 0,
(native->type == pcmk__client_ipc));
pcmk__xml_free(data);
return (rc < 0)? pcmk_legacy2rc(rc) : pcmk_rc_ok;
}
static xmlNode *
lrmd_handshake_hello_msg(const char *name, bool is_proxy)
{
xmlNode *hello = pcmk__xe_create(NULL, PCMK__XE_LRMD_COMMAND);
crm_xml_add(hello, PCMK__XA_T, PCMK__VALUE_LRMD);
crm_xml_add(hello, PCMK__XA_LRMD_OP, CRM_OP_REGISTER);
crm_xml_add(hello, PCMK__XA_LRMD_CLIENTNAME, name);
crm_xml_add(hello, PCMK__XA_LRMD_PROTOCOL_VERSION, LRMD_PROTOCOL_VERSION);
/* advertise that we are a proxy provider */
if (is_proxy) {
pcmk__xe_set_bool_attr(hello, PCMK__XA_LRMD_IS_IPC_PROVIDER, true);
}
return hello;
}
static int
process_lrmd_handshake_reply(xmlNode *reply, lrmd_private_t *native)
{
int rc = pcmk_rc_ok;
const char *version = crm_element_value(reply, PCMK__XA_LRMD_PROTOCOL_VERSION);
const char *msg_type = crm_element_value(reply, PCMK__XA_LRMD_OP);
const char *tmp_ticket = crm_element_value(reply, PCMK__XA_LRMD_CLIENTID);
const char *start_state = crm_element_value(reply, PCMK__XA_NODE_START_STATE);
long long uptime = -1;
crm_element_value_int(reply, PCMK__XA_LRMD_RC, &rc);
rc = pcmk_legacy2rc(rc);
/* The remote executor may add its uptime to the XML reply, which is useful
* in handling transient attributes when the connection to the remote node
* unexpectedly drops. If no parameter is given, just default to -1.
*/
crm_element_value_ll(reply, PCMK__XA_UPTIME, &uptime);
native->remote->uptime = uptime;
if (start_state) {
native->remote->start_state = strdup(start_state);
}
if (rc == EPROTO) {
crm_err("Executor protocol version mismatch between client (%s) and server (%s)",
LRMD_PROTOCOL_VERSION, version);
crm_log_xml_err(reply, "Protocol Error");
} else if (!pcmk__str_eq(msg_type, CRM_OP_REGISTER, pcmk__str_casei)) {
crm_err("Invalid registration message: %s", msg_type);
crm_log_xml_err(reply, "Bad reply");
rc = EPROTO;
} else if (tmp_ticket == NULL) {
crm_err("No registration token provided");
crm_log_xml_err(reply, "Bad reply");
rc = EPROTO;
} else {
crm_trace("Obtained registration token: %s", tmp_ticket);
native->token = strdup(tmp_ticket);
native->peer_version = strdup(version?version:"1.0"); /* Included since 1.1 */
rc = pcmk_rc_ok;
}
return rc;
}
static int
lrmd_handshake(lrmd_t * lrmd, const char *name)
{
int rc = pcmk_rc_ok;
lrmd_private_t *native = lrmd->lrmd_private;
xmlNode *reply = NULL;
xmlNode *hello = lrmd_handshake_hello_msg(name, native->proxy_callback != NULL);
rc = lrmd_send_xml(lrmd, hello, -1, &reply);
if (rc < 0) {
crm_perror(LOG_DEBUG, "Couldn't complete registration with the executor API: %d", rc);
rc = ECOMM;
} else if (reply == NULL) {
crm_err("Did not receive registration reply");
rc = EPROTO;
} else {
rc = process_lrmd_handshake_reply(reply, native);
}
pcmk__xml_free(reply);
pcmk__xml_free(hello);
if (rc != pcmk_rc_ok) {
lrmd_api_disconnect(lrmd);
}
return rc;
}
static int
lrmd_handshake_async(lrmd_t * lrmd, const char *name)
{
int rc = pcmk_rc_ok;
lrmd_private_t *native = lrmd->lrmd_private;
xmlNode *hello = lrmd_handshake_hello_msg(name, native->proxy_callback != NULL);
rc = send_remote_message(lrmd, hello);
if (rc == pcmk_rc_ok) {
native->expected_late_replies++;
} else {
lrmd_api_disconnect(lrmd);
}
pcmk__xml_free(hello);
return rc;
}
static int
lrmd_ipc_connect(lrmd_t * lrmd, int *fd)
{
int rc = pcmk_ok;
lrmd_private_t *native = lrmd->lrmd_private;
struct ipc_client_callbacks lrmd_callbacks = {
.dispatch = lrmd_ipc_dispatch,
.destroy = lrmd_ipc_connection_destroy
};
crm_info("Connecting to executor");
if (fd) {
/* No mainloop */
native->ipc = crm_ipc_new(CRM_SYSTEM_LRMD, 0);
if (native->ipc != NULL) {
rc = pcmk__connect_generic_ipc(native->ipc);
if (rc == pcmk_rc_ok) {
rc = pcmk__ipc_fd(native->ipc, fd);
}
if (rc != pcmk_rc_ok) {
crm_err("Connection to executor failed: %s", pcmk_rc_str(rc));
rc = -ENOTCONN;
}
}
} else {
native->source = mainloop_add_ipc_client(CRM_SYSTEM_LRMD, G_PRIORITY_HIGH, 0, lrmd, &lrmd_callbacks);
native->ipc = mainloop_get_ipc_client(native->source);
}
if (native->ipc == NULL) {
crm_debug("Could not connect to the executor API");
rc = -ENOTCONN;
}
return rc;
}
static void
copy_gnutls_datum(gnutls_datum_t *dest, gnutls_datum_t *source)
{
pcmk__assert((dest != NULL) && (source != NULL) && (source->data != NULL));
dest->data = gnutls_malloc(source->size);
pcmk__mem_assert(dest->data);
memcpy(dest->data, source->data, source->size);
dest->size = source->size;
}
static void
clear_gnutls_datum(gnutls_datum_t *datum)
{
gnutls_free(datum->data);
datum->data = NULL;
datum->size = 0;
}
#define KEY_READ_LEN 256 // Chunk size for reading key from file
// \return Standard Pacemaker return code
static int
read_gnutls_key(const char *location, gnutls_datum_t *key)
{
FILE *stream = NULL;
size_t buf_len = KEY_READ_LEN;
if ((location == NULL) || (key == NULL)) {
return EINVAL;
}
stream = fopen(location, "r");
if (stream == NULL) {
return errno;
}
key->data = gnutls_malloc(buf_len);
key->size = 0;
while (!feof(stream)) {
int next = fgetc(stream);
if (next == EOF) {
if (!feof(stream)) {
crm_warn("Pacemaker Remote key read was partially successful "
"(copy in memory may be corrupted)");
}
break;
}
if (key->size == buf_len) {
buf_len = key->size + KEY_READ_LEN;
key->data = gnutls_realloc(key->data, buf_len);
pcmk__assert(key->data);
}
key->data[key->size++] = (unsigned char) next;
}
fclose(stream);
if (key->size == 0) {
clear_gnutls_datum(key);
return ENOKEY;
}
return pcmk_rc_ok;
}
// Cache the most recently used Pacemaker Remote authentication key
struct key_cache_s {
time_t updated; // When cached key was read (valid for 1 minute)
const char *location; // Where cached key was read from
gnutls_datum_t key; // Cached key
};
static bool
key_is_cached(struct key_cache_s *key_cache)
{
return key_cache->updated != 0;
}
static bool
key_cache_expired(struct key_cache_s *key_cache)
{
return (time(NULL) - key_cache->updated) >= 60;
}
static void
clear_key_cache(struct key_cache_s *key_cache)
{
clear_gnutls_datum(&(key_cache->key));
if ((key_cache->updated != 0) || (key_cache->location != NULL)) {
key_cache->updated = 0;
key_cache->location = NULL;
crm_debug("Cleared Pacemaker Remote key cache");
}
}
static void
get_cached_key(struct key_cache_s *key_cache, gnutls_datum_t *key)
{
copy_gnutls_datum(key, &(key_cache->key));
crm_debug("Using cached Pacemaker Remote key from %s",
pcmk__s(key_cache->location, "unknown location"));
}
static void
cache_key(struct key_cache_s *key_cache, gnutls_datum_t *key,
const char *location)
{
key_cache->updated = time(NULL);
key_cache->location = location;
copy_gnutls_datum(&(key_cache->key), key);
crm_debug("Using (and cacheing) Pacemaker Remote key from %s",
pcmk__s(location, "unknown location"));
}
/*!
* \internal
* \brief Get Pacemaker Remote authentication key from file or cache
*
* \param[in] location Path to key file to try (this memory must
* persist across all calls of this function)
* \param[out] key Key from location or cache
*
* \return Standard Pacemaker return code
*/
static int
get_remote_key(const char *location, gnutls_datum_t *key)
{
static struct key_cache_s key_cache = { 0, };
int rc = pcmk_rc_ok;
if ((location == NULL) || (key == NULL)) {
return EINVAL;
}
if (key_is_cached(&key_cache)) {
if (key_cache_expired(&key_cache)) {
clear_key_cache(&key_cache);
} else {
get_cached_key(&key_cache, key);
return pcmk_rc_ok;
}
}
rc = read_gnutls_key(location, key);
if (rc != pcmk_rc_ok) {
return rc;
}
cache_key(&key_cache, key, location);
return pcmk_rc_ok;
}
/*!
* \internal
* \brief Initialize the Pacemaker Remote authentication key
*
* Try loading the Pacemaker Remote authentication key from cache if available,
* otherwise from these locations, in order of preference:
*
* - The value of the PCMK_authkey_location environment variable, if set
* - The Pacemaker default key file location
*
* \param[out] key Where to store key
*
* \return Standard Pacemaker return code
*/
int
lrmd__init_remote_key(gnutls_datum_t *key)
{
static const char *env_location = NULL;
static bool need_env = true;
int rc = pcmk_rc_ok;
if (need_env) {
env_location = pcmk__env_option(PCMK__ENV_AUTHKEY_LOCATION);
need_env = false;
}
// Try location in environment variable, if set
if (env_location != NULL) {
rc = get_remote_key(env_location, key);
if (rc == pcmk_rc_ok) {
return pcmk_rc_ok;
}
crm_warn("Could not read Pacemaker Remote key from %s: %s",
env_location, pcmk_rc_str(rc));
return ENOKEY;
}
// Try default location, if environment wasn't explicitly set to it
rc = get_remote_key(DEFAULT_REMOTE_KEY_LOCATION, key);
if (rc == pcmk_rc_ok) {
return pcmk_rc_ok;
}
crm_warn("Could not read Pacemaker Remote key from default location %s: %s",
DEFAULT_REMOTE_KEY_LOCATION, pcmk_rc_str(rc));
return ENOKEY;
}
static void
report_async_connection_result(lrmd_t * lrmd, int rc)
{
lrmd_private_t *native = lrmd->lrmd_private;
if (native->callback) {
lrmd_event_data_t event = { 0, };
event.type = lrmd_event_connect;
event.remote_nodename = native->remote_nodename;
event.connection_rc = rc;
native->callback(&event);
}
}
static void
tls_handshake_failed(lrmd_t *lrmd, int tls_rc, int rc)
{
lrmd_private_t *native = lrmd->lrmd_private;
crm_warn("Disconnecting after TLS handshake with "
"Pacemaker Remote server %s:%d failed: %s",
native->server, native->port,
(rc == EPROTO)? gnutls_strerror(tls_rc) : pcmk_rc_str(rc));
report_async_connection_result(lrmd, pcmk_rc2legacy(rc));
gnutls_deinit(native->remote->tls_session);
native->remote->tls_session = NULL;
lrmd_tls_connection_destroy(lrmd);
}
static void
tls_handshake_succeeded(lrmd_t *lrmd)
{
int rc = pcmk_rc_ok;
lrmd_private_t *native = lrmd->lrmd_private;
/* Now that the handshake is done, see if any client TLS certificate is
* close to its expiration date and log if so. If a TLS certificate is not
* in use, this function will just return so we don't need to check for the
* session type here.
*/
pcmk__tls_check_cert_expiration(native->remote->tls_session);
crm_info("TLS connection to Pacemaker Remote server %s:%d succeeded",
native->server, native->port);
rc = add_tls_to_mainloop(lrmd, true);
/* If add_tls_to_mainloop failed, report that right now. Otherwise, we have
* to wait until we read the async reply to report anything.
*/
if (rc != pcmk_rc_ok) {
report_async_connection_result(lrmd, pcmk_rc2legacy(rc));
}
}
/*!
* \internal
* \brief Perform a TLS client handshake with a Pacemaker Remote server
*
* \param[in] lrmd Newly established Pacemaker Remote executor connection
*
* \return Standard Pacemaker return code
*/
static int
tls_client_handshake(lrmd_t *lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
int tls_rc = GNUTLS_E_SUCCESS;
int rc = pcmk__tls_client_handshake(native->remote, TLS_HANDSHAKE_TIMEOUT,
&tls_rc);
if (rc != pcmk_rc_ok) {
tls_handshake_failed(lrmd, tls_rc, rc);
}
return rc;
}
/*!
* \internal
* \brief Add trigger and file descriptor mainloop sources for TLS
*
* \param[in,out] lrmd API connection with established TLS session
* \param[in] do_api_handshake Whether to perform executor handshake
*
* \return Standard Pacemaker return code
*/
static int
add_tls_to_mainloop(lrmd_t *lrmd, bool do_api_handshake)
{
lrmd_private_t *native = lrmd->lrmd_private;
int rc = pcmk_rc_ok;
char *name = crm_strdup_printf("pacemaker-remote-%s:%d",
native->server, native->port);
struct mainloop_fd_callbacks tls_fd_callbacks = {
.dispatch = lrmd_tls_dispatch,
.destroy = lrmd_tls_connection_destroy,
};
native->process_notify = mainloop_add_trigger(G_PRIORITY_HIGH,
process_pending_notifies, lrmd);
native->source = mainloop_add_fd(name, G_PRIORITY_HIGH, native->sock, lrmd,
&tls_fd_callbacks);
/* Async connections lose the client name provided by the API caller, so we
* have to use our generated name here to perform the executor handshake.
*
* @TODO Keep track of the caller-provided name. Perhaps we should be using
* that name in this function instead of generating one anyway.
*/
if (do_api_handshake) {
rc = lrmd_handshake_async(lrmd, name);
}
free(name);
return rc;
}
struct handshake_data_s {
lrmd_t *lrmd;
time_t start_time;
int timeout_sec;
};
static gboolean
try_handshake_cb(gpointer user_data)
{
struct handshake_data_s *hs = user_data;
lrmd_t *lrmd = hs->lrmd;
lrmd_private_t *native = lrmd->lrmd_private;
pcmk__remote_t *remote = native->remote;
int rc = pcmk_rc_ok;
int tls_rc = GNUTLS_E_SUCCESS;
if (time(NULL) >= hs->start_time + hs->timeout_sec) {
rc = ETIME;
tls_handshake_failed(lrmd, GNUTLS_E_TIMEDOUT, rc);
free(hs);
return 0;
}
rc = pcmk__tls_client_try_handshake(remote, &tls_rc);
if (rc == pcmk_rc_ok) {
tls_handshake_succeeded(lrmd);
free(hs);
return 0;
} else if (rc == EAGAIN) {
mainloop_set_trigger(native->handshake_trigger);
return 1;
} else {
rc = EKEYREJECTED;
tls_handshake_failed(lrmd, tls_rc, rc);
free(hs);
return 0;
}
}
static void
lrmd_tcp_connect_cb(void *userdata, int rc, int sock)
{
lrmd_t *lrmd = userdata;
lrmd_private_t *native = lrmd->lrmd_private;
int tls_rc = GNUTLS_E_SUCCESS;
bool use_cert = pcmk__x509_enabled();
native->async_timer = 0;
if (rc != pcmk_rc_ok) {
lrmd_tls_connection_destroy(lrmd);
crm_info("Could not connect to Pacemaker Remote at %s:%d: %s "
QB_XS " rc=%d",
native->server, native->port, pcmk_rc_str(rc), rc);
report_async_connection_result(lrmd, pcmk_rc2legacy(rc));
return;
}
/* The TCP connection was successful, so establish the TLS connection. */
native->sock = sock;
if (native->tls == NULL) {
rc = pcmk__init_tls(&native->tls, false, use_cert ? GNUTLS_CRD_CERTIFICATE : GNUTLS_CRD_PSK);
if (rc != pcmk_rc_ok) {
lrmd_tls_connection_destroy(lrmd);
report_async_connection_result(lrmd, pcmk_rc2legacy(rc));
return;
}
}
if (!use_cert) {
gnutls_datum_t psk_key = { NULL, 0 };
rc = lrmd__init_remote_key(&psk_key);
if (rc != pcmk_rc_ok) {
crm_info("Could not connect to Pacemaker Remote at %s:%d: %s "
QB_XS " rc=%d",
native->server, native->port, pcmk_rc_str(rc), rc);
lrmd_tls_connection_destroy(lrmd);
report_async_connection_result(lrmd, pcmk_rc2legacy(rc));
return;
}
pcmk__tls_add_psk_key(native->tls, &psk_key);
gnutls_free(psk_key.data);
}
native->remote->tls_session = pcmk__new_tls_session(native->tls, sock);
if (native->remote->tls_session == NULL) {
lrmd_tls_connection_destroy(lrmd);
report_async_connection_result(lrmd, -EPROTO);
return;
}
/* If the TLS handshake immediately succeeds or fails, we can handle that
* now without having to deal with mainloops and retries. Otherwise, add a
* trigger to keep trying until we get a result (or it times out).
*/
rc = pcmk__tls_client_try_handshake(native->remote, &tls_rc);
if (rc == EAGAIN) {
struct handshake_data_s *hs = NULL;
if (native->handshake_trigger != NULL) {
return;
}
hs = pcmk__assert_alloc(1, sizeof(struct handshake_data_s));
hs->lrmd = lrmd;
hs->start_time = time(NULL);
hs->timeout_sec = TLS_HANDSHAKE_TIMEOUT;
native->handshake_trigger = mainloop_add_trigger(G_PRIORITY_LOW, try_handshake_cb, hs);
mainloop_set_trigger(native->handshake_trigger);
} else if (rc == pcmk_rc_ok) {
tls_handshake_succeeded(lrmd);
} else {
tls_handshake_failed(lrmd, tls_rc, rc);
}
}
static int
lrmd_tls_connect_async(lrmd_t * lrmd, int timeout /*ms */ )
{
int rc = pcmk_rc_ok;
int timer_id = 0;
lrmd_private_t *native = lrmd->lrmd_private;
native->sock = -1;
rc = pcmk__connect_remote(native->server, native->port, timeout, &timer_id,
&(native->sock), lrmd, lrmd_tcp_connect_cb);
if (rc != pcmk_rc_ok) {
crm_warn("Pacemaker Remote connection to %s:%d failed: %s "
QB_XS " rc=%d",
native->server, native->port, pcmk_rc_str(rc), rc);
return rc;
}
native->async_timer = timer_id;
return rc;
}
static int
lrmd_tls_connect(lrmd_t * lrmd, int *fd)
{
int rc = pcmk_rc_ok;
bool use_cert = pcmk__x509_enabled();
lrmd_private_t *native = lrmd->lrmd_private;
native->sock = -1;
rc = pcmk__connect_remote(native->server, native->port, 0, NULL,
&(native->sock), NULL, NULL);
if (rc != pcmk_rc_ok) {
crm_warn("Pacemaker Remote connection to %s:%d failed: %s "
QB_XS " rc=%d",
native->server, native->port, pcmk_rc_str(rc), rc);
lrmd_tls_connection_destroy(lrmd);
return ENOTCONN;
}
if (native->tls == NULL) {
rc = pcmk__init_tls(&native->tls, false, use_cert ? GNUTLS_CRD_CERTIFICATE : GNUTLS_CRD_PSK);
if (rc != pcmk_rc_ok) {
lrmd_tls_connection_destroy(lrmd);
return rc;
}
}
if (!use_cert) {
gnutls_datum_t psk_key = { NULL, 0 };
rc = lrmd__init_remote_key(&psk_key);
if (rc != pcmk_rc_ok) {
lrmd_tls_connection_destroy(lrmd);
return rc;
}
pcmk__tls_add_psk_key(native->tls, &psk_key);
gnutls_free(psk_key.data);
}
native->remote->tls_session = pcmk__new_tls_session(native->tls, native->sock);
if (native->remote->tls_session == NULL) {
lrmd_tls_connection_destroy(lrmd);
return EPROTO;
}
if (tls_client_handshake(lrmd) != pcmk_rc_ok) {
return EKEYREJECTED;
}
crm_info("Client TLS connection established with Pacemaker Remote server %s:%d", native->server,
native->port);
if (fd) {
*fd = native->sock;
} else {
rc = add_tls_to_mainloop(lrmd, false);
}
return rc;
}
static int
lrmd_api_connect(lrmd_t * lrmd, const char *name, int *fd)
{
int rc = -ENOTCONN;
lrmd_private_t *native = lrmd->lrmd_private;
switch (native->type) {
case pcmk__client_ipc:
rc = lrmd_ipc_connect(lrmd, fd);
break;
case pcmk__client_tls:
rc = lrmd_tls_connect(lrmd, fd);
rc = pcmk_rc2legacy(rc);
break;
default:
crm_err("Unsupported executor connection type (bug?): %d",
native->type);
rc = -EPROTONOSUPPORT;
}
if (rc == pcmk_ok) {
rc = lrmd_handshake(lrmd, name);
rc = pcmk_rc2legacy(rc);
}
return rc;
}
static int
lrmd_api_connect_async(lrmd_t * lrmd, const char *name, int timeout)
{
int rc = pcmk_ok;
lrmd_private_t *native = lrmd->lrmd_private;
CRM_CHECK(native && native->callback, return -EINVAL);
switch (native->type) {
case pcmk__client_ipc:
/* fake async connection with ipc. it should be fast
* enough that we gain very little from async */
rc = lrmd_api_connect(lrmd, name, NULL);
if (!rc) {
report_async_connection_result(lrmd, rc);
}
break;
case pcmk__client_tls:
rc = lrmd_tls_connect_async(lrmd, timeout);
rc = pcmk_rc2legacy(rc);
break;
default:
crm_err("Unsupported executor connection type (bug?): %d",
native->type);
rc = -EPROTONOSUPPORT;
}
return rc;
}
static void
lrmd_ipc_disconnect(lrmd_t * lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
if (native->source != NULL) {
/* Attached to mainloop */
mainloop_del_ipc_client(native->source);
native->source = NULL;
native->ipc = NULL;
} else if (native->ipc) {
/* Not attached to mainloop */
crm_ipc_t *ipc = native->ipc;
native->ipc = NULL;
crm_ipc_close(ipc);
crm_ipc_destroy(ipc);
}
}
static void
lrmd_tls_disconnect(lrmd_t * lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
if (native->remote->tls_session) {
gnutls_bye(native->remote->tls_session, GNUTLS_SHUT_RDWR);
gnutls_deinit(native->remote->tls_session);
native->remote->tls_session = NULL;
}
if (native->async_timer) {
g_source_remove(native->async_timer);
native->async_timer = 0;
}
if (native->source != NULL) {
/* Attached to mainloop */
mainloop_del_ipc_client(native->source);
native->source = NULL;
} else if (native->sock >= 0) {
close(native->sock);
native->sock = -1;
}
if (native->pending_notify) {
g_list_free_full(native->pending_notify, lrmd_free_xml);
native->pending_notify = NULL;
}
}
static int
lrmd_api_disconnect(lrmd_t * lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
int rc = pcmk_ok;
switch (native->type) {
case pcmk__client_ipc:
crm_debug("Disconnecting from local executor");
lrmd_ipc_disconnect(lrmd);
break;
case pcmk__client_tls:
crm_debug("Disconnecting from remote executor on %s",
native->remote_nodename);
lrmd_tls_disconnect(lrmd);
break;
default:
crm_err("Unsupported executor connection type (bug?): %d",
native->type);
rc = -EPROTONOSUPPORT;
}
free(native->token);
native->token = NULL;
free(native->peer_version);
native->peer_version = NULL;
return rc;
}
static int
lrmd_api_register_rsc(lrmd_t * lrmd,
const char *rsc_id,
const char *class,
const char *provider, const char *type, enum lrmd_call_options options)
{
int rc = pcmk_ok;
xmlNode *data = NULL;
if (!class || !type || !rsc_id) {
return -EINVAL;
}
if (pcmk_is_set(pcmk_get_ra_caps(class), pcmk_ra_cap_provider)
&& (provider == NULL)) {
return -EINVAL;
}
data = pcmk__xe_create(NULL, PCMK__XE_LRMD_RSC);
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_LRMD_RSC_ID, rsc_id);
crm_xml_add(data, PCMK__XA_LRMD_CLASS, class);
crm_xml_add(data, PCMK__XA_LRMD_PROVIDER, provider);
crm_xml_add(data, PCMK__XA_LRMD_TYPE, type);
rc = lrmd_send_command(lrmd, LRMD_OP_RSC_REG, data, NULL, 0, options, true);
pcmk__xml_free(data);
return rc;
}
static int
lrmd_api_unregister_rsc(lrmd_t * lrmd, const char *rsc_id, enum lrmd_call_options options)
{
int rc = pcmk_ok;
xmlNode *data = pcmk__xe_create(NULL, PCMK__XE_LRMD_RSC);
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_LRMD_RSC_ID, rsc_id);
rc = lrmd_send_command(lrmd, LRMD_OP_RSC_UNREG, data, NULL, 0, options, true);
pcmk__xml_free(data);
return rc;
}
lrmd_rsc_info_t *
lrmd_new_rsc_info(const char *rsc_id, const char *standard,
const char *provider, const char *type)
{
lrmd_rsc_info_t *rsc_info = pcmk__assert_alloc(1, sizeof(lrmd_rsc_info_t));
rsc_info->id = pcmk__str_copy(rsc_id);
rsc_info->standard = pcmk__str_copy(standard);
rsc_info->provider = pcmk__str_copy(provider);
rsc_info->type = pcmk__str_copy(type);
return rsc_info;
}
lrmd_rsc_info_t *
lrmd_copy_rsc_info(lrmd_rsc_info_t * rsc_info)
{
return lrmd_new_rsc_info(rsc_info->id, rsc_info->standard,
rsc_info->provider, rsc_info->type);
}
void
lrmd_free_rsc_info(lrmd_rsc_info_t * rsc_info)
{
if (!rsc_info) {
return;
}
free(rsc_info->id);
free(rsc_info->type);
free(rsc_info->standard);
free(rsc_info->provider);
free(rsc_info);
}
static lrmd_rsc_info_t *
lrmd_api_get_rsc_info(lrmd_t * lrmd, const char *rsc_id, enum lrmd_call_options options)
{
lrmd_rsc_info_t *rsc_info = NULL;
xmlNode *data = pcmk__xe_create(NULL, PCMK__XE_LRMD_RSC);
xmlNode *output = NULL;
const char *class = NULL;
const char *provider = NULL;
const char *type = NULL;
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_LRMD_RSC_ID, rsc_id);
lrmd_send_command(lrmd, LRMD_OP_RSC_INFO, data, &output, 0, options, true);
pcmk__xml_free(data);
if (!output) {
return NULL;
}
class = crm_element_value(output, PCMK__XA_LRMD_CLASS);
provider = crm_element_value(output, PCMK__XA_LRMD_PROVIDER);
type = crm_element_value(output, PCMK__XA_LRMD_TYPE);
if (!class || !type) {
pcmk__xml_free(output);
return NULL;
} else if (pcmk_is_set(pcmk_get_ra_caps(class), pcmk_ra_cap_provider)
&& !provider) {
pcmk__xml_free(output);
return NULL;
}
rsc_info = lrmd_new_rsc_info(rsc_id, class, provider, type);
pcmk__xml_free(output);
return rsc_info;
}
void
lrmd_free_op_info(lrmd_op_info_t *op_info)
{
if (op_info) {
free(op_info->rsc_id);
free(op_info->action);
free(op_info->interval_ms_s);
free(op_info->timeout_ms_s);
free(op_info);
}
}
static int
lrmd_api_get_recurring_ops(lrmd_t *lrmd, const char *rsc_id, int timeout_ms,
enum lrmd_call_options options, GList **output)
{
xmlNode *data = NULL;
xmlNode *output_xml = NULL;
int rc = pcmk_ok;
if (output == NULL) {
return -EINVAL;
}
*output = NULL;
// Send request
if (rsc_id) {
data = pcmk__xe_create(NULL, PCMK__XE_LRMD_RSC);
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_LRMD_RSC_ID, rsc_id);
}
rc = lrmd_send_command(lrmd, LRMD_OP_GET_RECURRING, data, &output_xml,
timeout_ms, options, true);
if (data) {
pcmk__xml_free(data);
}
// Process reply
if ((rc != pcmk_ok) || (output_xml == NULL)) {
return rc;
}
for (const xmlNode *rsc_xml = pcmk__xe_first_child(output_xml,
PCMK__XE_LRMD_RSC, NULL,
NULL);
(rsc_xml != NULL) && (rc == pcmk_ok);
rsc_xml = pcmk__xe_next(rsc_xml, PCMK__XE_LRMD_RSC)) {
rsc_id = crm_element_value(rsc_xml, PCMK__XA_LRMD_RSC_ID);
if (rsc_id == NULL) {
crm_err("Could not parse recurring operation information from executor");
continue;
}
for (const xmlNode *op_xml = pcmk__xe_first_child(rsc_xml,
PCMK__XE_LRMD_RSC_OP,
NULL, NULL);
op_xml != NULL;
op_xml = pcmk__xe_next(op_xml, PCMK__XE_LRMD_RSC_OP)) {
lrmd_op_info_t *op_info = calloc(1, sizeof(lrmd_op_info_t));
if (op_info == NULL) {
rc = -ENOMEM;
break;
}
op_info->rsc_id = strdup(rsc_id);
op_info->action = crm_element_value_copy(op_xml,
PCMK__XA_LRMD_RSC_ACTION);
op_info->interval_ms_s =
crm_element_value_copy(op_xml, PCMK__XA_LRMD_RSC_INTERVAL);
op_info->timeout_ms_s =
crm_element_value_copy(op_xml, PCMK__XA_LRMD_TIMEOUT);
*output = g_list_prepend(*output, op_info);
}
}
pcmk__xml_free(output_xml);
return rc;
}
static void
lrmd_api_set_callback(lrmd_t * lrmd, lrmd_event_callback callback)
{
lrmd_private_t *native = lrmd->lrmd_private;
native->callback = callback;
}
void
lrmd_internal_set_proxy_callback(lrmd_t * lrmd, void *userdata, void (*callback)(lrmd_t *lrmd, void *userdata, xmlNode *msg))
{
lrmd_private_t *native = lrmd->lrmd_private;
native->proxy_callback = callback;
native->proxy_callback_userdata = userdata;
}
void
lrmd_internal_proxy_dispatch(lrmd_t *lrmd, xmlNode *msg)
{
lrmd_private_t *native = lrmd->lrmd_private;
if (native->proxy_callback) {
crm_log_xml_trace(msg, "PROXY_INBOUND");
native->proxy_callback(lrmd, native->proxy_callback_userdata, msg);
}
}
int
lrmd_internal_proxy_send(lrmd_t * lrmd, xmlNode *msg)
{
if (lrmd == NULL) {
return -ENOTCONN;
}
crm_xml_add(msg, PCMK__XA_LRMD_OP, CRM_OP_IPC_FWD);
crm_log_xml_trace(msg, "PROXY_OUTBOUND");
return lrmd_send_xml_no_reply(lrmd, msg);
}
static int
stonith_get_metadata(const char *type, char **output)
{
int rc = pcmk_ok;
stonith_t *stonith_api = stonith__api_new();
if (stonith_api == NULL) {
crm_err("Could not get fence agent meta-data: API memory allocation failed");
return -ENOMEM;
}
rc = stonith_api->cmds->metadata(stonith_api, st_opt_sync_call, type, NULL,
output, 0);
if ((rc == pcmk_ok) && (*output == NULL)) {
rc = -EIO;
}
stonith_api->cmds->free(stonith_api);
return rc;
}
static int
lrmd_api_get_metadata(lrmd_t *lrmd, const char *standard, const char *provider,
const char *type, char **output,
enum lrmd_call_options options)
{
return lrmd->cmds->get_metadata_params(lrmd, standard, provider, type,
output, options, NULL);
}
static int
lrmd_api_get_metadata_params(lrmd_t *lrmd, const char *standard,
const char *provider, const char *type,
char **output, enum lrmd_call_options options,
lrmd_key_value_t *params)
{
svc_action_t *action = NULL;
GHashTable *params_table = NULL;
if (!standard || !type) {
lrmd_key_value_freeall(params);
return -EINVAL;
}
if (pcmk__str_eq(standard, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei)) {
lrmd_key_value_freeall(params);
// stonith-class resources don't support a provider
return stonith_get_metadata(type, output);
}
params_table = pcmk__strkey_table(free, free);
for (const lrmd_key_value_t *param = params; param; param = param->next) {
pcmk__insert_dup(params_table, param->key, param->value);
}
action = services__create_resource_action(type, standard, provider, type,
PCMK_ACTION_META_DATA, 0,
PCMK_DEFAULT_ACTION_TIMEOUT_MS,
params_table, 0);
lrmd_key_value_freeall(params);
if (action == NULL) {
return -ENOMEM;
}
if (action->rc != PCMK_OCF_UNKNOWN) {
services_action_free(action);
return -EINVAL;
}
if (!services_action_sync(action)) {
crm_err("Failed to retrieve meta-data for %s:%s:%s",
standard, provider, type);
services_action_free(action);
return -EIO;
}
if (!action->stdout_data) {
crm_err("Failed to receive meta-data for %s:%s:%s",
standard, provider, type);
services_action_free(action);
return -EIO;
}
*output = strdup(action->stdout_data);
services_action_free(action);
return pcmk_ok;
}
static int
lrmd_api_exec(lrmd_t *lrmd, const char *rsc_id, const char *action,
const char *userdata, guint interval_ms,
int timeout, /* ms */
int start_delay, /* ms */
enum lrmd_call_options options, lrmd_key_value_t * params)
{
int rc = pcmk_ok;
xmlNode *data = pcmk__xe_create(NULL, PCMK__XE_LRMD_RSC);
xmlNode *args = pcmk__xe_create(data, PCMK__XE_ATTRIBUTES);
lrmd_key_value_t *tmp = NULL;
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_LRMD_RSC_ID, rsc_id);
crm_xml_add(data, PCMK__XA_LRMD_RSC_ACTION, action);
crm_xml_add(data, PCMK__XA_LRMD_RSC_USERDATA_STR, userdata);
crm_xml_add_ms(data, PCMK__XA_LRMD_RSC_INTERVAL, interval_ms);
crm_xml_add_int(data, PCMK__XA_LRMD_TIMEOUT, timeout);
crm_xml_add_int(data, PCMK__XA_LRMD_RSC_START_DELAY, start_delay);
for (tmp = params; tmp; tmp = tmp->next) {
hash2smartfield((gpointer) tmp->key, (gpointer) tmp->value, args);
}
rc = lrmd_send_command(lrmd, LRMD_OP_RSC_EXEC, data, NULL, timeout, options, true);
pcmk__xml_free(data);
lrmd_key_value_freeall(params);
return rc;
}
/* timeout is in ms */
static int
lrmd_api_exec_alert(lrmd_t *lrmd, const char *alert_id, const char *alert_path,
int timeout, lrmd_key_value_t *params)
{
int rc = pcmk_ok;
xmlNode *data = pcmk__xe_create(NULL, PCMK__XE_LRMD_ALERT);
xmlNode *args = pcmk__xe_create(data, PCMK__XE_ATTRIBUTES);
lrmd_key_value_t *tmp = NULL;
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_LRMD_ALERT_ID, alert_id);
crm_xml_add(data, PCMK__XA_LRMD_ALERT_PATH, alert_path);
crm_xml_add_int(data, PCMK__XA_LRMD_TIMEOUT, timeout);
for (tmp = params; tmp; tmp = tmp->next) {
hash2smartfield((gpointer) tmp->key, (gpointer) tmp->value, args);
}
rc = lrmd_send_command(lrmd, LRMD_OP_ALERT_EXEC, data, NULL, timeout,
lrmd_opt_notify_orig_only, true);
pcmk__xml_free(data);
lrmd_key_value_freeall(params);
return rc;
}
static int
lrmd_api_cancel(lrmd_t *lrmd, const char *rsc_id, const char *action,
guint interval_ms)
{
int rc = pcmk_ok;
xmlNode *data = pcmk__xe_create(NULL, PCMK__XE_LRMD_RSC);
crm_xml_add(data, PCMK__XA_LRMD_ORIGIN, __func__);
crm_xml_add(data, PCMK__XA_LRMD_RSC_ACTION, action);
crm_xml_add(data, PCMK__XA_LRMD_RSC_ID, rsc_id);
crm_xml_add_ms(data, PCMK__XA_LRMD_RSC_INTERVAL, interval_ms);
rc = lrmd_send_command(lrmd, LRMD_OP_RSC_CANCEL, data, NULL, 0, 0, true);
pcmk__xml_free(data);
return rc;
}
static int
list_stonith_agents(lrmd_list_t ** resources)
{
int rc = 0;
stonith_t *stonith_api = stonith__api_new();
stonith_key_value_t *stonith_resources = NULL;
stonith_key_value_t *dIter = NULL;
if (stonith_api == NULL) {
crm_err("Could not list fence agents: API memory allocation failed");
return -ENOMEM;
}
stonith_api->cmds->list_agents(stonith_api, st_opt_sync_call, NULL,
&stonith_resources, 0);
stonith_api->cmds->free(stonith_api);
for (dIter = stonith_resources; dIter; dIter = dIter->next) {
rc++;
if (resources) {
*resources = lrmd_list_add(*resources, dIter->value);
}
}
stonith__key_value_freeall(stonith_resources, true, false);
return rc;
}
static int
lrmd_api_list_agents(lrmd_t * lrmd, lrmd_list_t ** resources, const char *class,
const char *provider)
{
int rc = 0;
int stonith_count = 0; // Initially, whether to include stonith devices
if (pcmk__str_eq(class, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei)) {
stonith_count = 1;
} else {
GList *gIter = NULL;
GList *agents = resources_list_agents(class, provider);
for (gIter = agents; gIter != NULL; gIter = gIter->next) {
*resources = lrmd_list_add(*resources, (const char *)gIter->data);
rc++;
}
g_list_free_full(agents, free);
if (!class) {
stonith_count = 1;
}
}
if (stonith_count) {
// Now, if stonith devices are included, how many there are
stonith_count = list_stonith_agents(resources);
if (stonith_count > 0) {
rc += stonith_count;
}
}
if (rc == 0) {
crm_notice("No agents found for class %s", class);
rc = -EPROTONOSUPPORT;
}
return rc;
}
static bool
does_provider_have_agent(const char *agent, const char *provider, const char *class)
{
bool found = false;
GList *agents = NULL;
GList *gIter2 = NULL;
agents = resources_list_agents(class, provider);
for (gIter2 = agents; gIter2 != NULL; gIter2 = gIter2->next) {
if (pcmk__str_eq(agent, gIter2->data, pcmk__str_casei)) {
found = true;
}
}
g_list_free_full(agents, free);
return found;
}
static int
lrmd_api_list_ocf_providers(lrmd_t * lrmd, const char *agent, lrmd_list_t ** providers)
{
int rc = pcmk_ok;
char *provider = NULL;
GList *ocf_providers = NULL;
GList *gIter = NULL;
ocf_providers = resources_list_providers(PCMK_RESOURCE_CLASS_OCF);
for (gIter = ocf_providers; gIter != NULL; gIter = gIter->next) {
provider = gIter->data;
if (!agent || does_provider_have_agent(agent, provider,
PCMK_RESOURCE_CLASS_OCF)) {
*providers = lrmd_list_add(*providers, (const char *)gIter->data);
rc++;
}
}
g_list_free_full(ocf_providers, free);
return rc;
}
static int
lrmd_api_list_standards(lrmd_t * lrmd, lrmd_list_t ** supported)
{
int rc = 0;
GList *standards = NULL;
GList *gIter = NULL;
standards = resources_list_standards();
for (gIter = standards; gIter != NULL; gIter = gIter->next) {
*supported = lrmd_list_add(*supported, (const char *)gIter->data);
rc++;
}
if (list_stonith_agents(NULL) > 0) {
*supported = lrmd_list_add(*supported, PCMK_RESOURCE_CLASS_STONITH);
rc++;
}
g_list_free_full(standards, free);
return rc;
}
/*!
* \internal
* \brief Create an executor API object
*
* \param[out] api Will be set to newly created API object (it is the
* caller's responsibility to free this value with
* lrmd_api_delete() if this function succeeds)
* \param[in] nodename If the object will be used for a remote connection,
* the node name to use in cluster for remote executor
* \param[in] server If the object will be used for a remote connection,
* the resolvable host name to connect to
* \param[in] port If the object will be used for a remote connection,
* port number on \p server to connect to
*
* \return Standard Pacemaker return code
* \note If the caller leaves one of \p nodename or \p server NULL, the other's
* value will be used for both. If the caller leaves both NULL, an API
* object will be created for a local executor connection.
*/
int
lrmd__new(lrmd_t **api, const char *nodename, const char *server, int port)
{
lrmd_private_t *pvt = NULL;
if (api == NULL) {
return EINVAL;
}
*api = NULL;
// Allocate all memory needed
*api = calloc(1, sizeof(lrmd_t));
if (*api == NULL) {
return ENOMEM;
}
pvt = calloc(1, sizeof(lrmd_private_t));
if (pvt == NULL) {
lrmd_api_delete(*api);
*api = NULL;
return ENOMEM;
}
(*api)->lrmd_private = pvt;
// @TODO Do we need to do this for local connections?
pvt->remote = calloc(1, sizeof(pcmk__remote_t));
(*api)->cmds = calloc(1, sizeof(lrmd_api_operations_t));
if ((pvt->remote == NULL) || ((*api)->cmds == NULL)) {
lrmd_api_delete(*api);
*api = NULL;
return ENOMEM;
}
// Set methods
(*api)->cmds->connect = lrmd_api_connect;
(*api)->cmds->connect_async = lrmd_api_connect_async;
(*api)->cmds->is_connected = lrmd_api_is_connected;
(*api)->cmds->poke_connection = lrmd_api_poke_connection;
(*api)->cmds->disconnect = lrmd_api_disconnect;
(*api)->cmds->register_rsc = lrmd_api_register_rsc;
(*api)->cmds->unregister_rsc = lrmd_api_unregister_rsc;
(*api)->cmds->get_rsc_info = lrmd_api_get_rsc_info;
(*api)->cmds->get_recurring_ops = lrmd_api_get_recurring_ops;
(*api)->cmds->set_callback = lrmd_api_set_callback;
(*api)->cmds->get_metadata = lrmd_api_get_metadata;
(*api)->cmds->exec = lrmd_api_exec;
(*api)->cmds->cancel = lrmd_api_cancel;
(*api)->cmds->list_agents = lrmd_api_list_agents;
(*api)->cmds->list_ocf_providers = lrmd_api_list_ocf_providers;
(*api)->cmds->list_standards = lrmd_api_list_standards;
(*api)->cmds->exec_alert = lrmd_api_exec_alert;
(*api)->cmds->get_metadata_params = lrmd_api_get_metadata_params;
if ((nodename == NULL) && (server == NULL)) {
pvt->type = pcmk__client_ipc;
} else {
if (nodename == NULL) {
nodename = server;
} else if (server == NULL) {
server = nodename;
}
pvt->type = pcmk__client_tls;
pvt->remote_nodename = strdup(nodename);
pvt->server = strdup(server);
if ((pvt->remote_nodename == NULL) || (pvt->server == NULL)) {
lrmd_api_delete(*api);
*api = NULL;
return ENOMEM;
}
pvt->port = port;
if (pvt->port == 0) {
pvt->port = crm_default_remote_port();
}
}
return pcmk_rc_ok;
}
lrmd_t *
lrmd_api_new(void)
{
lrmd_t *api = NULL;
pcmk__assert(lrmd__new(&api, NULL, NULL, 0) == pcmk_rc_ok);
return api;
}
lrmd_t *
lrmd_remote_api_new(const char *nodename, const char *server, int port)
{
lrmd_t *api = NULL;
pcmk__assert(lrmd__new(&api, nodename, server, port) == pcmk_rc_ok);
return api;
}
void
lrmd_api_delete(lrmd_t * lrmd)
{
if (lrmd == NULL) {
return;
}
if (lrmd->cmds != NULL) { // Never NULL, but make static analysis happy
if (lrmd->cmds->disconnect != NULL) { // Also never really NULL
lrmd->cmds->disconnect(lrmd); // No-op if already disconnected
}
free(lrmd->cmds);
}
if (lrmd->lrmd_private != NULL) {
lrmd_private_t *native = lrmd->lrmd_private;
free(native->server);
free(native->remote_nodename);
free(native->remote);
free(native->token);
free(native->peer_version);
free(lrmd->lrmd_private);
}
free(lrmd);
}
struct metadata_cb {
void (*callback)(int pid, const pcmk__action_result_t *result,
void *user_data);
void *user_data;
};
/*!
* \internal
* \brief Process asynchronous metadata completion
*
* \param[in,out] action Metadata action that completed
*/
static void
metadata_complete(svc_action_t *action)
{
struct metadata_cb *metadata_cb = (struct metadata_cb *) action->cb_data;
pcmk__action_result_t result = PCMK__UNKNOWN_RESULT;
services__copy_result(action, &result);
pcmk__set_result_output(&result, action->stdout_data, action->stderr_data);
metadata_cb->callback(0, &result, metadata_cb->user_data);
result.action_stdout = NULL; // Prevent free, because action owns it
result.action_stderr = NULL; // Prevent free, because action owns it
pcmk__reset_result(&result);
free(metadata_cb);
}
/*!
* \internal
* \brief Retrieve agent metadata asynchronously
*
* \param[in] rsc Resource agent specification
* \param[in] callback Function to call with result (this will always be
* called, whether by this function directly or later
* via the main loop, and on success the metadata will
* be in its result argument's action_stdout)
* \param[in,out] user_data User data to pass to callback
*
* \return Standard Pacemaker return code
* \note This function is not a lrmd_api_operations_t method because it does not
* need an lrmd_t object and does not go through the executor, but
* executes the agent directly.
*/
int
lrmd__metadata_async(const lrmd_rsc_info_t *rsc,
void (*callback)(int pid,
const pcmk__action_result_t *result,
void *user_data),
void *user_data)
{
svc_action_t *action = NULL;
struct metadata_cb *metadata_cb = NULL;
pcmk__action_result_t result = PCMK__UNKNOWN_RESULT;
CRM_CHECK(callback != NULL, return EINVAL);
if ((rsc == NULL) || (rsc->standard == NULL) || (rsc->type == NULL)) {
pcmk__set_result(&result, PCMK_OCF_NOT_CONFIGURED,
PCMK_EXEC_ERROR_FATAL,
"Invalid resource specification");
callback(0, &result, user_data);
pcmk__reset_result(&result);
return EINVAL;
}
if (strcmp(rsc->standard, PCMK_RESOURCE_CLASS_STONITH) == 0) {
return stonith__metadata_async(rsc->type,
pcmk__timeout_ms2s(PCMK_DEFAULT_ACTION_TIMEOUT_MS),
callback, user_data);
}
action = services__create_resource_action(pcmk__s(rsc->id, rsc->type),
rsc->standard, rsc->provider,
rsc->type,
PCMK_ACTION_META_DATA, 0,
PCMK_DEFAULT_ACTION_TIMEOUT_MS,
NULL, 0);
if (action == NULL) {
pcmk__set_result(&result, PCMK_OCF_UNKNOWN_ERROR, PCMK_EXEC_ERROR,
"Out of memory");
callback(0, &result, user_data);
pcmk__reset_result(&result);
return ENOMEM;
}
if (action->rc != PCMK_OCF_UNKNOWN) {
services__copy_result(action, &result);
callback(0, &result, user_data);
pcmk__reset_result(&result);
services_action_free(action);
return EINVAL;
}
action->cb_data = calloc(1, sizeof(struct metadata_cb));
if (action->cb_data == NULL) {
services_action_free(action);
pcmk__set_result(&result, PCMK_OCF_UNKNOWN_ERROR, PCMK_EXEC_ERROR,
"Out of memory");
callback(0, &result, user_data);
pcmk__reset_result(&result);
return ENOMEM;
}
metadata_cb = (struct metadata_cb *) action->cb_data;
metadata_cb->callback = callback;
metadata_cb->user_data = user_data;
if (!services_action_async(action, metadata_complete)) {
services_action_free(action);
return pcmk_rc_error; // @TODO Derive from action->rc and ->status
}
// The services library has taken responsibility for action
return pcmk_rc_ok;
}
/*!
* \internal
* \brief Set the result of an executor event
*
* \param[in,out] event Executor event to set
* \param[in] rc OCF exit status of event
* \param[in] op_status Executor status of event
* \param[in] exit_reason Human-friendly description of event
*/
void
lrmd__set_result(lrmd_event_data_t *event, enum ocf_exitcode rc, int op_status,
const char *exit_reason)
{
if (event == NULL) {
return;
}
event->rc = rc;
event->op_status = op_status;
// lrmd_event_data_t has (const char *) members that lrmd_free_event() frees
pcmk__str_update((char **) &event->exit_reason, exit_reason);
}
/*!
* \internal
* \brief Clear an executor event's exit reason, output, and error output
*
* \param[in,out] event Executor event to reset
*/
void
lrmd__reset_result(lrmd_event_data_t *event)
{
if (event == NULL) {
return;
}
free((void *) event->exit_reason);
event->exit_reason = NULL;
free((void *) event->output);
event->output = NULL;
}
/*!
* \internal
* \brief Get the uptime of a remote resource connection
*
* When the cluster connects to a remote resource, part of that resource's
* handshake includes the uptime of the remote resource's connection. This
* uptime is stored in the lrmd_t object.
*
* \return The connection's uptime, or -1 if unknown
*/
time_t
lrmd__uptime(lrmd_t *lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
if (native->remote == NULL) {
return -1;
} else {
return native->remote->uptime;
}
}
const char *
lrmd__node_start_state(lrmd_t *lrmd)
{
lrmd_private_t *native = lrmd->lrmd_private;
if (native->remote == NULL) {
return NULL;
} else {
return native->remote->start_state;
}
}