diff --git a/cib/io.c b/cib/io.c index b283b619ae..e2873a8967 100644 --- a/cib/io.c +++ b/cib/io.c @@ -1,488 +1,492 @@ /* * Copyright (C) 2004 Andrew Beekhof * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public * License as published by the Free Software Foundation; either * version 2 of the License, or (at your option) any later version. * * This software is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * General Public License for more details. * * You should have received a copy of the GNU General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include extern const char *cib_root; crm_trigger_t *cib_writer = NULL; gboolean initialized = FALSE; extern int cib_status; int write_cib_contents(gpointer p); static void cib_rename(const char *old) { int new_fd; char *new = crm_strdup_printf("%s/cib.auto.XXXXXX", cib_root); crm_err("Archiving unusable file %s as %s", old, new); umask(S_IWGRP | S_IWOTH | S_IROTH); if ((new_fd = mkstemp(new) < 0) || (rename(old, new) < 0)) { crm_perror(LOG_ERR, "Couldn't rename %s as %s", old, new); crm_err("Disabling disk writes and continuing"); cib_writes_enabled = FALSE; } if (new_fd > 0) { close(new_fd); } free(new); } /* * It is the callers responsibility to free the output of this function */ static xmlNode * retrieveCib(const char *filename, const char *sigfile) { xmlNode *root = NULL; + crm_info("Reading cluster configuration file %s (digest: %s)", + filename, sigfile); switch (cib_file_read_and_verify(filename, sigfile, &root)) { case -pcmk_err_cib_corrupt: crm_warn("Continuing but %s will NOT be used.", filename); break; case -pcmk_err_cib_modified: /* Archive the original files so the contents are not lost */ crm_warn("Continuing but %s will NOT be used.", filename); cib_rename(filename); cib_rename(sigfile); break; } return root; } /* * for OSs without support for direntry->d_type, like Solaris */ #ifndef DT_UNKNOWN # define DT_UNKNOWN 0 # define DT_FIFO 1 # define DT_CHR 2 # define DT_DIR 4 # define DT_BLK 6 # define DT_REG 8 # define DT_LNK 10 # define DT_SOCK 12 # define DT_WHT 14 #endif /*DT_UNKNOWN*/ static int cib_archive_filter(const struct dirent * a) { int rc = 0; /* Looking for regular files (d_type = 8) starting with 'cib-' and not ending in .sig */ struct stat s; char *a_path = crm_strdup_printf("%s/%s", cib_root, a->d_name); if(stat(a_path, &s) != 0) { rc = errno; crm_trace("%s - stat failed: %s (%d)", a->d_name, pcmk_strerror(rc), rc); rc = 0; } else if ((s.st_mode & S_IFREG) != S_IFREG) { unsigned char dtype; #ifdef HAVE_STRUCT_DIRENT_D_TYPE dtype = a->d_type; #else switch (s.st_mode & S_IFMT) { case S_IFREG: dtype = DT_REG; break; case S_IFDIR: dtype = DT_DIR; break; case S_IFCHR: dtype = DT_CHR; break; case S_IFBLK: dtype = DT_BLK; break; case S_IFLNK: dtype = DT_LNK; break; case S_IFIFO: dtype = DT_FIFO; break; case S_IFSOCK: dtype = DT_SOCK; break; default: dtype = DT_UNKNOWN; break; } #endif crm_trace("%s - wrong type (%d)", a->d_name, dtype); } else if(strstr(a->d_name, "cib-") != a->d_name) { crm_trace("%s - wrong prefix", a->d_name); } else if(strstr(a->d_name, ".sig") != NULL) { crm_trace("%s - wrong suffix", a->d_name); } else { crm_debug("%s - candidate", a->d_name); rc = 1; } free(a_path); return rc; } static int cib_archive_sort(const struct dirent ** a, const struct dirent **b) { /* Order by creation date - most recently created file first */ int rc = 0; struct stat buf; time_t a_age = 0; time_t b_age = 0; char *a_path = crm_strdup_printf("%s/%s", cib_root, a[0]->d_name); char *b_path = crm_strdup_printf("%s/%s", cib_root, b[0]->d_name); if(stat(a_path, &buf) == 0) { a_age = buf.st_ctime; } if(stat(b_path, &buf) == 0) { b_age = buf.st_ctime; } free(a_path); free(b_path); if(a_age > b_age) { rc = 1; } else if(a_age < b_age) { rc = -1; } crm_trace("%s (%u) vs. %s (%u) : %d", a[0]->d_name, a_age, b[0]->d_name, b_age, rc); return rc; } xmlNode * readCibXmlFile(const char *dir, const char *file, gboolean discard_status) { struct dirent **namelist = NULL; int lpc = 0; char *sigfile = NULL; char *filename = NULL; const char *name = NULL; const char *value = NULL; const char *validation = NULL; const char *use_valgrind = getenv("PCMK_valgrind_enabled"); xmlNode *root = NULL; xmlNode *status = NULL; if (!crm_is_writable(dir, file, CRM_DAEMON_USER, NULL, FALSE)) { cib_status = -EACCES; return NULL; } filename = crm_concat(dir, file, '/'); sigfile = crm_concat(filename, "sig", '.'); cib_status = pcmk_ok; root = retrieveCib(filename, sigfile); free(filename); free(sigfile); if (root == NULL) { crm_warn("Primary configuration corrupt or unusable, trying backups in %s", cib_root); lpc = scandir(cib_root, &namelist, cib_archive_filter, cib_archive_sort); if (lpc < 0) { crm_perror(LOG_NOTICE, "scandir(%s) failed", cib_root); } } while (root == NULL && lpc > 1) { crm_debug("Testing %d candidates", lpc); lpc--; filename = crm_strdup_printf("%s/%s", cib_root, namelist[lpc]->d_name); sigfile = crm_concat(filename, "sig", '.'); + crm_info("Reading cluster configuration file %s (digest: %s)", + filename, sigfile); if (cib_file_read_and_verify(filename, sigfile, &root) < 0) { crm_warn("Continuing but %s will NOT be used.", filename); } else { crm_notice("Continuing with last valid configuration archive: %s", filename); } free(namelist[lpc]); free(filename); free(sigfile); } free(namelist); if (root == NULL) { root = createEmptyCib(0); crm_warn("Continuing with an empty configuration."); } if (cib_writes_enabled && use_valgrind) { if (crm_is_true(use_valgrind) || strstr(use_valgrind, "cib")) { cib_writes_enabled = FALSE; crm_err("*********************************************************"); crm_err("*** Disabling disk writes to avoid confusing Valgrind ***"); crm_err("*********************************************************"); } } status = find_xml_node(root, XML_CIB_TAG_STATUS, FALSE); if (discard_status && status != NULL) { /* strip out the status section if there is one */ free_xml(status); status = NULL; } if (status == NULL) { create_xml_node(root, XML_CIB_TAG_STATUS); } /* Do this before DTD validation happens */ /* fill in some defaults */ name = XML_ATTR_GENERATION_ADMIN; value = crm_element_value(root, name); if (value == NULL) { crm_warn("No value for %s was specified in the configuration.", name); crm_warn("The reccomended course of action is to shutdown," " run crm_verify and fix any errors it reports."); crm_warn("We will default to zero and continue but may get" " confused about which configuration to use if" " multiple nodes are powered up at the same time."); crm_xml_add_int(root, name, 0); } name = XML_ATTR_GENERATION; value = crm_element_value(root, name); if (value == NULL) { crm_xml_add_int(root, name, 0); } name = XML_ATTR_NUMUPDATES; value = crm_element_value(root, name); if (value == NULL) { crm_xml_add_int(root, name, 0); } /* unset these and require the DC/CCM to update as needed */ xml_remove_prop(root, XML_ATTR_DC_UUID); if (discard_status) { crm_log_xml_trace(root, "[on-disk]"); } validation = crm_element_value(root, XML_ATTR_VALIDATION); if (validate_xml(root, NULL, TRUE) == FALSE) { crm_err("CIB does not validate with %s", crm_str(validation)); cib_status = -pcmk_err_schema_validation; } else if (validation == NULL) { int version = 0; update_validation(&root, &version, 0, FALSE, FALSE); if (version > 0) { crm_notice("Enabling %s validation on" " the existing (sane) configuration", get_schema_name(version)); } else { crm_err("CIB does not validate with any known DTD or schema"); cib_status = -pcmk_err_schema_validation; } } return root; } /* * The caller should never free the return value */ xmlNode * get_the_CIB(void) { return the_cib; } gboolean uninitializeCib(void) { xmlNode *tmp_cib = the_cib; if (tmp_cib == NULL) { crm_debug("The CIB has already been deallocated."); return FALSE; } initialized = FALSE; the_cib = NULL; crm_debug("Deallocating the CIB."); free_xml(tmp_cib); crm_debug("The CIB has been deallocated."); return TRUE; } /* * This method will not free the old CIB pointer or the new one. * We rely on the caller to have saved a pointer to the old CIB * and to free the old/bad one depending on what is appropriate. */ gboolean initializeCib(xmlNode * new_cib) { if (new_cib == NULL) { return FALSE; } the_cib = new_cib; initialized = TRUE; return TRUE; } /* * This method will free the old CIB pointer on success and the new one * on failure. */ int activateCibXml(xmlNode * new_cib, gboolean to_disk, const char *op) { xmlNode *saved_cib = the_cib; CRM_ASSERT(new_cib != saved_cib); if (initializeCib(new_cib) == FALSE) { free_xml(new_cib); crm_err("Ignoring invalid or NULL CIB"); if (saved_cib != NULL) { crm_warn("Reverting to last known CIB"); if (initializeCib(saved_cib) == FALSE) { /* oh we are so dead */ crm_crit("Couldn't re-initialize the old CIB!"); exit(1); } } else { crm_crit("Could not write out new CIB and no saved" " version to revert to"); } return -ENODATA; } free_xml(saved_cib); if (cib_writes_enabled && cib_status == pcmk_ok && to_disk) { crm_debug("Triggering CIB write for %s op", op); mainloop_set_trigger(cib_writer); } return pcmk_ok; } static void cib_diskwrite_complete(mainloop_child_t * p, pid_t pid, int core, int signo, int exitcode) { if (signo) { crm_notice("Disk write process terminated with signal %d (pid=%d, core=%d)", signo, pid, core); } else { do_crm_log(exitcode == 0 ? LOG_TRACE : LOG_ERR, "Disk write process exited (pid=%d, rc=%d)", pid, exitcode); } if (exitcode != 0 && cib_writes_enabled) { crm_err("Disabling disk writes after write failure"); cib_writes_enabled = FALSE; } mainloop_trigger_complete(cib_writer); } int write_cib_contents(gpointer p) { int exit_rc = pcmk_ok; xmlNode *cib_local = NULL; /* Make a copy of the CIB to write (possibly in a forked child) */ if (p) { /* Synchronous write out */ cib_local = copy_xml(p); } else { int pid = 0; int bb_state = qb_log_ctl(QB_LOG_BLACKBOX, QB_LOG_CONF_STATE_GET, 0); /* Turn it off before the fork() to avoid: * - 2 processes writing to the same shared mem * - the child needing to disable it * (which would close it from underneath the parent) * This way, the shared mem files are already closed */ qb_log_ctl(QB_LOG_BLACKBOX, QB_LOG_CONF_ENABLED, QB_FALSE); pid = fork(); if (pid < 0) { crm_perror(LOG_ERR, "Disabling disk writes after fork failure"); cib_writes_enabled = FALSE; return FALSE; } if (pid) { /* Parent */ mainloop_child_add(pid, 0, "disk-writer", NULL, cib_diskwrite_complete); if (bb_state == QB_LOG_STATE_ENABLED) { /* Re-enable now that it it safe */ qb_log_ctl(QB_LOG_BLACKBOX, QB_LOG_CONF_ENABLED, QB_TRUE); } return -1; /* -1 means 'still work to do' */ } /* A-synchronous write out after a fork() */ /* In theory we can scribble on "the_cib" here and not affect the parent * But lets be safe anyway */ cib_local = copy_xml(the_cib); } /* Write the CIB */ exit_rc = cib_file_write_with_digest(cib_local, cib_root, "cib.xml"); /* A nonzero exit code will cause further writes to be disabled */ free_xml(cib_local); if (p == NULL) { /* Use _exit() because exit() could affect the parent adversely */ _exit(exit_rc); } return exit_rc; } diff --git a/extra/resources/docker-wrapper b/extra/resources/docker-wrapper index 42511f4a4a..4b0b87bdd9 100755 --- a/extra/resources/docker-wrapper +++ b/extra/resources/docker-wrapper @@ -1,497 +1,536 @@ #!/bin/bash # # Copyright (c) 2015 David Vossel # All Rights Reserved. # # This program is free software; you can redistribute it and/or modify # it under the terms of version 2 of the GNU General Public License as # published by the Free Software Foundation. # # This program is distributed in the hope that it would be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. # # Further, this software is distributed without any warranty that it is # free of the rightful claim of any third person regarding infringement # or the like. Any license provided herein, whether implied or # otherwise, applies only to this software file. Patent licenses, if # any, provided herein do not apply to combinations of this program with # other software, or any other product whatsoever. # # You should have received a copy of the GNU General Public License # along with this program; if not, write the Free Software Foundation, # Inc., 59 Temple Place - Suite 330, Boston MA 02111-1307, USA. # ####################################################################### # Initialization: : ${OCF_FUNCTIONS_DIR=${OCF_ROOT}/lib/heartbeat} . ${OCF_FUNCTIONS_DIR}/ocf-shellfuncs ####################################################################### +CONF_PREFIX="pcmk_docker" + meta_data() { cat < 1.0 Docker technology wrapper for pacemaker remote. docker wrapper - + Docker image to run resources within docker image - + Give resources within container access to cluster resources such as the CIB and the ability to manage cluster attributes. + +NOTE: Do not confuse this with the docker run command's +'--priviledged' option which gives a container permission +to access system devices. To toggle the docker run option, + set --priviledged=true as part of the ${CONF_PREFIS}_run_opts +arguments. The ${CONF_PREFIX}_privileged option only pertains +to whether or not the container has access to the cluster's +CIB or not. Some multistate resources need to be able to write +values to the cib, which would require enabling ${CONF_PREFIX}_privileged is privileged - + Add options to be appended to the 'docker run' command which is used when creating the container during the start action. This option allows users to do things such as setting a custom entry point and injecting environment variables into the newly created container. Note the '-d' option is supplied regardless of this value to force containers to run in the background. NOTE: Do not explicitly specify the --name argument in the run_opts. This agent will set --name using the resource's instance name run options + + +Allow the container to be reused after stopping the container. By default +containers are removed after stop. With the reuse option containers +will persist after the container stops. + +reuse container + + + END } ####################################################################### CLIENT="/usr/libexec/pacemaker/lrmd_internal_ctl" DOCKER_AGENT="/usr/lib/ocf/resource.d/heartbeat/docker" KEY_VAL_STR="" PROVIDER=$OCF_RESKEY_CRM_meta_provider CLASS=$OCF_RESKEY_CRM_meta_class TYPE=$OCF_RESKEY_CRM_meta_type CONTAINER=$OCF_RESKEY_CRM_meta_isolation_instance if [ -z "$CONTAINER" ]; then CONTAINER=$OCF_RESOURCE_INSTANCE fi RSC_STATE_DIR="${HA_RSCTMP}/docker-wrapper/${CONTAINER}-data/" RSC_STATE_FILE="$RSC_STATE_DIR/$OCF_RESOURCE_INSTANCE.state" CONNECTION_FAILURE=0 +HOST_LOG_DIR="${HA_RSCTMP}/docker-wrapper/${CONTAINER}-logs" +HOST_LOG_FILE="$HOST_LOG_DIR/pacemaker.log" +GUEST_LOG_DIR="/var/log/pcmk" +GUEST_LOG_FILE="$GUEST_LOG_DIR/pacemaker.log" pcmk_docker_wrapper_usage() { cat < $RSC_STATE_FILE fi } clear_state_file() { if [ -f "$RSC_STATE_FILE" ]; then rm -f $RSC_STATE_FILE fi } clear_state_dir() { [ -d "$RSC_STATE_DIR" ] || return 0 rm -rf $RSC_STATE_DIR } num_active_resources() { local count [ -d "$RSC_STATE_DIR" ] || return 0 count="$(ls $RSC_STATE_DIR | wc -w)" if [ $? -ne 0 ] || [ -z "$count" ]; then return 0 fi return $count } random_port() { local port=$(python -c 'import socket; s=socket.socket(); s.bind(("localhost", 0)); print(s.getsockname()[1]); s.close()') if [ $? -eq 0 ] && [ -n "$port" ]; then echo "$port" fi } get_active_port() { PORT="$(docker port $CONTAINER 3121 | awk -F: '{ print $2 }')" } # separate docker args from ocf resource args. separate_args() { local env key value # write out arguments to key value string for ocf agent while read -r line; do key="$(echo $line | awk -F= '{print $1}' | sed 's/^OCF_RESKEY_//g')" val="$(echo $line | awk -F= '{print $2}')" KEY_VAL_STR="$KEY_VAL_STR -k '$key' -v '$val'" - done < <(printenv | grep "^OCF.*" | grep -v "^OCF_RESKEY_pcmk_docker_.*") + done < <(printenv | grep "^OCF.*" | grep -v "^OCF_RESKEY_${CONF_PREFIX}_.*") # sanitize args for DOCKER agent's consumption while read -r line; do env="$(echo $line | awk -F= '{print $1}')" val="$(echo $line | awk -F= '{print $2}')" - key="$(echo "$env" | sed 's/^OCF_RESKEY_pcmk_docker/OCF_RESKEY/g')" + key="$(echo "$env" | sed "s/^OCF_RESKEY_${CONF_PREFIX}/OCF_RESKEY/g")" export $key="$val" - done < <(printenv | grep "^OCF_RESKEY_pcmk_docker_.*") + done < <(printenv | grep "^OCF_RESKEY_${CONF_PREFIX}_.*") - if ocf_is_true $OCF_RESKEY_pcmk_docker_privileged ; then + if ocf_is_true $OCF_RESKEY_privileged ; then export OCF_RESKEY_run_cmd="/usr/sbin/pacemaker_remoted" # on start set random port to run_opts # write port to state file... or potentially get from ps? maybe docker info or inspect as well? else export OCF_RESKEY_run_cmd="/usr/libexec/pacemaker/lrmd" fi export OCF_RESKEY_name="$CONTAINER" } monitor_container() { local rc $DOCKER_AGENT monitor rc=$? if [ $rc -ne $OCF_SUCCESS ]; then clear_state_dir return $rc fi poke_remote rc=$? if [ $rc -ne $OCF_SUCCESS ]; then # container is up without an active daemon. this is bad ocf_log err "Container, $CONTAINER, is active without a responsive pacemaker_remote instance" CONNECTION_FAILURE=1 return $OCF_ERR_GENERIC fi CONNECTION_FAILURE=0 return $rc } pcmk_docker_wrapper_monitor() { local rc monitor_container rc=$? if [ $rc -ne $OCF_SUCCESS ]; then return $rc fi client_action "monitor" rc=$? if [ $rc -eq $OCF_SUCCESS ] || [ $rc -eq $OCF_RUNNING_MASTER ]; then write_state_file else clear_state_file fi return $rc } pcmk_docker_wrapper_generic_action() { local rc monitor_container rc=$? if [ $? -ne $OCF_SUCCESS ]; then return $rc fi client_action "$1" } client_action() { local action=$1 local agent_type="-T $TYPE -C $CLASS" local rc=0 if [ -n "$PROVIDER" ]; then agent_type="$agent_type -P $PROVIDER" fi - if ocf_is_true $OCF_RESKEY_pcmk_docker_privileged ; then + if ocf_is_true $OCF_RESKEY_privileged ; then if [ -z "$PORT" ]; then get_active_port fi + export PCMK_logfile=$HOST_LOG_FILE ocf_log info "$CLIENT -c 'exec' -S '127.0.0.1' -p '$PORT' -a '$action' -r '$OCF_RESOURCE_INSTANCE' -n '$CONTAINER' '$agent_type' $KEY_VAL_STR " eval $CLIENT -c 'exec' -S '127.0.0.1' -p '$PORT' -a '$action' -r '$OCF_RESOURCE_INSTANCE' -n '$CONTAINER' '$agent_type' $KEY_VAL_STR else + export PCMK_logfile=$GUEST_LOG_FILE ocf_log info "$CLIENT -c \"exec\" -a $action -r \"$OCF_RESOURCE_INSTANCE\" $agent_type $KEY_VAL_STR" - echo "$CLIENT -c \"exec\" -a $action -r \"$OCF_RESOURCE_INSTANCE\" $agent_type $KEY_VAL_STR " | nsenter --target $(docker inspect --format {{.State.Pid}} ${CONTAINER}) --mount --uts --ipc --net --pid 2>&1 + echo "$CLIENT -c \"exec\" -a $action -r \"$OCF_RESOURCE_INSTANCE\" $agent_type $KEY_VAL_STR " | nsenter --target $(docker inspect --format {{.State.Pid}} ${CONTAINER}) --mount --uts --ipc --net --pid fi rc=$? ocf_log debug "Client action $action with result $rc" return $rc } poke_remote() { # verifies daemon in container is active - if ocf_is_true $OCF_RESKEY_pcmk_docker_privileged ; then + if ocf_is_true $OCF_RESKEY_privileged ; then get_active_port ocf_log info "Attempting to contect $CONTAINER on port $PORT" $CLIENT -c "poke" -S "127.0.0.1" -p $PORT -n $CONTAINER fi # no op for non privileged containers since we handed the # client monitor action as the monitor_cmd for the docker agent } start_container() { local rc monitor_container rc=$? if [ $rc -eq $OCF_SUCCESS ]; then return $rc fi - export OCF_RESKEY_run_opts="-e PCMK_debug=yes -e PCMK_logfile=/var/log/pacemaker.log $OCF_RESKEY_run_opts" - if ocf_is_true $OCF_RESKEY_pcmk_docker_privileged ; then + mkdir -p $HOST_LOG_DIR + export OCF_RESKEY_run_opts="-e PCMK_logfile=$GUEST_LOG_FILE $OCF_RESKEY_run_opts" + export OCF_RESKEY_run_opts="-v $HOST_LOG_DIR:$GUEST_LOG_DIR $OCF_RESKEY_run_opts" + if ocf_is_true $OCF_RESKEY_privileged ; then if ! [ -f "/etc/pacemaker/authkey" ]; then # generate an authkey if it doesn't exist. mkdir -p /etc/pacemaker/ dd if=/dev/urandom of=/etc/pacemaker/authkey bs=4096 count=1 > /dev/null 2>&1 chmod 600 /etc/pacemaker/authkey fi PORT=$(random_port) if [ -z "$PORT" ]; then ocf_exit_reason "Unable to assign random port for pacemaker remote" return $OCF_ERR_GENERIC fi export OCF_RESKEY_run_opts="-p 127.0.0.1:${PORT}:3121 $OCF_RESKEY_run_opts" export OCF_RESKEY_run_opts="-v /etc/pacemaker/authkey:/etc/pacemaker/authkey $OCF_RESKEY_run_opts" ocf_log debug "using privileged mode: run_opts=$OCF_RESKEY_run_opts" else export OCF_RESKEY_monitor_cmd="$CLIENT -c poke" fi $DOCKER_AGENT start rc=$? if [ $rc -ne $OCF_SUCCESS ]; then - ocf_exit_reason "Docker container failed to start" + + docker ps > /dev/null 2>&1 + if [ $? -ne 0 ]; then + ocf_exit_reason "docker daemon is inactive." + fi return $rc fi monitor_container } pcmk_docker_wrapper_start() { local rc start_container rc=$? if [ $rc -ne $OCF_SUCCESS ]; then return $rc fi client_action "start" rc=$? if [ $? -ne "$OCF_SUCCESS" ]; then ocf_exit_reason "Failed to start agent within container" return $rc fi pcmk_docker_wrapper_monitor - return $? + rc=$? + if [ $rc -eq $OCF_SUCCESS ]; then + ocf_log notice "$OCF_RESOURCE_INSTANCE started successfully. Container's logfile can be found at $HOST_LOG_FILE" + fi + + return $rc } stop_container() { local rc local count num_active_resources count=$? if [ $count -ne 0 ]; then ocf_log err "Failed to stop agent within container. Killing container $CONTAINER with $count active resources" fi $DOCKER_AGENT "stop" rc=$? if [ $rc -ne $OCF_SUCCESS ]; then ocf_exit_reason "Docker container failed to stop" return $rc fi clear_state_dir return $rc } stop_resource() { local rc client_action "stop" rc=$? if [ $? -ne "$OCF_SUCCESS" ]; then export OCF_RESKEY_force_stop="true" kill_now=1 else clear_state_file fi } pcmk_docker_wrapper_stop() { local rc local kill_now=0 local all_stopped=0 pcmk_docker_wrapper_monitor rc=$? if [ $rc -eq $OCF_NOT_RUNNING ]; then rc=$OCF_SUCCESS num_active_resources if [ $? -eq 0 ]; then # stop container if no more resources are running ocf_log info "Gracefully stopping container $CONTAINER because no resources are left running." stop_container rc=$? fi return $rc fi # if we can't talk to the remote daemon but the container is # active, we have to force kill the container. if [ $CONNECTION_FAILURE -eq 1 ]; then export OCF_RESKEY_force_kill="true" stop_container return $? fi # If we've gotten this far, the container is up, and we # need to gracefully stop a resource within the container. client_action "stop" rc=$? if [ $? -ne "$OCF_SUCCESS" ]; then export OCF_RESKEY_force_stop="true" # force kill the container if we fail to stop a resource. stop_container rc=$? else clear_state_file num_active_resources if [ $? -eq 0 ]; then # stop container if no more resources are running ocf_log info "Gracefully stopping container $CONTAINER because last resource has stopped" stop_container rc=$? fi fi return $rc } pcmk_docker_wrapper_validate() { + check_binary docker if [ -z "$CLASS" ] || [ -z "$TYPE" ]; then ocf_exit_reason "Update pacemaker to a version that supports container wrappers." return $OCF_ERR_CONFIGURED fi if ! [ -f "$DOCKER_AGENT" ]; then ocf_exit_reason "Requires $DOCKER_AGENT to be installed. update the resource-agents package" return $OCF_ERR_INSTALLED fi $DOCKER_AGENT validate-all return $? } case $__OCF_ACTION in meta-data) meta_data exit $OCF_SUCCESS ;; usage|help) pcmk_docker_wrapper_usage exit $OCF_SUCCESS ;; esac separate_args pcmk_docker_wrapper_validate rc=$? if [ $rc -ne 0 ]; then case $__OCF_ACTION in stop) exit $OCF_SUCCESS;; monitor) exit $OCF_NOT_RUNNING;; *) exit $rc;; esac fi case $__OCF_ACTION in start) pcmk_docker_wrapper_start;; stop) pcmk_docker_wrapper_stop;; monitor|status) pcmk_docker_wrapper_monitor;; reload|promote|demote|notify) pcmk_docker_wrapper_generic_action $__OCF_ACTION;; validate-all) pcmk_docker_wrapper_validate;; *) pcmk_docker_wrapper_usage exit $OCF_ERR_UNIMPLEMENTED ;; esac rc=$? ocf_log debug "Docker-wrapper ${OCF_RESOURCE_INSTANCE} $__OCF_ACTION : $rc" exit $rc diff --git a/lib/cib/cib_file.c b/lib/cib/cib_file.c index d2da2a20f4..5b8a64b4b6 100644 --- a/lib/cib/cib_file.c +++ b/lib/cib/cib_file.c @@ -1,860 +1,861 @@ /* * Copyright (c) 2004 International Business Machines * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * */ #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #include #define cib_flag_dirty 0x00001 #define cib_flag_live 0x00002 typedef struct cib_file_opaque_s { int flags; char *filename; } cib_file_opaque_t; int cib_file_perform_op(cib_t * cib, const char *op, const char *host, const char *section, xmlNode * data, xmlNode ** output_data, int call_options); int cib_file_perform_op_delegate(cib_t * cib, const char *op, const char *host, const char *section, xmlNode * data, xmlNode ** output_data, int call_options, const char *user_name); int cib_file_signon(cib_t * cib, const char *name, enum cib_conn_type type); int cib_file_signoff(cib_t * cib); int cib_file_free(cib_t * cib); static int cib_file_inputfd(cib_t * cib) { return -EPROTONOSUPPORT; } static int cib_file_set_connection_dnotify(cib_t * cib, void (*dnotify) (gpointer user_data)) { return -EPROTONOSUPPORT; } static int cib_file_register_notification(cib_t * cib, const char *callback, int enabled) { return -EPROTONOSUPPORT; } /*! * \internal * \brief Compare the calculated digest of an XML tree against a signature file * * \param[in] root Root of XML tree to compare * \param[in] sigfile Name of signature file containing digest to compare * * \return TRUE if digests match or signature file does not exist, else FALSE */ static gboolean cib_file_verify_digest(xmlNode *root, const char *sigfile) { gboolean passed = FALSE; char *expected = crm_read_contents(sigfile); if (expected == NULL) { switch (errno) { case 0: crm_err("On-disk digest at %s is empty", sigfile); return FALSE; case ENOENT: crm_warn("No on-disk digest present at %s", sigfile); return TRUE; default: crm_perror(LOG_ERR, "Could not read on-disk digest from %s", sigfile); return FALSE; } } passed = crm_digest_verify(root, expected); free(expected); return passed; } /*! * \internal * \brief Read an XML tree from a file and verify its digest * * \param[in] filename Name of XML file to read * \param[in] sigfile Name of signature file containing digest to compare * \param[in] root If non-NULL, will be set to pointer to parsed XML tree * * \return 0 if file was successfully read, parsed and verified, otherwise: * -errno on stat() failure, * -pcmk_err_cib_corrupt if file size is 0 or XML is not parseable, or * -pcmk_err_cib_modified if digests do not match * \note If root is non-NULL, it is the caller's responsibility to free *root on * successful return. */ int cib_file_read_and_verify(const char *filename, const char *sigfile, xmlNode **root) { int s_res; struct stat buf; char *local_sigfile = NULL; xmlNode *local_root = NULL; CRM_ASSERT(filename != NULL); if (root) { *root = NULL; } /* Verify that file exists and its size is nonzero */ s_res = stat(filename, &buf); if (s_res < 0) { crm_perror(LOG_WARNING, "Could not verify cluster configuration file %s", filename); return -errno; } else if (buf.st_size == 0) { crm_warn("Cluster configuration file %s is corrupt (size is zero)", filename); return -pcmk_err_cib_corrupt; } /* Parse XML */ - crm_info("Reading cluster configuration file %s", filename); local_root = filename2xml(filename); if (local_root == NULL) { crm_warn("Cluster configuration file %s is corrupt (unparseable as XML)", filename); return -pcmk_err_cib_corrupt; } /* If sigfile is not specified, use original file name plus .sig */ if (sigfile == NULL) { sigfile = local_sigfile = crm_concat(filename, "sig", '.'); } /* Verify that digests match */ - crm_debug("Verifying cluster configuration signature from %s", sigfile); if (cib_file_verify_digest(local_root, sigfile) == FALSE) { free(local_sigfile); free_xml(local_root); return -pcmk_err_cib_modified; } free(local_sigfile); if (root) { *root = local_root; } else { free_xml(local_root); } return pcmk_ok; } #define CIB_SERIES "cib" #define CIB_SERIES_MAX 100 #define CIB_SERIES_BZIP FALSE /* Must be false because archived copies are created with hard links */ #define CIB_LIVE_NAME CIB_SERIES ".xml" /*! * \internal * \brief Check whether a file is the live CIB * * \param[in] filename Name of file to check * * \return TRUE if file exists and has one of the possible live CIB filenames */ static gboolean cib_file_is_live(const char *filename) { if (filename != NULL) { /* Canonicalize all file names for true comparison */ char *real_filename = crm_compat_realpath(filename); if (real_filename != NULL) { const char *livenames[] = { CRM_CONFIG_DIR "/" CIB_LIVE_NAME, CRM_LEGACY_CONFIG_DIR "/" CIB_LIVE_NAME }; char *real_livename; int i; /* Compare against each possible live CIB name */ for (i = 0; i < sizeof(livenames)/sizeof(const char*); ++i) { real_livename = crm_compat_realpath(livenames[i]); if (real_livename && !strcmp(real_filename, real_livename)) { free(real_livename); return TRUE; } free(real_livename); } free(real_filename); } } return FALSE; } /* cib_file_backup() and cib_file_write_with_digest() need to chown the * written files only in limited circumstances, so these variables allow * that to be indicated without affecting external callers */ static uid_t cib_file_owner = 0; static uid_t cib_file_group = 0; static gboolean cib_do_chown = FALSE; /*! * \internal * \brief Back up a CIB * * \param[in] cib_dirname Directory containing CIB file and backups * \param[in] cib_filename Name (relative to cib_dirname) of CIB file to back up * * \return 0 on success, -1 on error */ static int cib_file_backup(const char *cib_dirname, const char *cib_filename) { int rc = 0; char *cib_path = crm_concat(cib_dirname, cib_filename, '/'); char *cib_digest = crm_concat(cib_path, "sig", '.'); /* Figure out what backup file sequence number to use */ int seq = get_last_sequence(cib_dirname, CIB_SERIES); char *backup_path = generate_series_filename(cib_dirname, CIB_SERIES, seq, CIB_SERIES_BZIP); char *backup_digest = crm_concat(backup_path, "sig", '.'); CRM_ASSERT((cib_path != NULL) && (cib_digest != NULL) && (backup_path != NULL) && (backup_digest != NULL)); /* Remove the old backups if they exist */ unlink(backup_path); unlink(backup_digest); /* Back up the CIB, by hard-linking it to the backup name */ if ((link(cib_path, backup_path) < 0) && (errno != ENOENT)) { crm_perror(LOG_ERR, "Could not archive %s by linking to %s", cib_path, backup_path); rc = -1; /* Back up the CIB signature similarly */ } else if ((link(cib_digest, backup_digest) < 0) && (errno != ENOENT)) { crm_perror(LOG_ERR, "Could not archive %s by linking to %s", cib_digest, backup_digest); rc = -1; /* Update the last counter and ensure everything is sync'd to media */ } else { write_last_sequence(cib_dirname, CIB_SERIES, seq + 1, CIB_SERIES_MAX); if (cib_do_chown) { if ((chown(backup_path, cib_file_owner, cib_file_group) < 0) && (errno != ENOENT)) { crm_perror(LOG_ERR, "Could not set owner of %s", backup_path); rc = -1; } if ((chown(backup_digest, cib_file_owner, cib_file_group) < 0) && (errno != ENOENT)) { crm_perror(LOG_ERR, "Could not set owner of %s", backup_digest); rc = -1; } if (crm_chown_last_sequence(cib_dirname, CIB_SERIES, cib_file_owner, cib_file_group) < 0) { crm_perror(LOG_ERR, "Could not set owner of %s last sequence file", cib_dirname); rc = -1; } } crm_sync_directory(cib_dirname); crm_info("Archived previous version as %s", backup_path); } free(cib_path); free(cib_digest); free(backup_path); free(backup_digest); return rc; } /*! * \internal * \brief Prepare CIB XML to be written to disk * * Set num_updates to 0, set cib-last-written to the current timestamp, * and strip out the status section. * * \param[in] root Root of CIB XML tree * * \return void */ static void cib_file_prepare_xml(xmlNode *root) { xmlNode *cib_status_root = NULL; /* Always write out with num_updates=0 and current last-written timestamp */ crm_xml_add(root, XML_ATTR_NUMUPDATES, "0"); crm_xml_add_last_written(root); /* Delete status section before writing to file, because * we discard it on startup anyway, and users get confused by it */ cib_status_root = find_xml_node(root, XML_CIB_TAG_STATUS, TRUE); CRM_LOG_ASSERT(cib_status_root != NULL); if (cib_status_root != NULL) { free_xml(cib_status_root); } } /*! * \internal * \brief Write CIB to disk, along with a signature file containing its digest * * \param[in] cib_root Root of XML tree to write * \param[in] cib_dirname Directory containing CIB and signature files * \param[in] cib_filename Name (relative to cib_dirname) of file to write * * \return pcmk_ok on success, * pcmk_err_cib_modified if existing cib_filename doesn't match digest, * pcmk_err_cib_backup if existing cib_filename couldn't be backed up, * or pcmk_err_cib_save if new cib_filename couldn't be saved */ int cib_file_write_with_digest(xmlNode *cib_root, const char *cib_dirname, const char *cib_filename) { int exit_rc = pcmk_ok; int rc, fd; char *digest = NULL; /* Detect CIB version for diagnostic purposes */ const char *epoch = crm_element_value(cib_root, XML_ATTR_GENERATION); const char *admin_epoch = crm_element_value(cib_root, XML_ATTR_GENERATION_ADMIN); /* Determine full CIB and signature pathnames */ char *cib_path = crm_concat(cib_dirname, cib_filename, '/'); char *digest_path = crm_concat(cib_path, "sig", '.'); /* Create temporary file name patterns for writing out CIB and signature */ char *tmp_cib = crm_strdup_printf("%s/cib.XXXXXX", cib_dirname); char *tmp_digest = crm_strdup_printf("%s/cib.XXXXXX", cib_dirname); CRM_ASSERT((cib_path != NULL) && (digest_path != NULL) && (tmp_cib != NULL) && (tmp_digest != NULL)); /* Ensure the admin didn't modify the existing CIB underneath us */ + crm_trace("Reading cluster configuration file %s", cib_path); rc = cib_file_read_and_verify(cib_path, NULL, NULL); if ((rc != pcmk_ok) && (rc != -ENOENT)) { crm_err("%s was manually modified while the cluster was active!", cib_path); exit_rc = pcmk_err_cib_modified; goto cleanup; } /* Back up the existing CIB */ if (cib_file_backup(cib_dirname, cib_filename) < 0) { exit_rc = pcmk_err_cib_backup; goto cleanup; } crm_debug("Writing CIB to disk"); umask(S_IWGRP | S_IWOTH | S_IROTH); cib_file_prepare_xml(cib_root); /* Write the CIB to a temporary file, so we can deploy (near) atomically */ fd = mkstemp(tmp_cib); if (fd < 0) { crm_perror(LOG_ERR, "Couldn't open temporary file %s for writing CIB", tmp_cib); exit_rc = pcmk_err_cib_save; goto cleanup; } /* Protect the temporary file */ if (fchmod(fd, S_IRUSR | S_IWUSR) < 0) { crm_perror(LOG_ERR, "Couldn't protect temporary file %s for writing CIB", tmp_cib); exit_rc = pcmk_err_cib_save; goto cleanup; } if (cib_do_chown && (fchown(fd, cib_file_owner, cib_file_group) < 0)) { crm_perror(LOG_ERR, "Couldn't protect temporary file %s for writing CIB", tmp_cib); exit_rc = pcmk_err_cib_save; goto cleanup; } /* Write out the CIB */ if (write_xml_fd(cib_root, tmp_cib, fd, FALSE) <= 0) { crm_err("Changes couldn't be written to %s", tmp_cib); exit_rc = pcmk_err_cib_save; goto cleanup; } /* Calculate CIB digest */ digest = calculate_on_disk_digest(cib_root); CRM_ASSERT(digest != NULL); crm_info("Wrote version %s.%s.0 of the CIB to disk (digest: %s)", (admin_epoch ? admin_epoch : "0"), (epoch ? epoch : "0"), digest); /* Write the CIB digest to a temporary file */ fd = mkstemp(tmp_digest); if (cib_do_chown && (fchown(fd, cib_file_owner, cib_file_group) < 0)) { crm_perror(LOG_ERR, "Couldn't protect temporary file %s for writing CIB", tmp_cib); exit_rc = pcmk_err_cib_save; goto cleanup; } if ((fd < 0) || (crm_write_sync(fd, digest) < 0)) { crm_perror(LOG_ERR, "Could not write digest to file %s", tmp_digest); exit_rc = pcmk_err_cib_save; goto cleanup; } crm_debug("Wrote digest %s to disk", digest); /* Verify that what we wrote is sane */ + crm_info("Reading cluster configuration file %s (digest: %s)", + tmp_cib, tmp_digest); rc = cib_file_read_and_verify(tmp_cib, tmp_digest, NULL); CRM_ASSERT(rc == 0); /* Rename temporary files to live, and sync directory changes to media */ crm_debug("Activating %s", tmp_cib); if (rename(tmp_cib, cib_path) < 0) { crm_perror(LOG_ERR, "Couldn't rename %s as %s", tmp_cib, cib_path); exit_rc = pcmk_err_cib_save; } if (rename(tmp_digest, digest_path) < 0) { crm_perror(LOG_ERR, "Couldn't rename %s as %s", tmp_digest, digest_path); exit_rc = pcmk_err_cib_save; } crm_sync_directory(cib_dirname); cleanup: free(cib_path); free(digest_path); free(digest); free(tmp_digest); free(tmp_cib); return exit_rc; } cib_t * cib_file_new(const char *cib_location) { cib_file_opaque_t *private = NULL; cib_t *cib = cib_new_variant(); private = calloc(1, sizeof(cib_file_opaque_t)); CRM_ASSERT((cib != NULL) && (private != NULL)); cib->variant = cib_file; cib->variant_opaque = private; if (cib_location == NULL) { cib_location = getenv("CIB_file"); } private->flags = 0; if (cib_file_is_live(cib_location)) { set_bit(private->flags, cib_flag_live); crm_trace("File %s detected as live CIB", cib_location); } private->filename = strdup(cib_location); /* assign variant specific ops */ cib->delegate_fn = cib_file_perform_op_delegate; cib->cmds->signon = cib_file_signon; cib->cmds->signoff = cib_file_signoff; cib->cmds->free = cib_file_free; cib->cmds->inputfd = cib_file_inputfd; cib->cmds->register_notification = cib_file_register_notification; cib->cmds->set_connection_dnotify = cib_file_set_connection_dnotify; return cib; } static xmlNode *in_mem_cib = NULL; /*! * \internal * \brief Read CIB from disk and validate it against XML DTD * * \param[in] filename Name of file to read CIB from * * \return pcmk_ok on success, * -ENXIO if file does not exist (or stat() otherwise fails), or * -pcmk_err_schema_validation if XML doesn't parse or validate * \note If filename is the live CIB, this will *not* verify its digest, * though that functionality would be trivial to add here. * Also, this will *not* verify that the file is writeable, * because some callers might not need to write. */ static int load_file_cib(const char *filename) { struct stat buf; xmlNode *root = NULL; const char *ignore_dtd = NULL; /* Ensure file is readable */ if (stat(filename, &buf) < 0) { return -ENXIO; } /* Parse XML from file */ root = filename2xml(filename); if (root == NULL) { return -pcmk_err_schema_validation; } /* Add a status section if not already present */ if (find_xml_node(root, XML_CIB_TAG_STATUS, FALSE) == NULL) { create_xml_node(root, XML_CIB_TAG_STATUS); } /* Validate XML against its specified DTD */ ignore_dtd = crm_element_value(root, XML_ATTR_VALIDATION); if (validate_xml(root, NULL, TRUE) == FALSE) { crm_err("CIB does not validate against %s", ignore_dtd); free_xml(root); return -pcmk_err_schema_validation; } /* Remember the parsed XML for later use */ in_mem_cib = root; return pcmk_ok; } int cib_file_signon(cib_t * cib, const char *name, enum cib_conn_type type) { int rc = pcmk_ok; cib_file_opaque_t *private = cib->variant_opaque; if (private->filename == NULL) { rc = -EINVAL; } else { rc = load_file_cib(private->filename); } if (rc == pcmk_ok) { crm_debug("%s: Opened connection to local file '%s'", name, private->filename); cib->state = cib_connected_command; cib->type = cib_command; } else { fprintf(stderr, "%s: Connection to local file '%s' failed: %s\n", name, private->filename, pcmk_strerror(rc)); } return rc; } /*! * \internal * \brief Write out the in-memory CIB to a live CIB file * * param[in] path Full path to file to write * * \return 0 on success, -1 on failure */ static int cib_file_write_live(char *path) { uid_t uid = geteuid(); struct passwd *daemon_pwent; char *sep = strrchr(path, '/'); const char *cib_dirname, *cib_filename; int rc = 0; /* Get the desired uid/gid */ errno = 0; daemon_pwent = getpwnam(CRM_DAEMON_USER); if (daemon_pwent == NULL) { crm_perror(LOG_ERR, "Could not find %s user", CRM_DAEMON_USER); return -1; } /* If we're root, we can change the ownership; * if we're daemon, anything we create will be OK; * otherwise, block access so we don't create wrong owner */ if ((uid != 0) && (uid != daemon_pwent->pw_uid)) { crm_perror(LOG_ERR, "Must be root or %s to modify live CIB", CRM_DAEMON_USER); return 0; } /* fancy footwork to separate dirname from filename * (we know the canonical name maps to the live CIB, * but the given name might be relative, or symlinked) */ if (sep == NULL) { /* no directory component specified */ cib_dirname = "./"; cib_filename = path; } else if (sep == path) { /* given name is in / */ cib_dirname = "/"; cib_filename = path + 1; } else { /* typical case; split given name into parts */ *sep = '\0'; cib_dirname = path; cib_filename = sep + 1; } /* if we're root, we want to update the file ownership */ if (uid == 0) { cib_file_owner = daemon_pwent->pw_uid; cib_file_group = daemon_pwent->pw_gid; cib_do_chown = TRUE; } /* write the file */ if (cib_file_write_with_digest(in_mem_cib, cib_dirname, cib_filename) != pcmk_ok) { rc = -1; } /* turn off file ownership changes, for other callers */ if (uid == 0) { cib_do_chown = FALSE; } /* undo fancy stuff */ if ((sep != NULL) && (*sep == '\0')) { *sep = '/'; } return rc; } /*! * \internal * \brief Sign-off method for CIB file variants * * This will write the file to disk if needed, and free the in-memory CIB. If * the file is the live CIB, it will compute and write a signature as well. * * \param[in] cib CIB object to sign off * * \return pcmk_ok on success, pcmk_err_generic on failure * \todo This method should refuse to write the live CIB if the CIB daemon is * running. */ int cib_file_signoff(cib_t * cib) { int rc = pcmk_ok; cib_file_opaque_t *private = cib->variant_opaque; crm_debug("Signing out of the CIB Service"); cib->state = cib_disconnected; cib->type = cib_no_connection; /* If the in-memory CIB has been changed, write it to disk */ if (is_set(private->flags, cib_flag_dirty)) { /* If this is the live CIB, write it out with a digest */ if (is_set(private->flags, cib_flag_live)) { if (cib_file_write_live(private->filename) < 0) { rc = pcmk_err_generic; } /* Otherwise, it's a simple write */ } else { gboolean do_bzip = (strstr(private->filename, ".bz2") != NULL); if (write_xml_file(in_mem_cib, private->filename, do_bzip) <= 0) { rc = pcmk_err_generic; } } if (rc == pcmk_ok) { crm_info("Wrote CIB to %s", private->filename); clear_bit(private->flags, cib_flag_dirty); } else { crm_err("Could not write CIB to %s", private->filename); } } /* Free the in-memory CIB */ free_xml(in_mem_cib); in_mem_cib = NULL; return rc; } int cib_file_free(cib_t * cib) { int rc = pcmk_ok; if (cib->state != cib_disconnected) { rc = cib_file_signoff(cib); } if (rc == pcmk_ok) { cib_file_opaque_t *private = cib->variant_opaque; free(private->filename); free(cib->cmds); free(private); free(cib); } else { fprintf(stderr, "Couldn't sign off: %d\n", rc); } return rc; } struct cib_func_entry { const char *op; gboolean read_only; cib_op_t fn; }; /* *INDENT-OFF* */ static struct cib_func_entry cib_file_ops[] = { {CIB_OP_QUERY, TRUE, cib_process_query}, {CIB_OP_MODIFY, FALSE, cib_process_modify}, {CIB_OP_APPLY_DIFF, FALSE, cib_process_diff}, {CIB_OP_BUMP, FALSE, cib_process_bump}, {CIB_OP_REPLACE, FALSE, cib_process_replace}, {CIB_OP_CREATE, FALSE, cib_process_create}, {CIB_OP_DELETE, FALSE, cib_process_delete}, {CIB_OP_ERASE, FALSE, cib_process_erase}, {CIB_OP_UPGRADE, FALSE, cib_process_upgrade}, }; /* *INDENT-ON* */ int cib_file_perform_op(cib_t * cib, const char *op, const char *host, const char *section, xmlNode * data, xmlNode ** output_data, int call_options) { return cib_file_perform_op_delegate(cib, op, host, section, data, output_data, call_options, NULL); } int cib_file_perform_op_delegate(cib_t * cib, const char *op, const char *host, const char *section, xmlNode * data, xmlNode ** output_data, int call_options, const char *user_name) { int rc = pcmk_ok; char *effective_user = NULL; gboolean query = FALSE; gboolean changed = FALSE; xmlNode *request = NULL; xmlNode *output = NULL; xmlNode *cib_diff = NULL; xmlNode *result_cib = NULL; cib_op_t *fn = NULL; int lpc = 0; static int max_msg_types = DIMOF(cib_file_ops); cib_file_opaque_t *private = cib->variant_opaque; crm_info("%s on %s", op, section); call_options |= (cib_no_mtime | cib_inhibit_bcast | cib_scope_local); if (cib->state == cib_disconnected) { return -ENOTCONN; } if (output_data != NULL) { *output_data = NULL; } if (op == NULL) { return -EINVAL; } for (lpc = 0; lpc < max_msg_types; lpc++) { if (safe_str_eq(op, cib_file_ops[lpc].op)) { fn = &(cib_file_ops[lpc].fn); query = cib_file_ops[lpc].read_only; break; } } if (fn == NULL) { return -EPROTONOSUPPORT; } cib->call_id++; request = cib_create_op(cib->call_id, "dummy-token", op, host, section, data, call_options, user_name); #if ENABLE_ACL if(user_name) { crm_xml_add(request, XML_ACL_TAG_USER, user_name); } crm_trace("Performing %s operation as %s", op, user_name); #endif /* Mirror the logic in cib_prepare_common() */ if (section != NULL && data != NULL && crm_str_eq(crm_element_name(data), XML_TAG_CIB, TRUE)) { data = get_object_root(section, data); } rc = cib_perform_op(op, call_options, fn, query, section, request, data, TRUE, &changed, in_mem_cib, &result_cib, &cib_diff, &output); free_xml(request); if (rc == -pcmk_err_schema_validation) { validate_xml_verbose(result_cib); } if (rc != pcmk_ok) { free_xml(result_cib); } else if (query == FALSE) { xml_log_patchset(LOG_DEBUG, "cib:diff", cib_diff); free_xml(in_mem_cib); in_mem_cib = result_cib; set_bit(private->flags, cib_flag_dirty); } free_xml(cib_diff); if (cib->op_callback != NULL) { cib->op_callback(NULL, cib->call_id, rc, output); } if (output_data && output) { if(output == in_mem_cib) { *output_data = copy_xml(output); } else { *output_data = output; } } else if(output != in_mem_cib) { free_xml(output); } free(effective_user); return rc; } diff --git a/lib/pengine/clone.c b/lib/pengine/clone.c index 6213694a7b..741cabb897 100644 --- a/lib/pengine/clone.c +++ b/lib/pengine/clone.c @@ -1,603 +1,606 @@ /* * Copyright (C) 2004 Andrew Beekhof * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ #include #include #include #include #include #include #define VARIANT_CLONE 1 #include "./variant.h" void clone_create_notifications(resource_t * rsc, action_t * action, action_t * action_complete, pe_working_set_t * data_set); void force_non_unique_clone(resource_t * rsc, const char *rid, pe_working_set_t * data_set); resource_t *create_child_clone(resource_t * rsc, int sub_id, pe_working_set_t * data_set); static void mark_as_orphan(resource_t * rsc) { GListPtr gIter = rsc->children; set_bit(rsc->flags, pe_rsc_orphan); for (; gIter != NULL; gIter = gIter->next) { resource_t *child = (resource_t *) gIter->data; mark_as_orphan(child); } } void force_non_unique_clone(resource_t * rsc, const char *rid, pe_working_set_t * data_set) { if (rsc->variant == pe_clone || rsc->variant == pe_master) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); crm_config_warn("Clones %s contains non-OCF resource %s and so " "can only be used as an anonymous clone. " "Set the " XML_RSC_ATTR_UNIQUE " meta attribute to false", rsc->id, rid); clone_data->clone_node_max = 1; clone_data->clone_max = g_list_length(data_set->nodes); clear_bit_recursive(rsc, pe_rsc_unique); } } resource_t * find_clone_instance(resource_t * rsc, const char *sub_id, pe_working_set_t * data_set) { char *child_id = NULL; resource_t *child = NULL; const char *child_base = NULL; clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); child_base = ID(clone_data->xml_obj_child); child_id = crm_concat(child_base, sub_id, ':'); child = pe_find_resource(rsc->children, child_id); free(child_id); return child; } resource_t * create_child_clone(resource_t * rsc, int sub_id, pe_working_set_t * data_set) { gboolean as_orphan = FALSE; char *inc_num = NULL; char *inc_max = NULL; resource_t *child_rsc = NULL; xmlNode *child_copy = NULL; clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); CRM_CHECK(clone_data->xml_obj_child != NULL, return FALSE); if (sub_id < 0) { as_orphan = TRUE; sub_id = clone_data->total_clones; } inc_num = crm_itoa(sub_id); inc_max = crm_itoa(clone_data->clone_max); child_copy = copy_xml(clone_data->xml_obj_child); crm_xml_add(child_copy, XML_RSC_ATTR_INCARNATION, inc_num); if (common_unpack(child_copy, &child_rsc, rsc, data_set) == FALSE) { pe_err("Failed unpacking resource %s", crm_element_value(child_copy, XML_ATTR_ID)); child_rsc = NULL; goto bail; } /* child_rsc->globally_unique = rsc->globally_unique; */ CRM_ASSERT(child_rsc); clone_data->total_clones += 1; pe_rsc_trace(child_rsc, "Setting clone attributes for: %s", child_rsc->id); rsc->children = g_list_append(rsc->children, child_rsc); if (as_orphan) { mark_as_orphan(child_rsc); } add_hash_param(child_rsc->meta, XML_RSC_ATTR_INCARNATION_MAX, inc_max); print_resource(LOG_DEBUG_3, "Added ", child_rsc, FALSE); bail: free(inc_num); free(inc_max); return child_rsc; } gboolean master_unpack(resource_t * rsc, pe_working_set_t * data_set) { const char *master_max = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_MASTER_MAX); const char *master_node_max = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_MASTER_NODEMAX); g_hash_table_replace(rsc->meta, strdup("stateful"), strdup(XML_BOOLEAN_TRUE)); if (clone_unpack(rsc, data_set)) { clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); clone_data->master_max = crm_parse_int(master_max, "1"); clone_data->master_node_max = crm_parse_int(master_node_max, "1"); return TRUE; } return FALSE; } gboolean clone_unpack(resource_t * rsc, pe_working_set_t * data_set) { int lpc = 0; const char *type = NULL; int num_xml_children = 0; xmlNode *a_child = NULL; xmlNode *xml_obj = rsc->xml; clone_variant_data_t *clone_data = NULL; const char *ordered = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_ORDERED); const char *interleave = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_INTERLEAVE); const char *max_clones = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_INCARNATION_MAX); const char *max_clones_node = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_INCARNATION_NODEMAX); pe_rsc_trace(rsc, "Processing resource %s...", rsc->id); clone_data = calloc(1, sizeof(clone_variant_data_t)); rsc->variant_opaque = clone_data; clone_data->interleave = FALSE; clone_data->ordered = FALSE; clone_data->active_clones = 0; clone_data->xml_obj_child = NULL; clone_data->clone_node_max = crm_parse_int(max_clones_node, "1"); if (max_clones) { clone_data->clone_max = crm_parse_int(max_clones, "1"); } else if (g_list_length(data_set->nodes) > 0) { clone_data->clone_max = g_list_length(data_set->nodes); } else { clone_data->clone_max = 1; /* Handy during crm_verify */ } if (crm_is_true(interleave)) { clone_data->interleave = TRUE; } if (crm_is_true(ordered)) { clone_data->ordered = TRUE; } if ((rsc->flags & pe_rsc_unique) == 0 && clone_data->clone_node_max > 1) { crm_config_err("Anonymous clones (%s) may only support one copy per node", rsc->id); clone_data->clone_node_max = 1; } pe_rsc_trace(rsc, "Options for %s", rsc->id); pe_rsc_trace(rsc, "\tClone max: %d", clone_data->clone_max); pe_rsc_trace(rsc, "\tClone node max: %d", clone_data->clone_node_max); pe_rsc_trace(rsc, "\tClone is unique: %s", is_set(rsc->flags, pe_rsc_unique) ? "true" : "false"); clone_data->xml_obj_child = find_xml_node(xml_obj, XML_CIB_TAG_GROUP, FALSE); if (clone_data->xml_obj_child == NULL) { clone_data->xml_obj_child = find_xml_node(xml_obj, XML_CIB_TAG_RESOURCE, TRUE); for (a_child = __xml_first_child(xml_obj); a_child != NULL; a_child = __xml_next(a_child)) { if (crm_str_eq((const char *)a_child->name, XML_CIB_TAG_RESOURCE, TRUE)) { num_xml_children++; } } } if (clone_data->xml_obj_child == NULL) { crm_config_err("%s has nothing to clone", rsc->id); return FALSE; } for (a_child = __xml_first_child(xml_obj); a_child != NULL; a_child = __xml_next(a_child)) { if (crm_str_eq((const char *)a_child->name, type, TRUE)) { num_xml_children++; } } if (num_xml_children > 1) { crm_config_err("%s has too many children. Only the first (%s) will be cloned.", rsc->id, ID(clone_data->xml_obj_child)); } /* * Make clones ever so slightly sticky by default * * This helps ensure clone instances are not shuffled around the cluster * for no benefit in situations when pre-allocation is not appropriate */ if (g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_STICKINESS) == NULL) { add_hash_param(rsc->meta, XML_RSC_ATTR_STICKINESS, "1"); } pe_rsc_trace(rsc, "\tClone is unique (fixed): %s", is_set(rsc->flags, pe_rsc_unique) ? "true" : "false"); clone_data->notify_confirm = is_set(rsc->flags, pe_rsc_notify); add_hash_param(rsc->meta, XML_RSC_ATTR_UNIQUE, is_set(rsc->flags, pe_rsc_unique) ? XML_BOOLEAN_TRUE : XML_BOOLEAN_FALSE); for (lpc = 0; lpc < clone_data->clone_max; lpc++) { if (create_child_clone(rsc, lpc, data_set) == NULL) { return FALSE; } } if (clone_data->clone_max == 0) { /* create one so that unpack_find_resource() will hook up * any orphans up to the parent correctly */ if (create_child_clone(rsc, -1, data_set) == NULL) { return FALSE; } } pe_rsc_trace(rsc, "Added %d children to resource %s...", clone_data->clone_max, rsc->id); return TRUE; } gboolean clone_active(resource_t * rsc, gboolean all) { GListPtr gIter = rsc->children; for (; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; gboolean child_active = child_rsc->fns->active(child_rsc, all); if (all == FALSE && child_active) { return TRUE; } else if (all && child_active == FALSE) { return FALSE; } } if (all) { return TRUE; } else { return FALSE; } } static void short_print(char *list, const char *prefix, const char *type, long options, void *print_data) { if (list) { if (options & pe_print_html) { status_print("
  • "); } status_print("%s%s: [%s ]", prefix, type, list); if (options & pe_print_html) { status_print("
    • \n"); } else if (options & pe_print_suppres_nl) { /* nothing */ } else if ((options & pe_print_printf) || (options & pe_print_ncurses)) { status_print("\n"); } } } static void clone_print_xml(resource_t * rsc, const char *pre_text, long options, void *print_data) { int is_master_slave = rsc->variant == pe_master ? 1 : 0; char *child_text = crm_concat(pre_text, " ", ' '); GListPtr gIter = rsc->children; status_print("%sid); status_print("multi_state=\"%s\" ", is_master_slave ? "true" : "false"); status_print("unique=\"%s\" ", is_set(rsc->flags, pe_rsc_unique) ? "true" : "false"); status_print("managed=\"%s\" ", is_set(rsc->flags, pe_rsc_managed) ? "true" : "false"); status_print("failed=\"%s\" ", is_set(rsc->flags, pe_rsc_failed) ? "true" : "false"); status_print("failure_ignored=\"%s\" ", is_set(rsc->flags, pe_rsc_failure_ignored) ? "true" : "false"); status_print(">\n"); for (; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; child_rsc->fns->print(child_rsc, child_text, options, print_data); } status_print("%s\n", pre_text); free(child_text); } bool is_set_recursive(resource_t * rsc, long long flag, bool any) { GListPtr gIter; bool all = !any; if(is_set(rsc->flags, flag)) { if(any) { return TRUE; } } else if(all) { return FALSE; } for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { if(is_set_recursive(gIter->data, flag, any)) { if(any) { return TRUE; } } else if(all) { return FALSE; } } if(all) { return TRUE; } return FALSE; } void clone_print(resource_t * rsc, const char *pre_text, long options, void *print_data) { char *list_text = NULL; char *child_text = NULL; char *stopped_list = NULL; const char *type = "Clone"; GListPtr master_list = NULL; GListPtr started_list = NULL; GListPtr gIter = rsc->children; clone_variant_data_t *clone_data = NULL; int active_instances = 0; if (pre_text == NULL) { pre_text = " "; } if (options & pe_print_xml) { clone_print_xml(rsc, pre_text, options, print_data); return; } get_clone_variant_data(clone_data, rsc); child_text = crm_concat(pre_text, " ", ' '); if (rsc->variant == pe_master) { type = "Master/Slave"; } status_print("%s%s Set: %s [%s]%s%s", pre_text ? pre_text : "", type, rsc->id, ID(clone_data->xml_obj_child), is_set(rsc->flags, pe_rsc_unique) ? " (unique)" : "", is_set(rsc->flags, pe_rsc_managed) ? "" : " (unmanaged)"); if (options & pe_print_html) { status_print("\n
      \n"); } else if ((options & pe_print_log) == 0) { status_print("\n"); } for (; gIter != NULL; gIter = gIter->next) { gboolean print_full = FALSE; resource_t *child_rsc = (resource_t *) gIter->data; if (options & pe_print_clone_details) { print_full = TRUE; } if (child_rsc->fns->active(child_rsc, FALSE) == FALSE) { /* Inactive clone */ if (is_set(child_rsc->flags, pe_rsc_orphan)) { continue; } else if (is_set(rsc->flags, pe_rsc_unique)) { print_full = TRUE; } else if (is_not_set(options, pe_print_clone_active)) { stopped_list = add_list_element(stopped_list, child_rsc->id); } } else if (is_set_recursive(child_rsc, pe_rsc_unique, TRUE) || is_set_recursive(child_rsc, pe_rsc_orphan, TRUE) || is_set_recursive(child_rsc, pe_rsc_managed, FALSE) == FALSE || is_set_recursive(child_rsc, pe_rsc_failed, TRUE)) { /* Unique, unmanaged or failed clone */ print_full = TRUE; } else if (is_set(options, pe_print_pending) && child_rsc->pending_task != NULL) { /* In a pending state */ print_full = TRUE; } else if (child_rsc->fns->active(child_rsc, TRUE)) { /* Fully active anonymous clone */ node_t *location = child_rsc->fns->location(child_rsc, NULL, TRUE); if (location) { enum rsc_role_e a_role = child_rsc->fns->state(child_rsc, TRUE); - if (a_role > RSC_ROLE_SLAVE) { + if (location->details->online == FALSE && location->details->unclean) { + print_full = TRUE; + + } else if (a_role > RSC_ROLE_SLAVE) { /* And active on a single node as master */ master_list = g_list_append(master_list, location); } else { /* And active on a single node as started/slave */ started_list = g_list_append(started_list, location); } } else { /* uncolocated group - bleh */ print_full = TRUE; } } else { /* Partially active anonymous clone */ print_full = TRUE; } if (print_full) { if (options & pe_print_html) { status_print("
    • \n"); } child_rsc->fns->print(child_rsc, child_text, options, print_data); if (options & pe_print_html) { status_print("
      • \n"); } } } /* Masters */ master_list = g_list_sort(master_list, sort_node_uname); for (gIter = master_list; gIter; gIter = gIter->next) { node_t *host = gIter->data; list_text = add_list_element(list_text, host->details->uname); active_instances++; } short_print(list_text, child_text, "Masters", options, print_data); g_list_free(master_list); free(list_text); list_text = NULL; /* Started/Slaves */ started_list = g_list_sort(started_list, sort_node_uname); for (gIter = started_list; gIter; gIter = gIter->next) { node_t *host = gIter->data; list_text = add_list_element(list_text, host->details->uname); active_instances++; } short_print(list_text, child_text, rsc->variant == pe_master ? "Slaves" : "Started", options, print_data); g_list_free(started_list); free(list_text); list_text = NULL; if (is_not_set(options, pe_print_clone_active)) { /* Stopped */ if (is_not_set(rsc->flags, pe_rsc_unique) && (clone_data->clone_max > active_instances)) { GListPtr nIter; GListPtr list = g_hash_table_get_values(rsc->allowed_nodes); /* Custom stopped list for non-unique clones */ free(stopped_list); stopped_list = NULL; if (g_list_length(list) == 0) { /* Clusters with symmetrical=false haven't calculated allowed_nodes yet * If we've not probed for them yet, the Stopped list will be empty */ list = g_hash_table_get_values(rsc->known_on); } list = g_list_sort(list, sort_node_uname); for (nIter = list; nIter != NULL; nIter = nIter->next) { node_t *node = (node_t *)nIter->data; if (pe_find_node(rsc->running_on, node->details->uname) == NULL) { stopped_list = add_list_element(stopped_list, node->details->uname); } } g_list_free(list); } short_print(stopped_list, child_text, "Stopped", options, print_data); free(stopped_list); } if (options & pe_print_html) { status_print("
    \n"); } free(child_text); } void clone_free(resource_t * rsc) { GListPtr gIter = rsc->children; clone_variant_data_t *clone_data = NULL; get_clone_variant_data(clone_data, rsc); pe_rsc_trace(rsc, "Freeing %s", rsc->id); for (; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; CRM_ASSERT(child_rsc); pe_rsc_trace(child_rsc, "Freeing child %s", child_rsc->id); free_xml(child_rsc->xml); child_rsc->xml = NULL; /* There could be a saved unexpanded xml */ free_xml(child_rsc->orig_xml); child_rsc->orig_xml = NULL; child_rsc->fns->free(child_rsc); } g_list_free(rsc->children); if (clone_data) { CRM_ASSERT(clone_data->demote_notify == NULL); CRM_ASSERT(clone_data->stop_notify == NULL); CRM_ASSERT(clone_data->start_notify == NULL); CRM_ASSERT(clone_data->promote_notify == NULL); } common_free(rsc); } enum rsc_role_e clone_resource_state(const resource_t * rsc, gboolean current) { enum rsc_role_e clone_role = RSC_ROLE_UNKNOWN; GListPtr gIter = rsc->children; for (; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; enum rsc_role_e a_role = child_rsc->fns->state(child_rsc, current); if (a_role > clone_role) { clone_role = a_role; } } pe_rsc_trace(rsc, "%s role: %s", rsc->id, role2text(clone_role)); return clone_role; } diff --git a/lib/pengine/complex.c b/lib/pengine/complex.c index 3701ec8181..16b59fb437 100644 --- a/lib/pengine/complex.c +++ b/lib/pengine/complex.c @@ -1,816 +1,845 @@ /* * Copyright (C) 2004 Andrew Beekhof * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ #include #include #include #include void populate_hash(xmlNode * nvpair_list, GHashTable * hash, const char **attrs, int attrs_length); resource_object_functions_t resource_class_functions[] = { { native_unpack, native_find_rsc, native_parameter, native_print, native_active, native_resource_state, native_location, native_free}, { group_unpack, native_find_rsc, native_parameter, group_print, group_active, group_resource_state, native_location, group_free}, { clone_unpack, native_find_rsc, native_parameter, clone_print, clone_active, clone_resource_state, native_location, clone_free}, { master_unpack, native_find_rsc, native_parameter, clone_print, clone_active, clone_resource_state, native_location, clone_free} }; enum pe_obj_types get_resource_type(const char *name) { if (safe_str_eq(name, XML_CIB_TAG_RESOURCE)) { return pe_native; } else if (safe_str_eq(name, XML_CIB_TAG_GROUP)) { return pe_group; } else if (safe_str_eq(name, XML_CIB_TAG_INCARNATION)) { return pe_clone; } else if (safe_str_eq(name, XML_CIB_TAG_MASTER)) { return pe_master; } return pe_unknown; } const char * get_resource_typename(enum pe_obj_types type) { switch (type) { case pe_native: return XML_CIB_TAG_RESOURCE; case pe_group: return XML_CIB_TAG_GROUP; case pe_clone: return XML_CIB_TAG_INCARNATION; case pe_master: return XML_CIB_TAG_MASTER; case pe_unknown: return "unknown"; } return ""; } static void dup_attr(gpointer key, gpointer value, gpointer user_data) { add_hash_param(user_data, key, value); } void get_meta_attributes(GHashTable * meta_hash, resource_t * rsc, node_t * node, pe_working_set_t * data_set) { GHashTable *node_hash = NULL; const char *version = crm_element_value(data_set->input, XML_ATTR_CRM_VERSION); if (node) { node_hash = node->details->attrs; } if (rsc->xml) { xmlAttrPtr xIter = NULL; for (xIter = rsc->xml->properties; xIter; xIter = xIter->next) { const char *prop_name = (const char *)xIter->name; const char *prop_value = crm_element_value(rsc->xml, prop_name); add_hash_param(meta_hash, prop_name, prop_value); } } unpack_instance_attributes(data_set->input, rsc->xml, XML_TAG_META_SETS, node_hash, meta_hash, NULL, FALSE, data_set->now); if(version == NULL || compare_version(version, "3.0.9") < 0) { /* populate from the regular attributes until the GUI can create * meta attributes */ unpack_instance_attributes(data_set->input, rsc->xml, XML_TAG_ATTR_SETS, node_hash, meta_hash, NULL, FALSE, data_set->now); } /* set anything else based on the parent */ if (rsc->parent != NULL) { g_hash_table_foreach(rsc->parent->meta, dup_attr, meta_hash); } /* and finally check the defaults */ unpack_instance_attributes(data_set->input, data_set->rsc_defaults, XML_TAG_META_SETS, node_hash, meta_hash, NULL, FALSE, data_set->now); } void get_rsc_attributes(GHashTable * meta_hash, resource_t * rsc, node_t * node, pe_working_set_t * data_set) { GHashTable *node_hash = NULL; if (node) { node_hash = node->details->attrs; } unpack_instance_attributes(data_set->input, rsc->xml, XML_TAG_ATTR_SETS, node_hash, meta_hash, NULL, FALSE, data_set->now); if (rsc->container) { g_hash_table_replace(meta_hash, strdup(CRM_META"_"XML_RSC_ATTR_CONTAINER), strdup(rsc->container->id)); } /* set anything else based on the parent */ if (rsc->parent != NULL) { get_rsc_attributes(meta_hash, rsc->parent, node, data_set); } else { /* and finally check the defaults */ unpack_instance_attributes(data_set->input, data_set->rsc_defaults, XML_TAG_ATTR_SETS, node_hash, meta_hash, NULL, FALSE, data_set->now); } } static char * template_op_key(xmlNode * op) { const char *name = crm_element_value(op, "name"); const char *role = crm_element_value(op, "role"); char *key = NULL; if (role == NULL || crm_str_eq(role, RSC_ROLE_STARTED_S, TRUE) || crm_str_eq(role, RSC_ROLE_SLAVE_S, TRUE)) { role = RSC_ROLE_UNKNOWN_S; } key = crm_concat(name, role, '-'); return key; } static gboolean unpack_template(xmlNode * xml_obj, xmlNode ** expanded_xml, pe_working_set_t * data_set) { xmlNode *cib_resources = NULL; xmlNode *template = NULL; xmlNode *new_xml = NULL; xmlNode *child_xml = NULL; xmlNode *rsc_ops = NULL; xmlNode *template_ops = NULL; const char *template_ref = NULL; const char *clone = NULL; const char *id = NULL; if (xml_obj == NULL) { pe_err("No resource object for template unpacking"); return FALSE; } template_ref = crm_element_value(xml_obj, XML_CIB_TAG_RSC_TEMPLATE); if (template_ref == NULL) { return TRUE; } id = ID(xml_obj); if (id == NULL) { pe_err("'%s' object must have a id", crm_element_name(xml_obj)); return FALSE; } if (crm_str_eq(template_ref, id, TRUE)) { pe_err("The resource object '%s' should not reference itself", id); return FALSE; } cib_resources = get_xpath_object("//"XML_CIB_TAG_RESOURCES, data_set->input, LOG_TRACE); if (cib_resources == NULL) { pe_err("No resources configured"); return FALSE; } template = find_entity(cib_resources, XML_CIB_TAG_RSC_TEMPLATE, template_ref); if (template == NULL) { pe_err("No template named '%s'", template_ref); return FALSE; } new_xml = copy_xml(template); xmlNodeSetName(new_xml, xml_obj->name); crm_xml_replace(new_xml, XML_ATTR_ID, id); clone = crm_element_value(xml_obj, XML_RSC_ATTR_INCARNATION); if(clone) { crm_xml_add(new_xml, XML_RSC_ATTR_INCARNATION, clone); } template_ops = find_xml_node(new_xml, "operations", FALSE); for (child_xml = __xml_first_child(xml_obj); child_xml != NULL; child_xml = __xml_next(child_xml)) { xmlNode *new_child = NULL; new_child = add_node_copy(new_xml, child_xml); if (crm_str_eq((const char *)new_child->name, "operations", TRUE)) { rsc_ops = new_child; } } if (template_ops && rsc_ops) { xmlNode *op = NULL; GHashTable *rsc_ops_hash = g_hash_table_new_full(crm_str_hash, g_str_equal, g_hash_destroy_str, NULL); for (op = __xml_first_child(rsc_ops); op != NULL; op = __xml_next(op)) { char *key = template_op_key(op); g_hash_table_insert(rsc_ops_hash, key, op); } for (op = __xml_first_child(template_ops); op != NULL; op = __xml_next(op)) { char *key = template_op_key(op); if (g_hash_table_lookup(rsc_ops_hash, key) == NULL) { add_node_copy(rsc_ops, op); } free(key); } if (rsc_ops_hash) { g_hash_table_destroy(rsc_ops_hash); } free_xml(template_ops); } /*free_xml(*expanded_xml); */ *expanded_xml = new_xml; /* Disable multi-level templates for now */ /*if(unpack_template(new_xml, expanded_xml, data_set) == FALSE) { free_xml(*expanded_xml); *expanded_xml = NULL; return FALSE; } */ return TRUE; } static gboolean add_template_rsc(xmlNode * xml_obj, pe_working_set_t * data_set) { const char *template_ref = NULL; const char *id = NULL; if (xml_obj == NULL) { pe_err("No resource object for processing resource list of template"); return FALSE; } template_ref = crm_element_value(xml_obj, XML_CIB_TAG_RSC_TEMPLATE); if (template_ref == NULL) { return TRUE; } id = ID(xml_obj); if (id == NULL) { pe_err("'%s' object must have a id", crm_element_name(xml_obj)); return FALSE; } if (crm_str_eq(template_ref, id, TRUE)) { pe_err("The resource object '%s' should not reference itself", id); return FALSE; } if (add_tag_ref(data_set->template_rsc_sets, template_ref, id) == FALSE) { return FALSE; } return TRUE; } +static void +handle_rsc_isolation(resource_t *rsc) +{ + resource_t *top = uber_parent(rsc); + resource_t *iso = rsc; + const char *wrapper = NULL; + const char *value; + + /* check for isolation wrapper mapping if the parent doesn't have one set + * isolation mapping is enabled by default. For safety, we are allowing isolation + * to be disabled by setting the meta attr, isolation=false. */ + value = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_ISOLATION); + if (top->isolation_wrapper == NULL && (value == NULL || crm_is_true(value))) { + if (g_hash_table_lookup(rsc->meta, "pcmk_docker_image")) { + wrapper = "docker-wrapper"; + } + /* add more isolation technologies here as we expand */ + } else if (top->isolation_wrapper) { + goto set_rsc_opts; + } + + if (wrapper == NULL) { + return; + } + + /* if this is a cloned primitive/group, go head and set the isolation wrapper at + * at the clone level. this is really the only sane thing to do in this situation. + * This allows someone to clone an isolated resource without having to shuffle + * around the isolation attributes to the clone parent */ + if (top == rsc->parent && top->variant >= pe_clone) { + iso = top; + } + + iso->isolation_wrapper = wrapper; + set_bit(top->flags, pe_rsc_unique); + +set_rsc_opts: + clear_bit(rsc->flags, pe_rsc_allow_migrate); + set_bit(rsc->flags, pe_rsc_unique); + if (top->variant >= pe_clone) { + add_hash_param(rsc->meta, XML_RSC_ATTR_UNIQUE, XML_BOOLEAN_TRUE); + } +} + gboolean common_unpack(xmlNode * xml_obj, resource_t ** rsc, resource_t * parent, pe_working_set_t * data_set) { bool isdefault = FALSE; xmlNode *expanded_xml = NULL; xmlNode *ops = NULL; resource_t *top = NULL; const char *value = NULL; const char *rclass = NULL; /* Look for this after any templates have been expanded */ const char *id = crm_element_value(xml_obj, XML_ATTR_ID); int container_remote_node = 0; int baremetal_remote_node = 0; crm_log_xml_trace(xml_obj, "Processing resource input..."); if (id == NULL) { pe_err("Must specify id tag in "); return FALSE; } else if (rsc == NULL) { pe_err("Nowhere to unpack resource into"); return FALSE; } if (unpack_template(xml_obj, &expanded_xml, data_set) == FALSE) { return FALSE; } *rsc = calloc(1, sizeof(resource_t)); if (expanded_xml) { crm_log_xml_trace(expanded_xml, "Expanded resource..."); (*rsc)->xml = expanded_xml; (*rsc)->orig_xml = xml_obj; } else { (*rsc)->xml = xml_obj; (*rsc)->orig_xml = NULL; } /* Do not use xml_obj from here on, use (*rsc)->xml in case templates are involved */ rclass = crm_element_value((*rsc)->xml, XML_AGENT_ATTR_CLASS); (*rsc)->parent = parent; ops = find_xml_node((*rsc)->xml, "operations", FALSE); (*rsc)->ops_xml = expand_idref(ops, data_set->input); (*rsc)->variant = get_resource_type(crm_element_name((*rsc)->xml)); if ((*rsc)->variant == pe_unknown) { pe_err("Unknown resource type: %s", crm_element_name((*rsc)->xml)); free(*rsc); return FALSE; } (*rsc)->parameters = g_hash_table_new_full(crm_str_hash, g_str_equal, g_hash_destroy_str, g_hash_destroy_str); (*rsc)->meta = g_hash_table_new_full(crm_str_hash, g_str_equal, g_hash_destroy_str, g_hash_destroy_str); (*rsc)->allowed_nodes = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, g_hash_destroy_str); (*rsc)->known_on = g_hash_table_new_full(crm_str_hash, g_str_equal, NULL, g_hash_destroy_str); value = crm_element_value((*rsc)->xml, XML_RSC_ATTR_INCARNATION); if (value) { (*rsc)->id = crm_concat(id, value, ':'); add_hash_param((*rsc)->meta, XML_RSC_ATTR_INCARNATION, value); } else { (*rsc)->id = strdup(id); } (*rsc)->fns = &resource_class_functions[(*rsc)->variant]; pe_rsc_trace((*rsc), "Unpacking resource..."); get_meta_attributes((*rsc)->meta, *rsc, NULL, data_set); (*rsc)->flags = 0; set_bit((*rsc)->flags, pe_rsc_runnable); set_bit((*rsc)->flags, pe_rsc_provisional); if (is_set(data_set->flags, pe_flag_is_managed_default)) { set_bit((*rsc)->flags, pe_rsc_managed); } (*rsc)->rsc_cons = NULL; (*rsc)->rsc_tickets = NULL; (*rsc)->actions = NULL; (*rsc)->role = RSC_ROLE_STOPPED; (*rsc)->next_role = RSC_ROLE_UNKNOWN; (*rsc)->recovery_type = recovery_stop_start; (*rsc)->stickiness = data_set->default_resource_stickiness; (*rsc)->migration_threshold = INFINITY; (*rsc)->failure_timeout = 0; value = g_hash_table_lookup((*rsc)->meta, XML_CIB_ATTR_PRIORITY); (*rsc)->priority = crm_parse_int(value, "0"); (*rsc)->effective_priority = (*rsc)->priority; value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_NOTIFY); if (crm_is_true(value)) { set_bit((*rsc)->flags, pe_rsc_notify); } if (xml_contains_remote_node((*rsc)->xml)) { if (g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_CONTAINER)) { container_remote_node = 1; } else { baremetal_remote_node = 1; } } value = g_hash_table_lookup((*rsc)->meta, XML_OP_ATTR_ALLOW_MIGRATE); if (crm_is_true(value)) { set_bit((*rsc)->flags, pe_rsc_allow_migrate); } else if (value == NULL && baremetal_remote_node) { /* by default, we want baremetal remote-nodes to be able * to float around the cluster without having to stop all the * resources within the remote-node before moving. Allowing * migration support enables this feature. If this ever causes * problems, migration support can be explicitly turned off with * allow-migrate=false. */ set_bit((*rsc)->flags, pe_rsc_allow_migrate); } value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_MANAGED); if (value != NULL && safe_str_neq("default", value)) { gboolean bool_value = TRUE; crm_str_to_boolean(value, &bool_value); if (bool_value == FALSE) { clear_bit((*rsc)->flags, pe_rsc_managed); } else { set_bit((*rsc)->flags, pe_rsc_managed); } } value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_MAINTENANCE); if (value != NULL && safe_str_neq("default", value)) { gboolean bool_value = FALSE; crm_str_to_boolean(value, &bool_value); if (bool_value == TRUE) { clear_bit((*rsc)->flags, pe_rsc_managed); set_bit((*rsc)->flags, pe_rsc_maintenance); } } else if (is_set(data_set->flags, pe_flag_maintenance_mode)) { clear_bit((*rsc)->flags, pe_rsc_managed); set_bit((*rsc)->flags, pe_rsc_maintenance); } pe_rsc_trace((*rsc), "Options for %s", (*rsc)->id); - top = uber_parent(*rsc); - - /* check for isolation wrapper mapping if the parent doesn't have one set - * isolation mapping is enabled by default. For safety, we are allowing isolation - * to be disabled by setting the meta attr, isolation=false. */ - value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_ISOLATION); - if (top->isolation_wrapper == NULL && (value == NULL || crm_is_true(value))) { - if (g_hash_table_lookup((*rsc)->meta, "pcmk_docker_image")) { - (*rsc)->isolation_wrapper = "docker-wrapper"; - clear_bit((*rsc)->flags, pe_rsc_allow_migrate); - } - /* add more isolation technologies here as we expand */ - } - if (top->isolation_wrapper) { - /* never allow resources with an isolation wrapper migrate */ - clear_bit((*rsc)->flags, pe_rsc_allow_migrate); - } + handle_rsc_isolation(*rsc); + top = uber_parent(*rsc); value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_UNIQUE); - if (crm_is_true(value) || top->variant < pe_clone || (*rsc)->isolation_wrapper) { + if (crm_is_true(value) || top->variant < pe_clone) { set_bit((*rsc)->flags, pe_rsc_unique); } value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_RESTART); if (safe_str_eq(value, "restart")) { (*rsc)->restart_type = pe_restart_restart; pe_rsc_trace((*rsc), "\tDependency restart handling: restart"); } else { (*rsc)->restart_type = pe_restart_ignore; pe_rsc_trace((*rsc), "\tDependency restart handling: ignore"); } value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_MULTIPLE); if (safe_str_eq(value, "stop_only")) { (*rsc)->recovery_type = recovery_stop_only; pe_rsc_trace((*rsc), "\tMultiple running resource recovery: stop only"); } else if (safe_str_eq(value, "block")) { (*rsc)->recovery_type = recovery_block; pe_rsc_trace((*rsc), "\tMultiple running resource recovery: block"); } else { (*rsc)->recovery_type = recovery_stop_start; pe_rsc_trace((*rsc), "\tMultiple running resource recovery: stop/start"); } value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_STICKINESS); if (value != NULL && safe_str_neq("default", value)) { (*rsc)->stickiness = char2score(value); } value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_FAIL_STICKINESS); if (value != NULL && safe_str_neq("default", value)) { (*rsc)->migration_threshold = char2score(value); } else if (value == NULL) { /* Make a best-effort guess at a migration threshold for people with 0.6 configs * try with underscores and hyphens, from both the resource and global defaults section */ value = g_hash_table_lookup((*rsc)->meta, "resource-failure-stickiness"); if (value == NULL) { value = g_hash_table_lookup((*rsc)->meta, "resource_failure_stickiness"); } if (value == NULL) { value = g_hash_table_lookup(data_set->config_hash, "default-resource-failure-stickiness"); } if (value == NULL) { value = g_hash_table_lookup(data_set->config_hash, "default_resource_failure_stickiness"); } if (value) { int fail_sticky = char2score(value); if (fail_sticky == -INFINITY) { (*rsc)->migration_threshold = 1; pe_rsc_info((*rsc), "Set a migration threshold of %d for %s based on a failure-stickiness of %s", (*rsc)->migration_threshold, (*rsc)->id, value); } else if ((*rsc)->stickiness != 0 && fail_sticky != 0) { (*rsc)->migration_threshold = (*rsc)->stickiness / fail_sticky; if ((*rsc)->migration_threshold < 0) { /* Make sure it's positive */ (*rsc)->migration_threshold = 0 - (*rsc)->migration_threshold; } (*rsc)->migration_threshold += 1; pe_rsc_info((*rsc), "Calculated a migration threshold for %s of %d based on a stickiness of %d/%s", (*rsc)->id, (*rsc)->migration_threshold, (*rsc)->stickiness, value); } } } if (safe_str_eq(rclass, "stonith")) { set_bit(data_set->flags, pe_flag_have_stonith_resource); set_bit((*rsc)->flags, pe_rsc_fence_device); } value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_REQUIRES); handle_requires_pref: if (safe_str_eq(value, "nothing")) { } else if (safe_str_eq(value, "quorum")) { set_bit((*rsc)->flags, pe_rsc_needs_quorum); } else if (safe_str_eq(value, "unfencing")) { if (is_set((*rsc)->flags, pe_rsc_fence_device)) { crm_config_warn("%s is a fencing device but requires (un)fencing", (*rsc)->id); value = "quorum"; isdefault = TRUE; goto handle_requires_pref; } else if (is_not_set(data_set->flags, pe_flag_stonith_enabled)) { crm_config_warn("%s requires (un)fencing but fencing is disabled", (*rsc)->id); value = "quorum"; isdefault = TRUE; goto handle_requires_pref; } else { set_bit((*rsc)->flags, pe_rsc_needs_fencing); set_bit((*rsc)->flags, pe_rsc_needs_unfencing); } } else if (safe_str_eq(value, "fencing")) { set_bit((*rsc)->flags, pe_rsc_needs_fencing); if (is_not_set(data_set->flags, pe_flag_stonith_enabled)) { crm_config_warn("%s requires fencing but fencing is disabled", (*rsc)->id); } } else { if (value) { crm_config_err("Invalid value for %s->requires: %s%s", (*rsc)->id, value, is_set(data_set->flags, pe_flag_stonith_enabled) ? "" : " (stonith-enabled=false)"); } isdefault = TRUE; if(is_set((*rsc)->flags, pe_rsc_fence_device)) { value = "quorum"; } else if (is_set(data_set->flags, pe_flag_enable_unfencing)) { value = "unfencing"; } else if (is_set(data_set->flags, pe_flag_stonith_enabled)) { value = "fencing"; } else if (data_set->no_quorum_policy == no_quorum_ignore) { value = "nothing"; } else { value = "quorum"; } goto handle_requires_pref; } pe_rsc_trace((*rsc), "\tRequired to start: %s%s", value, isdefault?" (default)":""); value = g_hash_table_lookup((*rsc)->meta, XML_RSC_ATTR_FAIL_TIMEOUT); if (value != NULL) { /* call crm_get_msec() and convert back to seconds */ (*rsc)->failure_timeout = (crm_get_msec(value) / 1000); } get_target_role(*rsc, &((*rsc)->next_role)); pe_rsc_trace((*rsc), "\tDesired next state: %s", (*rsc)->next_role != RSC_ROLE_UNKNOWN ? role2text((*rsc)->next_role) : "default"); if ((*rsc)->fns->unpack(*rsc, data_set) == FALSE) { return FALSE; } if (is_set(data_set->flags, pe_flag_symmetric_cluster)) { resource_location(*rsc, NULL, 0, "symmetric_default", data_set); } else if (container_remote_node) { /* remote resources tied to a container resource must always be allowed * to opt-in to the cluster. Whether the connection resource is actually * allowed to be placed on a node is dependent on the container resource */ resource_location(*rsc, NULL, 0, "remote_connection_default", data_set); } pe_rsc_trace((*rsc), "\tAction notification: %s", is_set((*rsc)->flags, pe_rsc_notify) ? "required" : "not required"); (*rsc)->utilization = g_hash_table_new_full(crm_str_hash, g_str_equal, g_hash_destroy_str, g_hash_destroy_str); unpack_instance_attributes(data_set->input, (*rsc)->xml, XML_TAG_UTILIZATION, NULL, (*rsc)->utilization, NULL, FALSE, data_set->now); /* data_set->resources = g_list_append(data_set->resources, (*rsc)); */ if (expanded_xml) { if (add_template_rsc(xml_obj, data_set) == FALSE) { return FALSE; } } return TRUE; } void common_update_score(resource_t * rsc, const char *id, int score) { node_t *node = NULL; node = pe_hash_table_lookup(rsc->allowed_nodes, id); if (node != NULL) { pe_rsc_trace(rsc, "Updating score for %s on %s: %d + %d", rsc->id, id, node->weight, score); node->weight = merge_weights(node->weight, score); } if (rsc->children) { GListPtr gIter = rsc->children; for (; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; common_update_score(child_rsc, id, score); } } } gboolean is_parent(resource_t *child, resource_t *rsc) { resource_t *parent = child; if (parent == NULL || rsc == NULL) { return FALSE; } while (parent->parent != NULL) { if (parent->parent == rsc) { return TRUE; } parent = parent->parent; } return FALSE; } resource_t * uber_parent(resource_t * rsc) { resource_t *parent = rsc; if (parent == NULL) { return NULL; } while (parent->parent != NULL) { parent = parent->parent; } return parent; } void common_free(resource_t * rsc) { if (rsc == NULL) { return; } pe_rsc_trace(rsc, "Freeing %s %d", rsc->id, rsc->variant); g_list_free(rsc->rsc_cons); g_list_free(rsc->rsc_cons_lhs); g_list_free(rsc->rsc_tickets); g_list_free(rsc->dangling_migrations); if (rsc->parameters != NULL) { g_hash_table_destroy(rsc->parameters); } if (rsc->meta != NULL) { g_hash_table_destroy(rsc->meta); } if (rsc->utilization != NULL) { g_hash_table_destroy(rsc->utilization); } if (rsc->parent == NULL && is_set(rsc->flags, pe_rsc_orphan)) { free_xml(rsc->xml); rsc->xml = NULL; free_xml(rsc->orig_xml); rsc->orig_xml = NULL; /* if rsc->orig_xml, then rsc->xml is an expanded xml from a template */ } else if (rsc->orig_xml) { free_xml(rsc->xml); rsc->xml = NULL; } if (rsc->running_on) { g_list_free(rsc->running_on); rsc->running_on = NULL; } if (rsc->known_on) { g_hash_table_destroy(rsc->known_on); rsc->known_on = NULL; } if (rsc->actions) { g_list_free(rsc->actions); rsc->actions = NULL; } if (rsc->allowed_nodes) { g_hash_table_destroy(rsc->allowed_nodes); rsc->allowed_nodes = NULL; } g_list_free(rsc->fillers); g_list_free(rsc->rsc_location); pe_rsc_trace(rsc, "Resource freed"); free(rsc->id); free(rsc->clone_name); free(rsc->allocated_to); free(rsc->variant_opaque); free(rsc->pending_task); free(rsc); } diff --git a/lib/pengine/native.c b/lib/pengine/native.c index 5c427e6a58..2489e44193 100644 --- a/lib/pengine/native.c +++ b/lib/pengine/native.c @@ -1,868 +1,872 @@ /* * Copyright (C) 2004 Andrew Beekhof * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ #include #include #include #include #include #include #include #define VARIANT_NATIVE 1 #include "./variant.h" void native_add_running(resource_t * rsc, node_t * node, pe_working_set_t * data_set) { GListPtr gIter = rsc->running_on; CRM_CHECK(node != NULL, return); for (; gIter != NULL; gIter = gIter->next) { node_t *a_node = (node_t *) gIter->data; CRM_CHECK(a_node != NULL, return); if (safe_str_eq(a_node->details->id, node->details->id)) { return; } } pe_rsc_trace(rsc, "Adding %s to %s %s", rsc->id, node->details->uname, is_set(rsc->flags, pe_rsc_managed)?"":"(unmanaged)"); rsc->running_on = g_list_append(rsc->running_on, node); if (rsc->variant == pe_native) { node->details->running_rsc = g_list_append(node->details->running_rsc, rsc); } if (rsc->variant == pe_native && node->details->maintenance) { clear_bit(rsc->flags, pe_rsc_managed); } if (is_not_set(rsc->flags, pe_rsc_managed)) { resource_t *p = rsc->parent; pe_rsc_info(rsc, "resource %s isnt managed", rsc->id); resource_location(rsc, node, INFINITY, "not_managed_default", data_set); while(p && node->details->online) { /* add without the additional location constraint */ p->running_on = g_list_append(p->running_on, node); p = p->parent; } return; } if (rsc->variant == pe_native && g_list_length(rsc->running_on) > 1) { switch (rsc->recovery_type) { case recovery_stop_only: { GHashTableIter gIter; node_t *local_node = NULL; /* make sure it doesnt come up again */ g_hash_table_destroy(rsc->allowed_nodes); rsc->allowed_nodes = node_hash_from_list(data_set->nodes); g_hash_table_iter_init(&gIter, rsc->allowed_nodes); while (g_hash_table_iter_next(&gIter, NULL, (void **)&local_node)) { local_node->weight = -INFINITY; } } break; case recovery_stop_start: break; case recovery_block: clear_bit(rsc->flags, pe_rsc_managed); set_bit(rsc->flags, pe_rsc_block); break; } crm_debug("%s is active on %d nodes including %s: %s", rsc->id, g_list_length(rsc->running_on), node->details->uname, recovery2text(rsc->recovery_type)); } else { pe_rsc_trace(rsc, "Resource %s is active on: %s", rsc->id, node->details->uname); } if (rsc->parent != NULL) { native_add_running(rsc->parent, node, data_set); } } extern void force_non_unique_clone(resource_t * rsc, const char *rid, pe_working_set_t * data_set); gboolean native_unpack(resource_t * rsc, pe_working_set_t * data_set) { resource_t *parent = uber_parent(rsc); native_variant_data_t *native_data = NULL; const char *class = crm_element_value(rsc->xml, XML_AGENT_ATTR_CLASS); pe_rsc_trace(rsc, "Processing resource %s...", rsc->id); native_data = calloc(1, sizeof(native_variant_data_t)); rsc->variant_opaque = native_data; if (is_set(rsc->flags, pe_rsc_unique) && rsc->parent) { if (safe_str_eq(class, "lsb")) { resource_t *top = uber_parent(rsc); force_non_unique_clone(top, rsc->id, data_set); } } if (safe_str_eq(class, "ocf") == FALSE) { const char *stateful = g_hash_table_lookup(parent->meta, "stateful"); if (safe_str_eq(stateful, XML_BOOLEAN_TRUE)) { pe_err ("Resource %s is of type %s and therefore cannot be used as a master/slave resource", rsc->id, class); return FALSE; } } return TRUE; } resource_t * native_find_rsc(resource_t * rsc, const char *id, node_t * on_node, int flags) { gboolean match = FALSE; resource_t *result = NULL; GListPtr gIter = rsc->children; CRM_ASSERT(id != NULL); if (flags & pe_find_clone) { const char *rid = ID(rsc->xml); if (rsc->parent == NULL) { match = FALSE; } else if (safe_str_eq(rsc->id, id)) { match = TRUE; } else if (safe_str_eq(rid, id)) { match = TRUE; } } else { if (strcmp(rsc->id, id) == 0) { match = TRUE; } else if (is_set(flags, pe_find_renamed) && rsc->clone_name && strcmp(rsc->clone_name, id) == 0) { match = TRUE; } } if (match && on_node) { pe_rsc_trace(rsc, "Now checking %s is on %s", rsc->id, on_node->details->uname); if (is_set(flags, pe_find_current) && rsc->running_on) { GListPtr gIter = rsc->running_on; for (; gIter != NULL; gIter = gIter->next) { node_t *loc = (node_t *) gIter->data; if (loc->details == on_node->details) { return rsc; } } } else if (is_set(flags, pe_find_inactive) && rsc->running_on == NULL) { return rsc; } else if (is_not_set(flags, pe_find_current) && rsc->allocated_to && rsc->allocated_to->details == on_node->details) { return rsc; } } else if (match) { return rsc; } for (; gIter != NULL; gIter = gIter->next) { resource_t *child = (resource_t *) gIter->data; result = rsc->fns->find_rsc(child, id, on_node, flags); if (result) { return result; } } return NULL; } char * native_parameter(resource_t * rsc, node_t * node, gboolean create, const char *name, pe_working_set_t * data_set) { char *value_copy = NULL; const char *value = NULL; GHashTable *hash = rsc->parameters; GHashTable *local_hash = NULL; CRM_CHECK(rsc != NULL, return NULL); CRM_CHECK(name != NULL && strlen(name) != 0, return NULL); pe_rsc_trace(rsc, "Looking up %s in %s", name, rsc->id); if (create || g_hash_table_size(rsc->parameters) == 0) { if (node != NULL) { pe_rsc_trace(rsc, "Creating hash with node %s", node->details->uname); } else { pe_rsc_trace(rsc, "Creating default hash"); } local_hash = g_hash_table_new_full(crm_str_hash, g_str_equal, g_hash_destroy_str, g_hash_destroy_str); get_rsc_attributes(local_hash, rsc, node, data_set); hash = local_hash; } value = g_hash_table_lookup(hash, name); if (value == NULL) { /* try meta attributes instead */ value = g_hash_table_lookup(rsc->meta, name); } if (value != NULL) { value_copy = strdup(value); } if (local_hash != NULL) { g_hash_table_destroy(local_hash); } return value_copy; } gboolean native_active(resource_t * rsc, gboolean all) { GListPtr gIter = rsc->running_on; for (; gIter != NULL; gIter = gIter->next) { node_t *a_node = (node_t *) gIter->data; if (a_node->details->unclean) { crm_debug("Resource %s: node %s is unclean", rsc->id, a_node->details->uname); return TRUE; } else if (a_node->details->online == FALSE) { crm_debug("Resource %s: node %s is offline", rsc->id, a_node->details->uname); } else { crm_debug("Resource %s active on %s", rsc->id, a_node->details->uname); return TRUE; } } return FALSE; } struct print_data_s { long options; void *print_data; }; static void native_print_attr(gpointer key, gpointer value, gpointer user_data) { long options = ((struct print_data_s *)user_data)->options; void *print_data = ((struct print_data_s *)user_data)->print_data; status_print("Option: %s = %s\n", (char *)key, (char *)value); } static const char * native_pending_state(resource_t * rsc) { const char *pending_state = NULL; if (safe_str_eq(rsc->pending_task, CRMD_ACTION_START)) { pending_state = "Starting"; } else if (safe_str_eq(rsc->pending_task, CRMD_ACTION_STOP)) { pending_state = "Stopping"; } else if (safe_str_eq(rsc->pending_task, CRMD_ACTION_MIGRATE)) { pending_state = "Migrating"; } else if (safe_str_eq(rsc->pending_task, CRMD_ACTION_MIGRATED)) { /* Work might be done in here. */ pending_state = "Migrating"; } else if (safe_str_eq(rsc->pending_task, CRMD_ACTION_PROMOTE)) { pending_state = "Promoting"; } else if (safe_str_eq(rsc->pending_task, CRMD_ACTION_DEMOTE)) { pending_state = "Demoting"; } return pending_state; } static const char * native_pending_task(resource_t * rsc) { const char *pending_task = NULL; if (safe_str_eq(rsc->pending_task, CRMD_ACTION_NOTIFY)) { /* "Notifying" is not very useful to be shown. */ pending_task = NULL; } else if (safe_str_eq(rsc->pending_task, CRMD_ACTION_STATUS)) { pending_task = "Monitoring"; /* Comment this out until someone requests it */ /* } else if (safe_str_eq(rsc->pending_task, "probe")) { pending_task = "Checking"; */ } return pending_task; } static void native_print_xml(resource_t * rsc, const char *pre_text, long options, void *print_data) { enum rsc_role_e role = rsc->role; const char *class = crm_element_value(rsc->xml, XML_AGENT_ATTR_CLASS); const char *prov = crm_element_value(rsc->xml, XML_AGENT_ATTR_PROVIDER); const char *rsc_state = NULL; if(role == RSC_ROLE_STARTED && uber_parent(rsc)->variant == pe_master) { role = RSC_ROLE_SLAVE; } /* resource information. */ status_print("%sxml, XML_ATTR_TYPE)); if (options & pe_print_pending) { rsc_state = native_pending_state(rsc); } if (rsc_state == NULL) { rsc_state = role2text(role); } status_print("role=\"%s\" ", rsc_state); status_print("active=\"%s\" ", rsc->fns->active(rsc, TRUE) ? "true" : "false"); status_print("orphaned=\"%s\" ", is_set(rsc->flags, pe_rsc_orphan) ? "true" : "false"); status_print("managed=\"%s\" ", is_set(rsc->flags, pe_rsc_managed) ? "true" : "false"); status_print("failed=\"%s\" ", is_set(rsc->flags, pe_rsc_failed) ? "true" : "false"); status_print("failure_ignored=\"%s\" ", is_set(rsc->flags, pe_rsc_failure_ignored) ? "true" : "false"); status_print("nodes_running_on=\"%d\" ", g_list_length(rsc->running_on)); if (options & pe_print_pending) { const char *pending_task = native_pending_task(rsc); if (pending_task) { status_print("pending=\"%s\" ", pending_task); } } if (options & pe_print_dev) { status_print("provisional=\"%s\" ", is_set(rsc->flags, pe_rsc_provisional) ? "true" : "false"); status_print("runnable=\"%s\" ", is_set(rsc->flags, pe_rsc_runnable) ? "true" : "false"); status_print("priority=\"%f\" ", (double)rsc->priority); status_print("variant=\"%s\" ", crm_element_name(rsc->xml)); } /* print out the nodes this resource is running on */ if (options & pe_print_rsconly) { status_print("/>\n"); /* do nothing */ } else if (g_list_length(rsc->running_on) > 0) { GListPtr gIter = rsc->running_on; status_print(">\n"); for (; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; status_print("%s \n", pre_text, node->details->uname, node->details->id, node->details->online ? "false" : "true"); } status_print("%s\n", pre_text); } else { status_print("/>\n"); } } void native_print(resource_t * rsc, const char *pre_text, long options, void *print_data) { node_t *node = NULL; const char *class = crm_element_value(rsc->xml, XML_AGENT_ATTR_CLASS); const char *kind = crm_element_value(rsc->xml, XML_ATTR_TYPE); const char *target_role = NULL; int offset = 0; char buffer[LINE_MAX]; CRM_ASSERT(rsc->variant == pe_native); CRM_ASSERT(kind != NULL); if (rsc->meta) { const char *is_internal = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_INTERNAL_RSC); if (crm_is_true(is_internal)) { crm_trace("skipping print of internal resource %s", rsc->id); return; } target_role = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_TARGET_ROLE); } if (pre_text == NULL && (options & pe_print_printf)) { pre_text = " "; } if (options & pe_print_xml) { native_print_xml(rsc, pre_text, options, print_data); return; } if (rsc->running_on != NULL) { node = rsc->running_on->data; } if ((options & pe_print_rsconly) || g_list_length(rsc->running_on) > 1) { node = NULL; } if (options & pe_print_html) { if (is_not_set(rsc->flags, pe_rsc_managed)) { status_print(""); } else if (is_set(rsc->flags, pe_rsc_failed)) { status_print(""); } else if (rsc->variant == pe_native && g_list_length(rsc->running_on) == 0) { status_print(""); } else if (g_list_length(rsc->running_on) > 1) { status_print(""); } else if (is_set(rsc->flags, pe_rsc_failure_ignored)) { status_print(""); } else { status_print(""); } } if(pre_text) { offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", pre_text); } offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", rsc_printable_id(rsc)); offset += snprintf(buffer + offset, LINE_MAX - offset, "\t(%s", class); if (safe_str_eq(class, "ocf")) { const char *prov = crm_element_value(rsc->xml, XML_AGENT_ATTR_PROVIDER); offset += snprintf(buffer + offset, LINE_MAX - offset, "::%s", prov); } offset += snprintf(buffer + offset, LINE_MAX - offset, ":%s):\t", kind); if(is_set(rsc->flags, pe_rsc_orphan)) { offset += snprintf(buffer + offset, LINE_MAX - offset, " ORPHANED "); } if(rsc->role > RSC_ROLE_SLAVE && is_set(rsc->flags, pe_rsc_failed)) { - offset += snprintf(buffer + offset, LINE_MAX - offset, "FAILED %s ", role2text(rsc->role)); + offset += snprintf(buffer + offset, LINE_MAX - offset, "FAILED %s", role2text(rsc->role)); } else if(is_set(rsc->flags, pe_rsc_failed)) { - offset += snprintf(buffer + offset, LINE_MAX - offset, "FAILED "); + offset += snprintf(buffer + offset, LINE_MAX - offset, "FAILED"); } else { const char *rsc_state = NULL; if (options & pe_print_pending) { rsc_state = native_pending_state(rsc); } if (rsc_state == NULL) { rsc_state = role2text(rsc->role); } if (target_role) { enum rsc_role_e target_role_e = text2role(target_role); /* Ignore target role Started, as it is the default anyways * (and would also allow a Master to be Master). * Show if current role differs from target role, * or if target role limits our abilities. */ if (target_role_e != RSC_ROLE_STARTED && ( target_role_e == RSC_ROLE_SLAVE || target_role_e == RSC_ROLE_STOPPED || safe_str_neq(target_role, rsc_state))) { offset += snprintf(buffer + offset, LINE_MAX - offset, "(target-role:%s) ", target_role); } } - offset += snprintf(buffer + offset, LINE_MAX - offset, "%s ", rsc_state); + offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", rsc_state); } if(node) { - offset += snprintf(buffer + offset, LINE_MAX - offset, "%s ", node->details->uname); + offset += snprintf(buffer + offset, LINE_MAX - offset, " %s", node->details->uname); + + if (node->details->online == FALSE && node->details->unclean) { + offset += snprintf(buffer + offset, LINE_MAX - offset, " (UNCLEAN)"); + } } if (options & pe_print_pending) { const char *pending_task = native_pending_task(rsc); if (pending_task) { - offset += snprintf(buffer + offset, LINE_MAX - offset, "(%s) ", pending_task); + offset += snprintf(buffer + offset, LINE_MAX - offset, " (%s)", pending_task); } } if(is_not_set(rsc->flags, pe_rsc_managed)) { - offset += snprintf(buffer + offset, LINE_MAX - offset, "(unmanaged) "); + offset += snprintf(buffer + offset, LINE_MAX - offset, " (unmanaged)"); } if(is_set(rsc->flags, pe_rsc_failure_ignored)) { - offset += snprintf(buffer + offset, LINE_MAX - offset, "(failure ignored)"); + offset += snprintf(buffer + offset, LINE_MAX - offset, " (failure ignored)"); } if ((options & pe_print_rsconly) || g_list_length(rsc->running_on) > 1) { const char *desc = crm_element_value(rsc->xml, XML_ATTR_DESC); if(desc) { - offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", desc); + offset += snprintf(buffer + offset, LINE_MAX - offset, " %s", desc); } } CRM_LOG_ASSERT(offset > 0); status_print("%s", buffer); #if CURSES_ENABLED if ((options & pe_print_rsconly) || g_list_length(rsc->running_on) > 1) { /* Done */ } else if (options & pe_print_ncurses) { /* coverity[negative_returns] False positive */ move(-1, 0); } #endif if (options & pe_print_html) { status_print(" "); } if ((options & pe_print_rsconly)) { } else if (g_list_length(rsc->running_on) > 1) { GListPtr gIter = rsc->running_on; int counter = 0; if (options & pe_print_html) { status_print("
      \n"); } else if ((options & pe_print_printf) || (options & pe_print_ncurses)) { status_print("["); } for (; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; counter++; if (options & pe_print_html) { status_print("
    • \n%s", node->details->uname); } else if ((options & pe_print_printf) || (options & pe_print_ncurses)) { status_print(" %s", node->details->uname); } else if ((options & pe_print_log)) { status_print("\t%d : %s", counter, node->details->uname); } else { status_print("%s", node->details->uname); } if (options & pe_print_html) { status_print("
      • \n"); } } if (options & pe_print_html) { status_print("
    \n"); } else if ((options & pe_print_printf) || (options & pe_print_ncurses)) { status_print(" ]"); } } if (options & pe_print_html) { status_print("
    \n"); } else if (options & pe_print_suppres_nl) { /* nothing */ } else if ((options & pe_print_printf) || (options & pe_print_ncurses)) { status_print("\n"); } if (options & pe_print_details) { struct print_data_s pdata; pdata.options = options; pdata.print_data = print_data; g_hash_table_foreach(rsc->parameters, native_print_attr, &pdata); } if (options & pe_print_dev) { GHashTableIter iter; node_t *node = NULL; status_print("%s\t(%s%svariant=%s, priority=%f)", pre_text, is_set(rsc->flags, pe_rsc_provisional) ? "provisional, " : "", is_set(rsc->flags, pe_rsc_runnable) ? "" : "non-startable, ", crm_element_name(rsc->xml), (double)rsc->priority); status_print("%s\tAllowed Nodes", pre_text); g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { status_print("%s\t * %s %d", pre_text, node->details->uname, node->weight); } } if (options & pe_print_max_details) { GHashTableIter iter; node_t *node = NULL; status_print("%s\t=== Allowed Nodes\n", pre_text); g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { print_node("\t", node, FALSE); } } } void native_free(resource_t * rsc) { pe_rsc_trace(rsc, "Freeing resource action list (not the data)"); common_free(rsc); } enum rsc_role_e native_resource_state(const resource_t * rsc, gboolean current) { enum rsc_role_e role = rsc->next_role; if (current) { role = rsc->role; } pe_rsc_trace(rsc, "%s state: %s", rsc->id, role2text(role)); return role; } node_t * native_location(resource_t * rsc, GListPtr * list, gboolean current) { node_t *one = NULL; GListPtr result = NULL; if (rsc->children) { GListPtr gIter = rsc->children; for (; gIter != NULL; gIter = gIter->next) { resource_t *child = (resource_t *) gIter->data; child->fns->location(child, &result, current); } } else if (current && rsc->running_on) { result = g_list_copy(rsc->running_on); } else if (current == FALSE && rsc->allocated_to) { result = g_list_append(NULL, rsc->allocated_to); } if (result && g_list_length(result) == 1) { one = g_list_nth_data(result, 0); } if (list) { GListPtr gIter = result; for (; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; if (*list == NULL || pe_find_node_id(*list, node->details->id) == NULL) { *list = g_list_append(*list, node); } } } g_list_free(result); return one; } static void get_rscs_brief(GListPtr rsc_list, GHashTable * rsc_table, GHashTable * active_table) { GListPtr gIter = rsc_list; for (; gIter != NULL; gIter = gIter->next) { resource_t *rsc = (resource_t *) gIter->data; const char *class = crm_element_value(rsc->xml, XML_AGENT_ATTR_CLASS); const char *kind = crm_element_value(rsc->xml, XML_ATTR_TYPE); int offset = 0; char buffer[LINE_MAX]; int *rsc_counter = NULL; int *active_counter = NULL; if (rsc->variant != pe_native) { continue; } offset += snprintf(buffer + offset, LINE_MAX - offset, "%s", class); if (safe_str_eq(class, "ocf")) { const char *prov = crm_element_value(rsc->xml, XML_AGENT_ATTR_PROVIDER); offset += snprintf(buffer + offset, LINE_MAX - offset, "::%s", prov); } offset += snprintf(buffer + offset, LINE_MAX - offset, ":%s", kind); CRM_LOG_ASSERT(offset > 0); if (rsc_table) { rsc_counter = g_hash_table_lookup(rsc_table, buffer); if (rsc_counter == NULL) { rsc_counter = calloc(1, sizeof(int)); *rsc_counter = 0; g_hash_table_insert(rsc_table, strdup(buffer), rsc_counter); } (*rsc_counter)++; } if (active_table) { GListPtr gIter2 = rsc->running_on; for (; gIter2 != NULL; gIter2 = gIter2->next) { node_t *node = (node_t *) gIter2->data; GHashTable *node_table = NULL; if (node->details->unclean == FALSE && node->details->online == FALSE) { continue; } node_table = g_hash_table_lookup(active_table, node->details->uname); if (node_table == NULL) { node_table = g_hash_table_new_full(crm_str_hash, g_str_equal, free, free); g_hash_table_insert(active_table, strdup(node->details->uname), node_table); } active_counter = g_hash_table_lookup(node_table, buffer); if (active_counter == NULL) { active_counter = calloc(1, sizeof(int)); *active_counter = 0; g_hash_table_insert(node_table, strdup(buffer), active_counter); } (*active_counter)++; } } } } static void destroy_node_table(gpointer data) { GHashTable *node_table = data; if (node_table) { g_hash_table_destroy(node_table); } } void print_rscs_brief(GListPtr rsc_list, const char *pre_text, long options, void *print_data, gboolean print_all) { GHashTable *rsc_table = g_hash_table_new_full(crm_str_hash, g_str_equal, free, free); GHashTable *active_table = g_hash_table_new_full(crm_str_hash, g_str_equal, free, destroy_node_table); GHashTableIter hash_iter; char *type = NULL; int *rsc_counter = NULL; get_rscs_brief(rsc_list, rsc_table, active_table); g_hash_table_iter_init(&hash_iter, rsc_table); while (g_hash_table_iter_next(&hash_iter, (gpointer *)&type, (gpointer *)&rsc_counter)) { GHashTableIter hash_iter2; char *node_name = NULL; GHashTable *node_table = NULL; int active_counter_all = 0; g_hash_table_iter_init(&hash_iter2, active_table); while (g_hash_table_iter_next(&hash_iter2, (gpointer *)&node_name, (gpointer *)&node_table)) { int *active_counter = g_hash_table_lookup(node_table, type); if (active_counter == NULL || *active_counter == 0) { continue; } else { active_counter_all += *active_counter; } if (options & pe_print_rsconly) { node_name = NULL; } if (options & pe_print_html) { status_print("
  • \n"); } if (print_all) { status_print("%s%d/%d\t(%s):\tActive %s\n", pre_text ? pre_text : "", active_counter ? *active_counter : 0, rsc_counter ? *rsc_counter : 0, type, active_counter && (*active_counter > 0) && node_name ? node_name : ""); } else { status_print("%s%d\t(%s):\tActive %s\n", pre_text ? pre_text : "", active_counter ? *active_counter : 0, type, active_counter && (*active_counter > 0) && node_name ? node_name : ""); } if (options & pe_print_html) { status_print("
    • \n"); } } if (print_all && active_counter_all == 0) { if (options & pe_print_html) { status_print("
  • \n"); } status_print("%s%d/%d\t(%s):\tActive\n", pre_text ? pre_text : "", active_counter_all, rsc_counter ? *rsc_counter : 0, type); if (options & pe_print_html) { status_print("
    • \n"); } } } if (rsc_table) { g_hash_table_destroy(rsc_table); rsc_table = NULL; } if (active_table) { g_hash_table_destroy(active_table); active_table = NULL; } } diff --git a/lrmd/remote_ctl.c b/lrmd/remote_ctl.c index 32151d744f..c5787239d7 100644 --- a/lrmd/remote_ctl.c +++ b/lrmd/remote_ctl.c @@ -1,526 +1,526 @@ /* * Copyright (c) 2015 David Vossel * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * * This library is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * */ #include #include #include #include #include #include #include #include #include #include extern GHashTable *proxy_table; void lrmd_internal_set_proxy_callback(lrmd_t * lrmd, void *userdata, void (*callback)(lrmd_t *lrmd, void *userdata, xmlNode *msg)); /* *INDENT-OFF* */ static struct crm_option long_options[] = { {"help", 0, 0, '?'}, {"verbose", 0, 0, 'V', "\t\tPrint out logs and events to screen"}, {"quiet", 0, 0, 'Q', "\t\tSuppress all output to screen"}, {"tls", 1, 0, 'S', "\t\tSet tls host to contact"}, {"tls-port", 1, 0, 'p', "\t\tUse custom tls port"}, {"node", 1, 0, 'n', "\tNode name to use for ipc proxy"}, {"api-call", 1, 0, 'c', "\tDirectly relates to lrmd api functions"}, {"-spacer-", 1, 0, '-', "\nParameters for api-call option"}, {"action", 1, 0, 'a'}, {"rsc-id", 1, 0, 'r'}, {"provider", 1, 0, 'P'}, {"class", 1, 0, 'C'}, {"type", 1, 0, 'T'}, {"timeout", 1, 0, 't'}, {"param-key", 1, 0, 'k'}, {"param-val", 1, 0, 'v'}, {"-spacer-", 1, 0, '-'}, {0, 0, 0, 0} }; /* *INDENT-ON* */ static int wait_poke = 0; static int exec_call_id = 0; static gboolean client_start(gpointer user_data); static void try_connect(void); static struct { int verbose; int quiet; int print; int interval; int timeout; int port; const char *node_name; const char *api_call; const char *rsc_id; const char *provider; const char *class; const char *type; const char *action; const char *listen; const char *tls_host; lrmd_key_value_t *params; } options; GMainLoop *mainloop = NULL; lrmd_t *lrmd_conn = NULL; static void client_exit(int rc) { lrmd_api_delete(lrmd_conn); if (proxy_table) { g_hash_table_destroy(proxy_table); proxy_table = NULL; } exit(rc); } static void client_shutdown(int nsig) { lrmd_api_delete(lrmd_conn); lrmd_conn = NULL; } static void read_events(lrmd_event_data_t * event) { if (wait_poke && event->type == lrmd_event_poke) { client_exit(PCMK_OCF_OK); } if ((event->call_id == exec_call_id) && (event->type == lrmd_event_exec_complete)) { if (event->output) { crm_info("%s", event->output); } if (event->exit_reason) { fprintf(stderr, "%s%s\n", PCMK_OCF_REASON_PREFIX, event->exit_reason); } client_exit(event->rc); } } static gboolean timeout_err(gpointer data) { crm_err("timed out in remote_client\n"); client_exit(PCMK_OCF_TIMEOUT); return FALSE; } static void connection_events(lrmd_event_data_t * event) { int rc = event->connection_rc; if (event->type != lrmd_event_connect) { /* ignore */ return; } if (!rc) { client_start(NULL); return; } else { sleep(1); try_connect(); } } static void try_connect(void) { int tries = 10; static int num_tries = 0; int rc = 0; lrmd_conn->cmds->set_callback(lrmd_conn, connection_events); for (; num_tries < tries; num_tries++) { rc = lrmd_conn->cmds->connect_async(lrmd_conn, "lrmd", 10000); if (!rc) { num_tries++; return; /* we'll hear back in async callback */ } sleep(1); } crm_err("Failed to connect to pacemaker remote.\n"); client_exit(PCMK_OCF_UNKNOWN_ERROR); } static gboolean client_start(gpointer user_data) { int rc = 0; if (!lrmd_conn->cmds->is_connected(lrmd_conn)) { try_connect(); /* async connect, this funciton will get called back into. */ return 0; } lrmd_conn->cmds->set_callback(lrmd_conn, read_events); if (safe_str_eq(options.api_call, "ipc_debug")) { /* Do nothing, leave connection up just for debugging ipc proxy */ return 0; } if (options.timeout) { g_timeout_add(options.timeout, timeout_err, NULL); } if (safe_str_eq(options.api_call, "metadata")) { char *output = NULL; rc = lrmd_conn->cmds->get_metadata(lrmd_conn, options.class, options.provider, options.type, &output, 0); if (rc == pcmk_ok) { printf("%s", output); free(output); client_exit(PCMK_OCF_OK); } client_exit(PCMK_OCF_UNKNOWN_ERROR); } else if (safe_str_eq(options.api_call, "poke")) { rc = lrmd_conn->cmds->poke_connection(lrmd_conn); if (rc != pcmk_ok) { client_exit(PCMK_OCF_UNKNOWN_ERROR); } wait_poke = 1; } else { lrmd_rsc_info_t *rsc_info = NULL; rsc_info = lrmd_conn->cmds->get_rsc_info(lrmd_conn, options.rsc_id, 0); if (rsc_info == NULL) { rc = lrmd_conn->cmds->register_rsc(lrmd_conn, options.rsc_id, options.class, options.provider, options.type, 0); if (rc != 0){ crm_err("failed to register resource %s with pacemaker_remote. rc: %d\n", options.rsc_id, rc); client_exit(1); } } lrmd_free_rsc_info(rsc_info); rc = lrmd_conn->cmds->exec(lrmd_conn, options.rsc_id, options.action, NULL, options.interval, options.timeout, 0, 0, options.params); if (rc > 0) { exec_call_id = rc; } else { crm_err("execution of rsc %s failed. rc = %d\n", options.rsc_id, rc); client_exit(PCMK_OCF_UNKNOWN_ERROR); } } return 0; } static int remote_proxy_dispatch_internal(const char *buffer, ssize_t length, gpointer userdata) { /* Async responses from cib and friends back to clients via pacemaker_remoted */ xmlNode *xml = NULL; remote_proxy_t *proxy = userdata; uint32_t flags; xml = string2xml(buffer); if (xml == NULL) { crm_warn("Received a NULL msg from IPC service."); return 1; } flags = crm_ipc_buffer_flags(proxy->ipc); if (flags & crm_ipc_proxied_relay_response) { crm_trace("Passing response back to %.8s on %s: %.200s - request id: %d", proxy->session_id, proxy->node_name, buffer, proxy->last_request_id); remote_proxy_relay_response(lrmd_conn, proxy->session_id, xml, proxy->last_request_id); proxy->last_request_id = 0; } else { crm_trace("Passing event back to %.8s on %s: %.200s", proxy->session_id, proxy->node_name, buffer); remote_proxy_relay_event(lrmd_conn, proxy->session_id, xml); } free_xml(xml); return 1; } static void remote_proxy_disconnected(void *userdata) { remote_proxy_t *proxy = userdata; crm_trace("destroying %p", userdata); proxy->source = NULL; proxy->ipc = NULL; remote_proxy_notify_destroy(lrmd_conn, proxy->session_id); g_hash_table_remove(proxy_table, proxy->session_id); } static remote_proxy_t * remote_proxy_new(const char *node_name, const char *session_id, const char *channel) { static struct ipc_client_callbacks proxy_callbacks = { .dispatch = remote_proxy_dispatch_internal, .destroy = remote_proxy_disconnected }; remote_proxy_t *proxy = calloc(1, sizeof(remote_proxy_t)); proxy->node_name = strdup(node_name); proxy->session_id = strdup(session_id); if (safe_str_eq(channel, CRM_SYSTEM_CRMD)) { proxy->is_local = TRUE; } else { proxy->source = mainloop_add_ipc_client(channel, G_PRIORITY_LOW, 0, proxy, &proxy_callbacks); proxy->ipc = mainloop_get_ipc_client(proxy->source); if (proxy->source == NULL) { remote_proxy_free(proxy); return NULL; } } crm_trace("created proxy session ID %s", proxy->session_id); g_hash_table_insert(proxy_table, proxy->session_id, proxy); return proxy; } static void remote_proxy_cb(lrmd_t *lrmd, void *userdata, xmlNode *msg) { const char *op = crm_element_value(msg, F_LRMD_IPC_OP); const char *session = crm_element_value(msg, F_LRMD_IPC_SESSION); int msg_id = 0; /* sessions are raw ipc connections to IPC, * all we do is proxy requests/responses exactly * like they are given to us at the ipc level. */ CRM_CHECK(op != NULL, return); CRM_CHECK(session != NULL, return); crm_element_value_int(msg, F_LRMD_IPC_MSG_ID, &msg_id); /* This is msg from remote ipc client going to real ipc server */ if (safe_str_eq(op, "new")) { const char *channel = crm_element_value(msg, F_LRMD_IPC_IPC_SERVER); CRM_CHECK(channel != NULL, return); if (remote_proxy_new(options.node_name, session, channel) == NULL) { remote_proxy_notify_destroy(lrmd, session); } crm_info("new remote proxy client established to %s, session id %s", channel, session); } else if (safe_str_eq(op, "destroy")) { remote_proxy_end_session(session); } else if (safe_str_eq(op, "request")) { int flags = 0; xmlNode *request = get_message_xml(msg, F_LRMD_IPC_MSG); const char *name = crm_element_value(msg, F_LRMD_IPC_CLIENT); remote_proxy_t *proxy = g_hash_table_lookup(proxy_table, session); CRM_CHECK(request != NULL, return); if (proxy == NULL) { /* proxy connection no longer exists */ remote_proxy_notify_destroy(lrmd, session); return; } else if ((proxy->is_local == FALSE) && (crm_ipc_connected(proxy->ipc) == FALSE)) { remote_proxy_end_session(session); return; } proxy->last_request_id = 0; crm_element_value_int(msg, F_LRMD_IPC_MSG_FLAGS, &flags); crm_xml_add(request, XML_ACL_TAG_ROLE, "pacemaker-remote"); #if ENABLE_ACL CRM_ASSERT(options.node_name); crm_acl_get_set_user(request, F_LRMD_IPC_USER, options.node_name); #endif if (is_set(flags, crm_ipc_proxied)) { int rc = crm_ipc_send(proxy->ipc, request, flags, 5000, NULL); if(rc < 0) { xmlNode *op_reply = create_xml_node(NULL, "nack"); crm_err("Could not relay %s request %d from %s to %s for %s: %s (%d)", op, msg_id, proxy->node_name, crm_ipc_name(proxy->ipc), name, pcmk_strerror(rc), rc); /* Send a n'ack so the caller doesn't block */ crm_xml_add(op_reply, "function", __FUNCTION__); crm_xml_add_int(op_reply, "line", __LINE__); crm_xml_add_int(op_reply, "rc", rc); remote_proxy_relay_response(lrmd, session, op_reply, msg_id); free_xml(op_reply); } else { crm_trace("Relayed %s request %d from %s to %s for %s", op, msg_id, proxy->node_name, crm_ipc_name(proxy->ipc), name); proxy->last_request_id = msg_id; } } } else { crm_err("Unknown proxy operation: %s", op); } } int main(int argc, char **argv) { int option_index = 0; int argerr = 0; int flag; char *key = NULL; char *val = NULL; gboolean use_tls = FALSE; crm_trigger_t *trig; crm_set_options(NULL, "mode [options]", long_options, "Inject commands into the lrmd and watch for events\n"); while (1) { flag = crm_get_option(argc, argv, &option_index); if (flag == -1) break; switch (flag) { case '?': crm_help(flag, EX_OK); break; case 'V': options.verbose = 1; break; case 'Q': options.quiet = 1; options.verbose = 0; break; case 'n': options.node_name = optarg; break; case 'c': options.api_call = optarg; break; case 'a': options.action = optarg; break; case 'r': options.rsc_id = optarg; break; case 'P': options.provider = optarg; break; case 'C': options.class = optarg; break; case 'T': options.type = optarg; break; case 't': if(optarg) { options.timeout = atoi(optarg); } break; case 'k': key = optarg; if (key && val) { options.params = lrmd_key_value_add(options.params, key, val); key = val = NULL; } break; case 'v': val = optarg; if (key && val) { options.params = lrmd_key_value_add(options.params, key, val); key = val = NULL; } break; case 'S': options.tls_host = optarg; use_tls = TRUE; break; case 'p': if(optarg) { options.port = atoi(optarg); } use_tls = TRUE; break; default: ++argerr; break; } } if (argerr) { crm_help('?', EX_USAGE); } if (optind > argc) { ++argerr; } - crm_log_init("remote_client", LOG_INFO, TRUE, options.verbose ? TRUE : FALSE, argc, argv, FALSE); + crm_log_init("remote_client", LOG_INFO, FALSE, options.verbose ? TRUE : FALSE, argc, argv, FALSE); /* if we can't perform an api_call or listen for events, * there is nothing to do */ if (!options.api_call ) { crm_err("Nothing to be done. Please specify 'api-call'\n"); return PCMK_OCF_UNKNOWN_ERROR; } if (!options.timeout ) { options.timeout = 20000; } if (use_tls) { if (options.node_name == NULL) { crm_err("\"node\" option required when tls is in use.\n"); return PCMK_OCF_UNKNOWN_ERROR; } proxy_table = g_hash_table_new_full(crm_strcase_hash, crm_strcase_equal, NULL, remote_proxy_free); lrmd_conn = lrmd_remote_api_new(NULL, options.tls_host ? options.tls_host : "localhost", options.port); lrmd_internal_set_proxy_callback(lrmd_conn, NULL, remote_proxy_cb); } else { lrmd_conn = lrmd_api_new(); } trig = mainloop_add_trigger(G_PRIORITY_HIGH, client_start, NULL); mainloop_set_trigger(trig); mainloop_add_signal(SIGTERM, client_shutdown); mainloop = g_main_new(FALSE); g_main_run(mainloop); client_exit(0); return 0; } diff --git a/mcp/pacemaker.combined.upstart.in b/mcp/pacemaker.combined.upstart.in index 6301d1068f..a4749139ec 100644 --- a/mcp/pacemaker.combined.upstart.in +++ b/mcp/pacemaker.combined.upstart.in @@ -1,70 +1,73 @@ # pacemaker-corosync - High-Availability cluster # # Starts Corosync cluster engine and Pacemaker cluster manager. +# if you use automatic start, uncomment the line below. +#start on started local and runlevel [2345] + stop on runlevel [0123456] kill timeout 3600 respawn env prog=pacemakerd env rpm_sysconf=@sysconfdir@/sysconfig/pacemaker env rpm_lockfile=@localstatedir@/lock/subsys/pacemaker env deb_sysconf=@sysconfdir@/default/pacemaker env deb_lockfile=@localstatedir@/lock/pacemaker script [ -f "$rpm_sysconf" ] && . $rpm_sysconf [ -f "$deb_sysconf" ] && . $deb_sysconf exec $prog end script pre-start script # setup the software watchdog which corosync uses. # rewrite according to environment. [ -c /dev/watchdog ] || modprobe softdog soft_margin=60 pidof corosync || start corosync # if you use corosync-notifyd, uncomment the line below. #start corosync-notifyd # give it time to fail. sleep 2 pidof corosync || { exit 1; } # if you use crm_mon, uncomment the line below. #start crm_mon end script post-start script [ -f "$rpm_sysconf" ] && . $rpm_sysconf [ -f "$deb_sysconf" ] && . $deb_sysconf [ -z "$LOCK_FILE" -a -d @sysconfdir@/sysconfig ] && LOCK_FILE="$rpm_lockfile" [ -z "$LOCK_FILE" -a -d @sysconfdir@/default ] && LOCK_FILE="$deb_lockfile" touch $LOCK_FILE pidof $prog > @localstatedir@/run/$prog.pid end script post-stop script [ -f "$rpm_sysconf" ] && . $rpm_sysconf [ -f "$deb_sysconf" ] && . $deb_sysconf [ -z "$LOCK_FILE" -a -d @sysconfdir@/sysconfig ] && LOCK_FILE="$rpm_lockfile" [ -z "$LOCK_FILE" -a -d @sysconfdir@/default ] && LOCK_FILE="$deb_lockfile" rm -f $LOCK_FILE rm -f @localstatedir@/run/$prog.pid # if you use watchdog of corosync, uncomment the line below. #pidof corosync || false pidof crmd || stop corosync # if you want to reboot a machine by watchdog of corosync when # pacemakerd disappeared unexpectedly, uncomment the line below # and invalidate above "respawn" stanza. #pidof crmd && killall -q -9 corosync # if you use crm_mon, uncomment the line below. #stop crm_mon # if you use corosync-notifyd, uncomment the line below. #stop corosync-notifyd || true end script diff --git a/pengine/native.c b/pengine/native.c index 7faf692798..05c9d0f922 100644 --- a/pengine/native.c +++ b/pengine/native.c @@ -1,3324 +1,3366 @@ /* * Copyright (C) 2004 Andrew Beekhof * * This program is free software; you can redistribute it and/or * modify it under the terms of the GNU General Public * License as published by the Free Software Foundation; either * version 2 of the License, or (at your option) any later version. * * This software is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * General Public License for more details. * * You should have received a copy of the GNU General Public * License along with this library; if not, write to the Free Software * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ #include #include #include #include #include #include #include /* #define DELETE_THEN_REFRESH 1 // The crmd will remove the resource from the CIB itself, making this redundant */ #define INFINITY_HACK (INFINITY * -100) #define VARIANT_NATIVE 1 #include gboolean update_action(action_t * then); void native_rsc_colocation_rh_must(resource_t * rsc_lh, gboolean update_lh, resource_t * rsc_rh, gboolean update_rh); void native_rsc_colocation_rh_mustnot(resource_t * rsc_lh, gboolean update_lh, resource_t * rsc_rh, gboolean update_rh); void Recurring(resource_t * rsc, action_t * start, node_t * node, pe_working_set_t * data_set); void RecurringOp(resource_t * rsc, action_t * start, node_t * node, xmlNode * operation, pe_working_set_t * data_set); void Recurring_Stopped(resource_t * rsc, action_t * start, node_t * node, pe_working_set_t * data_set); void RecurringOp_Stopped(resource_t * rsc, action_t * start, node_t * node, xmlNode * operation, pe_working_set_t * data_set); void pe_post_notify(resource_t * rsc, node_t * node, action_t * op, notify_data_t * n_data, pe_working_set_t * data_set); gboolean DeleteRsc(resource_t * rsc, node_t * node, gboolean optional, pe_working_set_t * data_set); gboolean StopRsc(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set); gboolean StartRsc(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set); gboolean DemoteRsc(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set); gboolean PromoteRsc(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set); gboolean RoleError(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set); gboolean NullOp(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set); /* *INDENT-OFF* */ enum rsc_role_e rsc_state_matrix[RSC_ROLE_MAX][RSC_ROLE_MAX] = { /* Current State */ /* Next State: Unknown Stopped Started Slave Master */ /* Unknown */ { RSC_ROLE_UNKNOWN, RSC_ROLE_STOPPED, RSC_ROLE_STOPPED, RSC_ROLE_STOPPED, RSC_ROLE_STOPPED, }, /* Stopped */ { RSC_ROLE_STOPPED, RSC_ROLE_STOPPED, RSC_ROLE_STARTED, RSC_ROLE_SLAVE, RSC_ROLE_SLAVE, }, /* Started */ { RSC_ROLE_STOPPED, RSC_ROLE_STOPPED, RSC_ROLE_STARTED, RSC_ROLE_SLAVE, RSC_ROLE_MASTER, }, /* Slave */ { RSC_ROLE_STOPPED, RSC_ROLE_STOPPED, RSC_ROLE_STOPPED, RSC_ROLE_SLAVE, RSC_ROLE_MASTER, }, /* Master */ { RSC_ROLE_STOPPED, RSC_ROLE_SLAVE, RSC_ROLE_SLAVE, RSC_ROLE_SLAVE, RSC_ROLE_MASTER, }, }; gboolean (*rsc_action_matrix[RSC_ROLE_MAX][RSC_ROLE_MAX])(resource_t*,node_t*,gboolean,pe_working_set_t*) = { /* Current State */ /* Next State: Unknown Stopped Started Slave Master */ /* Unknown */ { RoleError, StopRsc, RoleError, RoleError, RoleError, }, /* Stopped */ { RoleError, NullOp, StartRsc, StartRsc, RoleError, }, /* Started */ { RoleError, StopRsc, NullOp, NullOp, PromoteRsc, }, /* Slave */ { RoleError, StopRsc, StopRsc, NullOp, PromoteRsc, }, /* Master */ { RoleError, DemoteRsc, DemoteRsc, DemoteRsc, NullOp, }, }; /* *INDENT-ON* */ struct capacity_data { node_t *node; resource_t *rsc; gboolean is_enough; }; static action_t * get_first_named_action(resource_t * rsc, const char *action, gboolean only_valid, node_t * current); static void check_capacity(gpointer key, gpointer value, gpointer user_data) { int required = 0; int remaining = 0; struct capacity_data *data = user_data; required = crm_parse_int(value, "0"); remaining = crm_parse_int(g_hash_table_lookup(data->node->details->utilization, key), "0"); if (required > remaining) { CRM_ASSERT(data->rsc); CRM_ASSERT(data->node); pe_rsc_debug(data->rsc, "Node %s has no enough %s for resource %s: required=%d remaining=%d", data->node->details->uname, (char *)key, data->rsc->id, required, remaining); data->is_enough = FALSE; } } static gboolean have_enough_capacity(node_t * node, resource_t * rsc) { struct capacity_data data; data.node = node; data.rsc = rsc; data.is_enough = TRUE; g_hash_table_foreach(rsc->utilization, check_capacity, &data); return data.is_enough; } static gboolean native_choose_node(resource_t * rsc, node_t * prefer, pe_working_set_t * data_set) { /* 1. Sort by weight 2. color.chosen_node = the node (of those with the highest wieght) with the fewest resources 3. remove color.chosen_node from all other colors */ int alloc_details = scores_log_level + 1; GListPtr nodes = NULL; node_t *chosen = NULL; int lpc = 0; int multiple = 0; int length = 0; gboolean result = FALSE; if (safe_str_neq(data_set->placement_strategy, "default")) { GListPtr gIter = NULL; for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; if (have_enough_capacity(node, rsc) == FALSE) { pe_rsc_debug(rsc, "Resource %s cannot be allocated to node %s: none of enough capacity", rsc->id, node->details->uname); resource_location(rsc, node, -INFINITY, "__limit_utilization_", data_set); } } dump_node_scores(alloc_details, rsc, "Post-utilization", rsc->allowed_nodes); } length = g_hash_table_size(rsc->allowed_nodes); if (is_not_set(rsc->flags, pe_rsc_provisional)) { return rsc->allocated_to ? TRUE : FALSE; } if (prefer) { chosen = g_hash_table_lookup(rsc->allowed_nodes, prefer->details->id); if (chosen && chosen->weight >= 0 && can_run_resources(chosen)) { pe_rsc_trace(rsc, "Using preferred node %s for %s instead of choosing from %d candidates", chosen->details->uname, rsc->id, length); } else if (chosen && chosen->weight < 0) { pe_rsc_trace(rsc, "Preferred node %s for %s was unavailable", chosen->details->uname, rsc->id); chosen = NULL; } else if (chosen && can_run_resources(chosen)) { pe_rsc_trace(rsc, "Preferred node %s for %s was unsuitable", chosen->details->uname, rsc->id); chosen = NULL; } else { pe_rsc_trace(rsc, "Preferred node %s for %s was unknown", prefer->details->uname, rsc->id); } } if (chosen == NULL && rsc->allowed_nodes) { nodes = g_hash_table_get_values(rsc->allowed_nodes); nodes = g_list_sort_with_data(nodes, sort_node_weight, g_list_nth_data(rsc->running_on, 0)); chosen = g_list_nth_data(nodes, 0); pe_rsc_trace(rsc, "Chose node %s for %s from %d candidates", chosen ? chosen->details->uname : "", rsc->id, length); if (chosen && chosen->weight > 0 && can_run_resources(chosen)) { node_t *running = g_list_nth_data(rsc->running_on, 0); if (running && can_run_resources(running) == FALSE) { pe_rsc_trace(rsc, "Current node for %s (%s) can't run resources", rsc->id, running->details->uname); running = NULL; } for (lpc = 1; lpc < length && running; lpc++) { node_t *tmp = g_list_nth_data(nodes, lpc); if (tmp->weight == chosen->weight) { multiple++; if (tmp->details == running->details) { /* prefer the existing node if scores are equal */ chosen = tmp; } } } } } if (multiple > 1) { int log_level = LOG_INFO; static char score[33]; score2char_stack(chosen->weight, score, sizeof(score)); if (chosen->weight >= INFINITY) { log_level = LOG_WARNING; } do_crm_log(log_level, "%d nodes with equal score (%s) for" " running %s resources. Chose %s.", multiple, score, rsc->id, chosen->details->uname); } result = native_assign_node(rsc, nodes, chosen, FALSE); g_list_free(nodes); return result; } static int node_list_attr_score(GHashTable * list, const char *attr, const char *value) { GHashTableIter iter; node_t *node = NULL; int best_score = -INFINITY; const char *best_node = NULL; if (attr == NULL) { attr = "#" XML_ATTR_UNAME; } g_hash_table_iter_init(&iter, list); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { int weight = node->weight; if (can_run_resources(node) == FALSE) { weight = -INFINITY; } if (weight > best_score || best_node == NULL) { const char *tmp = g_hash_table_lookup(node->details->attrs, attr); if (safe_str_eq(value, tmp)) { best_score = weight; best_node = node->details->uname; } } } if (safe_str_neq(attr, "#" XML_ATTR_UNAME)) { crm_info("Best score for %s=%s was %s with %d", attr, value, best_node ? best_node : "", best_score); } return best_score; } static void node_hash_update(GHashTable * list1, GHashTable * list2, const char *attr, float factor, gboolean only_positive) { int score = 0; int new_score = 0; GHashTableIter iter; node_t *node = NULL; if (attr == NULL) { attr = "#" XML_ATTR_UNAME; } g_hash_table_iter_init(&iter, list1); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { CRM_LOG_ASSERT(node != NULL); if(node == NULL) { continue; }; score = node_list_attr_score(list2, attr, g_hash_table_lookup(node->details->attrs, attr)); new_score = merge_weights(factor * score, node->weight); if (factor < 0 && score < 0) { /* Negative preference for a node with a negative score * should not become a positive preference * * TODO - Decide if we want to filter only if weight == -INFINITY * */ crm_trace("%s: Filtering %d + %f*%d (factor * score)", node->details->uname, node->weight, factor, score); } else if (node->weight == INFINITY_HACK) { crm_trace("%s: Filtering %d + %f*%d (node < 0)", node->details->uname, node->weight, factor, score); } else if (only_positive && new_score < 0 && node->weight > 0) { node->weight = INFINITY_HACK; crm_trace("%s: Filtering %d + %f*%d (score > 0)", node->details->uname, node->weight, factor, score); } else if (only_positive && new_score < 0 && node->weight == 0) { crm_trace("%s: Filtering %d + %f*%d (score == 0)", node->details->uname, node->weight, factor, score); } else { crm_trace("%s: %d + %f*%d", node->details->uname, node->weight, factor, score); node->weight = new_score; } } } GHashTable * node_hash_dup(GHashTable * hash) { /* Hack! */ GListPtr list = g_hash_table_get_values(hash); GHashTable *result = node_hash_from_list(list); g_list_free(list); return result; } GHashTable * native_merge_weights(resource_t * rsc, const char *rhs, GHashTable * nodes, const char *attr, float factor, enum pe_weights flags) { return rsc_merge_weights(rsc, rhs, nodes, attr, factor, flags); } GHashTable * rsc_merge_weights(resource_t * rsc, const char *rhs, GHashTable * nodes, const char *attr, float factor, enum pe_weights flags) { GHashTable *work = NULL; int multiplier = 1; if (factor < 0) { multiplier = -1; } if (is_set(rsc->flags, pe_rsc_merging)) { pe_rsc_info(rsc, "%s: Breaking dependency loop at %s", rhs, rsc->id); return nodes; } set_bit(rsc->flags, pe_rsc_merging); if (is_set(flags, pe_weights_init)) { if (rsc->variant == pe_group && rsc->children) { GListPtr last = rsc->children; while (last->next != NULL) { last = last->next; } pe_rsc_trace(rsc, "Merging %s as a group %p %p", rsc->id, rsc->children, last); work = rsc_merge_weights(last->data, rhs, NULL, attr, factor, flags); } else { work = node_hash_dup(rsc->allowed_nodes); } clear_bit(flags, pe_weights_init); } else if (rsc->variant == pe_group && rsc->children) { GListPtr iter = rsc->children; pe_rsc_trace(rsc, "%s: Combining scores from %d children of %s", rhs, g_list_length(iter), rsc->id); work = node_hash_dup(nodes); for(iter = rsc->children; iter->next != NULL; iter = iter->next) { work = rsc_merge_weights(iter->data, rhs, work, attr, factor, flags); } } else { pe_rsc_trace(rsc, "%s: Combining scores from %s", rhs, rsc->id); work = node_hash_dup(nodes); node_hash_update(work, rsc->allowed_nodes, attr, factor, is_set(flags, pe_weights_positive)); } if (is_set(flags, pe_weights_rollback) && can_run_any(work) == FALSE) { pe_rsc_info(rsc, "%s: Rolling back scores from %s", rhs, rsc->id); g_hash_table_destroy(work); clear_bit(rsc->flags, pe_rsc_merging); return nodes; } if (can_run_any(work)) { GListPtr gIter = NULL; if (is_set(flags, pe_weights_forward)) { gIter = rsc->rsc_cons; crm_trace("Checking %d additional colocation constraints", g_list_length(gIter)); } else if(rsc->variant == pe_group && rsc->children) { GListPtr last = rsc->children; while (last->next != NULL) { last = last->next; } gIter = ((resource_t*)last->data)->rsc_cons_lhs; crm_trace("Checking %d additional optional group colocation constraints from %s", g_list_length(gIter), ((resource_t*)last->data)->id); } else { gIter = rsc->rsc_cons_lhs; crm_trace("Checking %d additional optional colocation constraints %s", g_list_length(gIter), rsc->id); } for (; gIter != NULL; gIter = gIter->next) { resource_t *other = NULL; rsc_colocation_t *constraint = (rsc_colocation_t *) gIter->data; if (is_set(flags, pe_weights_forward)) { other = constraint->rsc_rh; } else { other = constraint->rsc_lh; } pe_rsc_trace(rsc, "Applying %s (%s)", constraint->id, other->id); work = rsc_merge_weights(other, rhs, work, constraint->node_attribute, multiplier * (float)constraint->score / INFINITY, flags|pe_weights_rollback); dump_node_scores(LOG_TRACE, NULL, rhs, work); } } if (is_set(flags, pe_weights_positive)) { node_t *node = NULL; GHashTableIter iter; g_hash_table_iter_init(&iter, work); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { if (node->weight == INFINITY_HACK) { node->weight = 1; } } } if (nodes) { g_hash_table_destroy(nodes); } clear_bit(rsc->flags, pe_rsc_merging); return work; } node_t * native_color(resource_t * rsc, node_t * prefer, pe_working_set_t * data_set) { GListPtr gIter = NULL; int alloc_details = scores_log_level + 1; if (rsc->parent && is_not_set(rsc->parent->flags, pe_rsc_allocating)) { /* never allocate children on their own */ pe_rsc_debug(rsc, "Escalating allocation of %s to its parent: %s", rsc->id, rsc->parent->id); rsc->parent->cmds->allocate(rsc->parent, prefer, data_set); } if (is_not_set(rsc->flags, pe_rsc_provisional)) { return rsc->allocated_to; } if (is_set(rsc->flags, pe_rsc_allocating)) { pe_rsc_debug(rsc, "Dependency loop detected involving %s", rsc->id); return NULL; } set_bit(rsc->flags, pe_rsc_allocating); print_resource(alloc_details, "Allocating: ", rsc, FALSE); dump_node_scores(alloc_details, rsc, "Pre-allloc", rsc->allowed_nodes); for (gIter = rsc->rsc_cons; gIter != NULL; gIter = gIter->next) { rsc_colocation_t *constraint = (rsc_colocation_t *) gIter->data; GHashTable *archive = NULL; resource_t *rsc_rh = constraint->rsc_rh; pe_rsc_trace(rsc, "%s: Pre-Processing %s (%s, %d, %s)", rsc->id, constraint->id, rsc_rh->id, constraint->score, role2text(constraint->role_lh)); if (constraint->role_lh >= RSC_ROLE_MASTER || (constraint->score < 0 && constraint->score > -INFINITY)) { archive = node_hash_dup(rsc->allowed_nodes); } rsc_rh->cmds->allocate(rsc_rh, NULL, data_set); rsc->cmds->rsc_colocation_lh(rsc, rsc_rh, constraint); if (archive && can_run_any(rsc->allowed_nodes) == FALSE) { pe_rsc_info(rsc, "%s: Rolling back scores from %s", rsc->id, rsc_rh->id); g_hash_table_destroy(rsc->allowed_nodes); rsc->allowed_nodes = archive; archive = NULL; } if (archive) { g_hash_table_destroy(archive); } } dump_node_scores(alloc_details, rsc, "Post-coloc", rsc->allowed_nodes); for (gIter = rsc->rsc_cons_lhs; gIter != NULL; gIter = gIter->next) { rsc_colocation_t *constraint = (rsc_colocation_t *) gIter->data; rsc->allowed_nodes = constraint->rsc_lh->cmds->merge_weights(constraint->rsc_lh, rsc->id, rsc->allowed_nodes, constraint->node_attribute, (float)constraint->score / INFINITY, pe_weights_rollback); } print_resource(LOG_DEBUG_2, "Allocating: ", rsc, FALSE); if (rsc->next_role == RSC_ROLE_STOPPED) { pe_rsc_trace(rsc, "Making sure %s doesn't get allocated", rsc->id); /* make sure it doesnt come up again */ resource_location(rsc, NULL, -INFINITY, XML_RSC_ATTR_TARGET_ROLE, data_set); } else if(rsc->next_role > rsc->role && is_set(data_set->flags, pe_flag_have_quorum) == FALSE && data_set->no_quorum_policy == no_quorum_freeze) { crm_notice("Resource %s cannot be elevated from %s to %s: no-quorum-policy=freeze", rsc->id, role2text(rsc->role), role2text(rsc->next_role)); rsc->next_role = rsc->role; } dump_node_scores(show_scores ? 0 : scores_log_level, rsc, __FUNCTION__, rsc->allowed_nodes); if (is_set(data_set->flags, pe_flag_stonith_enabled) && is_set(data_set->flags, pe_flag_have_stonith_resource) == FALSE) { clear_bit(rsc->flags, pe_rsc_managed); } if (is_not_set(rsc->flags, pe_rsc_managed)) { const char *reason = NULL; node_t *assign_to = NULL; rsc->next_role = rsc->role; if (rsc->running_on == NULL) { reason = "inactive"; } else if (rsc->role == RSC_ROLE_MASTER) { assign_to = rsc->running_on->data; reason = "master"; } else if (is_set(rsc->flags, pe_rsc_failed)) { assign_to = rsc->running_on->data; reason = "failed"; } else { assign_to = rsc->running_on->data; reason = "active"; } pe_rsc_info(rsc, "Unmanaged resource %s allocated to %s: %s", rsc->id, assign_to ? assign_to->details->uname : "'nowhere'", reason); native_assign_node(rsc, NULL, assign_to, TRUE); } else if (is_set(data_set->flags, pe_flag_stop_everything)) { pe_rsc_debug(rsc, "Forcing %s to stop", rsc->id); native_assign_node(rsc, NULL, NULL, TRUE); } else if (is_set(rsc->flags, pe_rsc_provisional) && native_choose_node(rsc, prefer, data_set)) { pe_rsc_trace(rsc, "Allocated resource %s to %s", rsc->id, rsc->allocated_to->details->uname); } else if (rsc->allocated_to == NULL) { if (is_not_set(rsc->flags, pe_rsc_orphan)) { pe_rsc_info(rsc, "Resource %s cannot run anywhere", rsc->id); } else if (rsc->running_on != NULL) { pe_rsc_info(rsc, "Stopping orphan resource %s", rsc->id); } } else { pe_rsc_debug(rsc, "Pre-Allocated resource %s to %s", rsc->id, rsc->allocated_to->details->uname); } clear_bit(rsc->flags, pe_rsc_allocating); print_resource(LOG_DEBUG_3, "Allocated ", rsc, TRUE); if (rsc->is_remote_node) { node_t *remote_node = pe_find_node(data_set->nodes, rsc->id); CRM_ASSERT(remote_node != NULL); if (rsc->allocated_to && rsc->next_role != RSC_ROLE_STOPPED) { crm_trace("Setting remote node %s to ONLINE", remote_node->details->id); remote_node->details->online = TRUE; /* We shouldn't consider an unseen remote-node unclean if we are going * to try and connect to it. Otherwise we get an unnecessary fence */ if (remote_node->details->unseen == TRUE) { remote_node->details->unclean = FALSE; } } else { crm_trace("Setting remote node %s to SHUTDOWN. next role = %s, allocated=%s", remote_node->details->id, role2text(rsc->next_role), rsc->allocated_to ? "true" : "false"); remote_node->details->shutdown = TRUE; } } return rsc->allocated_to; } static gboolean is_op_dup(resource_t * rsc, const char *name, const char *interval) { gboolean dup = FALSE; const char *id = NULL; const char *value = NULL; xmlNode *operation = NULL; CRM_ASSERT(rsc); for (operation = __xml_first_child(rsc->ops_xml); operation != NULL; operation = __xml_next(operation)) { if (crm_str_eq((const char *)operation->name, "op", TRUE)) { value = crm_element_value(operation, "name"); if (safe_str_neq(value, name)) { continue; } value = crm_element_value(operation, XML_LRM_ATTR_INTERVAL); if (value == NULL) { value = "0"; } if (safe_str_neq(value, interval)) { continue; } if (id == NULL) { id = ID(operation); } else { crm_config_err("Operation %s is a duplicate of %s", ID(operation), id); crm_config_err ("Do not use the same (name, interval) combination more than once per resource"); dup = TRUE; } } } return dup; } void RecurringOp(resource_t * rsc, action_t * start, node_t * node, xmlNode * operation, pe_working_set_t * data_set) { char *key = NULL; const char *name = NULL; const char *value = NULL; const char *interval = NULL; const char *node_uname = NULL; unsigned long long interval_ms = 0; action_t *mon = NULL; gboolean is_optional = TRUE; GListPtr possible_matches = NULL; /* Only process for the operations without role="Stopped" */ value = crm_element_value(operation, "role"); if (value && text2role(value) == RSC_ROLE_STOPPED) { return; } CRM_ASSERT(rsc); pe_rsc_trace(rsc, "Creating recurring action %s for %s in role %s on %s", ID(operation), rsc->id, role2text(rsc->next_role), node ? node->details->uname : "n/a"); if (node != NULL) { node_uname = node->details->uname; } interval = crm_element_value(operation, XML_LRM_ATTR_INTERVAL); interval_ms = crm_get_interval(interval); if (interval_ms == 0) { return; } name = crm_element_value(operation, "name"); if (is_op_dup(rsc, name, interval)) { return; } if (safe_str_eq(name, RSC_STOP) || safe_str_eq(name, RSC_START) || safe_str_eq(name, RSC_DEMOTE) || safe_str_eq(name, RSC_PROMOTE) ) { crm_config_err("Invalid recurring action %s wth name: '%s'", ID(operation), name); return; } key = generate_op_key(rsc->id, name, interval_ms); if (find_rsc_op_entry(rsc, key) == NULL) { /* disabled */ free(key); return; } if (start != NULL) { pe_rsc_trace(rsc, "Marking %s %s due to %s", key, is_set(start->flags, pe_action_optional) ? "optional" : "manditory", start->uuid); is_optional = (rsc->cmds->action_flags(start, NULL) & pe_action_optional); } else { pe_rsc_trace(rsc, "Marking %s optional", key); is_optional = TRUE; } /* start a monitor for an already active resource */ possible_matches = find_actions_exact(rsc->actions, key, node); if (possible_matches == NULL) { is_optional = FALSE; pe_rsc_trace(rsc, "Marking %s manditory: not active", key); } else { GListPtr gIter = NULL; for (gIter = possible_matches; gIter != NULL; gIter = gIter->next) { action_t *op = (action_t *) gIter->data; if (is_set(op->flags, pe_action_reschedule)) { is_optional = FALSE; break; } } g_list_free(possible_matches); } if ((rsc->next_role == RSC_ROLE_MASTER && value == NULL) || (value != NULL && text2role(value) != rsc->next_role)) { int log_level = LOG_DEBUG_2; const char *result = "Ignoring"; if (is_optional) { char *local_key = strdup(key); log_level = LOG_INFO; result = "Cancelling"; /* its running : cancel it */ mon = custom_action(rsc, local_key, RSC_CANCEL, node, FALSE, TRUE, data_set); free(mon->task); free(mon->cancel_task); mon->task = strdup(RSC_CANCEL); mon->cancel_task = strdup(name); add_hash_param(mon->meta, XML_LRM_ATTR_INTERVAL, interval); add_hash_param(mon->meta, XML_LRM_ATTR_TASK, name); local_key = NULL; switch (rsc->role) { case RSC_ROLE_SLAVE: case RSC_ROLE_STARTED: if (rsc->next_role == RSC_ROLE_MASTER) { local_key = promote_key(rsc); } else if (rsc->next_role == RSC_ROLE_STOPPED) { local_key = stop_key(rsc); } break; case RSC_ROLE_MASTER: local_key = demote_key(rsc); break; default: break; } if (local_key) { custom_action_order(rsc, NULL, mon, rsc, local_key, NULL, pe_order_runnable_left, data_set); } mon = NULL; } do_crm_log(log_level, "%s action %s (%s vs. %s)", result, key, value ? value : role2text(RSC_ROLE_SLAVE), role2text(rsc->next_role)); free(key); key = NULL; return; } mon = custom_action(rsc, key, name, node, is_optional, TRUE, data_set); key = mon->uuid; if (is_optional) { pe_rsc_trace(rsc, "%s\t %s (optional)", crm_str(node_uname), mon->uuid); } if (start == NULL || is_set(start->flags, pe_action_runnable) == FALSE) { pe_rsc_debug(rsc, "%s\t %s (cancelled : start un-runnable)", crm_str(node_uname), mon->uuid); update_action_flags(mon, pe_action_runnable | pe_action_clear); } else if (node == NULL || node->details->online == FALSE || node->details->unclean) { pe_rsc_debug(rsc, "%s\t %s (cancelled : no node available)", crm_str(node_uname), mon->uuid); update_action_flags(mon, pe_action_runnable | pe_action_clear); } else if (is_set(mon->flags, pe_action_optional) == FALSE) { pe_rsc_info(rsc, " Start recurring %s (%llus) for %s on %s", mon->task, interval_ms / 1000, rsc->id, crm_str(node_uname)); } if (rsc->next_role == RSC_ROLE_MASTER) { char *running_master = crm_itoa(PCMK_OCF_RUNNING_MASTER); add_hash_param(mon->meta, XML_ATTR_TE_TARGET_RC, running_master); free(running_master); } if (node == NULL || is_set(rsc->flags, pe_rsc_managed)) { custom_action_order(rsc, start_key(rsc), NULL, NULL, strdup(key), mon, pe_order_implies_then | pe_order_runnable_left, data_set); if (rsc->next_role == RSC_ROLE_MASTER) { custom_action_order(rsc, promote_key(rsc), NULL, rsc, NULL, mon, pe_order_optional | pe_order_runnable_left, data_set); } else if (rsc->role == RSC_ROLE_MASTER) { custom_action_order(rsc, demote_key(rsc), NULL, rsc, NULL, mon, pe_order_optional | pe_order_runnable_left, data_set); } } } void Recurring(resource_t * rsc, action_t * start, node_t * node, pe_working_set_t * data_set) { if (is_not_set(rsc->flags, pe_rsc_maintenance) && (node == NULL || node->details->maintenance == FALSE)) { xmlNode *operation = NULL; for (operation = __xml_first_child(rsc->ops_xml); operation != NULL; operation = __xml_next(operation)) { if (crm_str_eq((const char *)operation->name, "op", TRUE)) { RecurringOp(rsc, start, node, operation, data_set); } } } } void RecurringOp_Stopped(resource_t * rsc, action_t * start, node_t * node, xmlNode * operation, pe_working_set_t * data_set) { char *key = NULL; const char *name = NULL; const char *role = NULL; const char *interval = NULL; const char *node_uname = NULL; unsigned long long interval_ms = 0; GListPtr possible_matches = NULL; GListPtr gIter = NULL; /* TODO: Support of non-unique clone */ if (is_set(rsc->flags, pe_rsc_unique) == FALSE) { return; } /* Only process for the operations with role="Stopped" */ role = crm_element_value(operation, "role"); if (role == NULL || text2role(role) != RSC_ROLE_STOPPED) { return; } pe_rsc_trace(rsc, "Creating recurring actions %s for %s in role %s on nodes where it'll not be running", ID(operation), rsc->id, role2text(rsc->next_role)); if (node != NULL) { node_uname = node->details->uname; } interval = crm_element_value(operation, XML_LRM_ATTR_INTERVAL); interval_ms = crm_get_interval(interval); if (interval_ms == 0) { return; } name = crm_element_value(operation, "name"); if (is_op_dup(rsc, name, interval)) { return; } if (safe_str_eq(name, RSC_STOP) || safe_str_eq(name, RSC_START) || safe_str_eq(name, RSC_DEMOTE) || safe_str_eq(name, RSC_PROMOTE) ) { crm_config_err("Invalid recurring action %s wth name: '%s'", ID(operation), name); return; } key = generate_op_key(rsc->id, name, interval_ms); if (find_rsc_op_entry(rsc, key) == NULL) { /* disabled */ free(key); return; } /* if the monitor exists on the node where the resource will be running, cancel it */ if (node != NULL) { possible_matches = find_actions_exact(rsc->actions, key, node); if (possible_matches) { action_t *cancel_op = NULL; char *local_key = strdup(key); g_list_free(possible_matches); cancel_op = custom_action(rsc, local_key, RSC_CANCEL, node, FALSE, TRUE, data_set); free(cancel_op->task); free(cancel_op->cancel_task); cancel_op->task = strdup(RSC_CANCEL); cancel_op->cancel_task = strdup(name); add_hash_param(cancel_op->meta, XML_LRM_ATTR_INTERVAL, interval); add_hash_param(cancel_op->meta, XML_LRM_ATTR_TASK, name); local_key = NULL; if (rsc->next_role == RSC_ROLE_STARTED || rsc->next_role == RSC_ROLE_SLAVE) { /* rsc->role == RSC_ROLE_STOPPED: cancel the monitor before start */ /* rsc->role == RSC_ROLE_STARTED: for a migration, cancel the monitor on the target node before start */ custom_action_order(rsc, NULL, cancel_op, rsc, start_key(rsc), NULL, pe_order_runnable_left, data_set); } pe_rsc_info(rsc, "Cancel action %s (%s vs. %s) on %s", key, role, role2text(rsc->next_role), crm_str(node_uname)); } } for (gIter = data_set->nodes; gIter != NULL; gIter = gIter->next) { node_t *stop_node = (node_t *) gIter->data; const char *stop_node_uname = stop_node->details->uname; gboolean is_optional = TRUE; gboolean probe_is_optional = TRUE; gboolean stop_is_optional = TRUE; action_t *stopped_mon = NULL; char *rc_inactive = NULL; GListPtr probe_complete_ops = NULL; GListPtr stop_ops = NULL; GListPtr local_gIter = NULL; char *stop_op_key = NULL; if (node_uname && safe_str_eq(stop_node_uname, node_uname)) { continue; } pe_rsc_trace(rsc, "Creating recurring action %s for %s on %s", ID(operation), rsc->id, crm_str(stop_node_uname)); /* start a monitor for an already stopped resource */ possible_matches = find_actions_exact(rsc->actions, key, stop_node); if (possible_matches == NULL) { pe_rsc_trace(rsc, "Marking %s manditory on %s: not active", key, crm_str(stop_node_uname)); is_optional = FALSE; } else { pe_rsc_trace(rsc, "Marking %s optional on %s: already active", key, crm_str(stop_node_uname)); is_optional = TRUE; g_list_free(possible_matches); } stopped_mon = custom_action(rsc, strdup(key), name, stop_node, is_optional, TRUE, data_set); rc_inactive = crm_itoa(PCMK_OCF_NOT_RUNNING); add_hash_param(stopped_mon->meta, XML_ATTR_TE_TARGET_RC, rc_inactive); free(rc_inactive); probe_complete_ops = find_actions(data_set->actions, CRM_OP_PROBED, NULL); for (local_gIter = probe_complete_ops; local_gIter != NULL; local_gIter = local_gIter->next) { action_t *probe_complete = (action_t *) local_gIter->data; if (probe_complete->node == NULL) { if (is_set(probe_complete->flags, pe_action_optional) == FALSE) { probe_is_optional = FALSE; } if (is_set(probe_complete->flags, pe_action_runnable) == FALSE) { crm_debug("%s\t %s (cancelled : probe un-runnable)", crm_str(stop_node_uname), stopped_mon->uuid); update_action_flags(stopped_mon, pe_action_runnable | pe_action_clear); } if (is_set(rsc->flags, pe_rsc_managed)) { custom_action_order(NULL, NULL, probe_complete, NULL, strdup(key), stopped_mon, pe_order_optional, data_set); } break; } } if (probe_complete_ops) { g_list_free(probe_complete_ops); } stop_op_key = stop_key(rsc); stop_ops = find_actions_exact(rsc->actions, stop_op_key, stop_node); for (local_gIter = stop_ops; local_gIter != NULL; local_gIter = local_gIter->next) { action_t *stop = (action_t *) local_gIter->data; if (is_set(stop->flags, pe_action_optional) == FALSE) { stop_is_optional = FALSE; } if (is_set(stop->flags, pe_action_runnable) == FALSE) { crm_debug("%s\t %s (cancelled : stop un-runnable)", crm_str(stop_node_uname), stopped_mon->uuid); update_action_flags(stopped_mon, pe_action_runnable | pe_action_clear); } if (is_set(rsc->flags, pe_rsc_managed)) { custom_action_order(rsc, strdup(stop_op_key), stop, NULL, strdup(key), stopped_mon, pe_order_implies_then | pe_order_runnable_left, data_set); } } if (stop_ops) { g_list_free(stop_ops); } free(stop_op_key); if (is_optional == FALSE && probe_is_optional && stop_is_optional && is_set(rsc->flags, pe_rsc_managed) == FALSE) { pe_rsc_trace(rsc, "Marking %s optional on %s due to unmanaged", key, crm_str(stop_node_uname)); update_action_flags(stopped_mon, pe_action_optional); } if (is_set(stopped_mon->flags, pe_action_optional)) { pe_rsc_trace(rsc, "%s\t %s (optional)", crm_str(stop_node_uname), stopped_mon->uuid); } if (stop_node->details->online == FALSE || stop_node->details->unclean) { pe_rsc_debug(rsc, "%s\t %s (cancelled : no node available)", crm_str(stop_node_uname), stopped_mon->uuid); update_action_flags(stopped_mon, pe_action_runnable | pe_action_clear); } if (is_set(stopped_mon->flags, pe_action_runnable) && is_set(stopped_mon->flags, pe_action_optional) == FALSE) { crm_notice(" Start recurring %s (%llus) for %s on %s", stopped_mon->task, interval_ms / 1000, rsc->id, crm_str(stop_node_uname)); } } free(key); } void Recurring_Stopped(resource_t * rsc, action_t * start, node_t * node, pe_working_set_t * data_set) { if (is_not_set(rsc->flags, pe_rsc_maintenance) && (node == NULL || node->details->maintenance == FALSE)) { xmlNode *operation = NULL; for (operation = __xml_first_child(rsc->ops_xml); operation != NULL; operation = __xml_next(operation)) { if (crm_str_eq((const char *)operation->name, "op", TRUE)) { RecurringOp_Stopped(rsc, start, node, operation, data_set); } } } } static void handle_migration_actions(resource_t * rsc, node_t *current, node_t *chosen, pe_working_set_t * data_set) { action_t *migrate_to = NULL; action_t *migrate_from = NULL; action_t *start = NULL; action_t *stop = NULL; gboolean partial = rsc->partial_migration_target ? TRUE : FALSE; pe_rsc_trace(rsc, "Processing migration actions %s moving from %s to %s . partial migration = %s", rsc->id, current->details->id, chosen->details->id, partial ? "TRUE" : "FALSE"); start = start_action(rsc, chosen, TRUE); stop = stop_action(rsc, current, TRUE); if (partial == FALSE) { migrate_to = custom_action(rsc, generate_op_key(rsc->id, RSC_MIGRATE, 0), RSC_MIGRATE, current, TRUE, TRUE, data_set); } migrate_from = custom_action(rsc, generate_op_key(rsc->id, RSC_MIGRATED, 0), RSC_MIGRATED, chosen, TRUE, TRUE, data_set); if ((migrate_to && migrate_from) || (migrate_from && partial)) { set_bit(start->flags, pe_action_migrate_runnable); set_bit(stop->flags, pe_action_migrate_runnable); update_action_flags(start, pe_action_pseudo); /* easier than trying to delete it from the graph */ /* order probes before migrations */ if (partial) { set_bit(migrate_from->flags, pe_action_migrate_runnable); migrate_from->needs = start->needs; custom_action_order(rsc, generate_op_key(rsc->id, RSC_STATUS, 0), NULL, rsc, generate_op_key(rsc->id, RSC_MIGRATED, 0), NULL, pe_order_optional, data_set); } else { set_bit(migrate_from->flags, pe_action_migrate_runnable); set_bit(migrate_to->flags, pe_action_migrate_runnable); migrate_to->needs = start->needs; custom_action_order(rsc, generate_op_key(rsc->id, RSC_STATUS, 0), NULL, rsc, generate_op_key(rsc->id, RSC_MIGRATE, 0), NULL, pe_order_optional, data_set); custom_action_order(rsc, generate_op_key(rsc->id, RSC_MIGRATE, 0), NULL, rsc, generate_op_key(rsc->id, RSC_MIGRATED, 0), NULL, pe_order_optional | pe_order_implies_first_migratable, data_set); } custom_action_order(rsc, generate_op_key(rsc->id, RSC_MIGRATED, 0), NULL, rsc, generate_op_key(rsc->id, RSC_STOP, 0), NULL, pe_order_optional | pe_order_implies_first_migratable, data_set); custom_action_order(rsc, generate_op_key(rsc->id, RSC_MIGRATED, 0), NULL, rsc, generate_op_key(rsc->id, RSC_START, 0), NULL, pe_order_optional | pe_order_implies_first_migratable | pe_order_pseudo_left, data_set); } if (migrate_to) { add_hash_param(migrate_to->meta, XML_LRM_ATTR_MIGRATE_SOURCE, current->details->uname); add_hash_param(migrate_to->meta, XML_LRM_ATTR_MIGRATE_TARGET, chosen->details->uname); /* pcmk remote connections don't require pending to be recorded in cib. * We can optimize cib writes by only setting PENDING for non pcmk remote * connection resources */ if (rsc->is_remote_node == FALSE) { /* migrate_to takes place on the source node, but can * have an effect on the target node depending on how * the agent is written. Because of this, we have to maintain * a record that the migrate_to occurred incase the source node * loses membership while the migrate_to action is still in-flight. */ add_hash_param(migrate_to->meta, XML_OP_ATTR_PENDING, "true"); } } if (migrate_from) { add_hash_param(migrate_from->meta, XML_LRM_ATTR_MIGRATE_SOURCE, current->details->uname); add_hash_param(migrate_from->meta, XML_LRM_ATTR_MIGRATE_TARGET, chosen->details->uname); } } void native_create_actions(resource_t * rsc, pe_working_set_t * data_set) { action_t *start = NULL; node_t *chosen = NULL; node_t *current = NULL; gboolean need_stop = FALSE; gboolean is_moving = FALSE; gboolean allow_migrate = is_set(rsc->flags, pe_rsc_allow_migrate) ? TRUE : FALSE; GListPtr gIter = NULL; int num_active_nodes = 0; enum rsc_role_e role = RSC_ROLE_UNKNOWN; enum rsc_role_e next_role = RSC_ROLE_UNKNOWN; CRM_ASSERT(rsc); chosen = rsc->allocated_to; if (chosen != NULL && rsc->next_role == RSC_ROLE_UNKNOWN) { rsc->next_role = RSC_ROLE_STARTED; pe_rsc_trace(rsc, "Fixed next_role: unknown -> %s", role2text(rsc->next_role)); } else if (rsc->next_role == RSC_ROLE_UNKNOWN) { rsc->next_role = RSC_ROLE_STOPPED; pe_rsc_trace(rsc, "Fixed next_role: unknown -> %s", role2text(rsc->next_role)); } pe_rsc_trace(rsc, "Processing state transition for %s %p: %s->%s", rsc->id, rsc, role2text(rsc->role), role2text(rsc->next_role)); if (rsc->running_on) { current = rsc->running_on->data; } for (gIter = rsc->running_on; gIter != NULL; gIter = gIter->next) { node_t *n = (node_t *) gIter->data; if (rsc->partial_migration_source && (n->details == rsc->partial_migration_source->details)) { current = rsc->partial_migration_source; } num_active_nodes++; } get_rsc_attributes(rsc->parameters, rsc, chosen, data_set); for (gIter = rsc->dangling_migrations; gIter != NULL; gIter = gIter->next) { node_t *current = (node_t *) gIter->data; action_t *stop = stop_action(rsc, current, FALSE); set_bit(stop->flags, pe_action_dangle); pe_rsc_trace(rsc, "Forcing a cleanup of %s on %s", rsc->id, current->details->uname); if (is_set(data_set->flags, pe_flag_remove_after_stop)) { DeleteRsc(rsc, current, FALSE, data_set); } } if (num_active_nodes > 1) { if (num_active_nodes == 2 && chosen && rsc->partial_migration_target && rsc->partial_migration_source && (current->details == rsc->partial_migration_source->details) && (chosen->details == rsc->partial_migration_target->details)) { /* Here the chosen node is still the migration target from a partial * migration. Attempt to continue the migration instead of recovering * by stopping the resource everywhere and starting it on a single node. */ pe_rsc_trace(rsc, "Will attempt to continue with a partial migration to target %s from %s", rsc->partial_migration_target->details->id, rsc->partial_migration_source->details->id); } else { const char *type = crm_element_value(rsc->xml, XML_ATTR_TYPE); const char *class = crm_element_value(rsc->xml, XML_AGENT_ATTR_CLASS); if(rsc->partial_migration_target && rsc->partial_migration_source) { crm_notice("Resource %s can no longer migrate to %s. Stopping on %s too", rsc->id, rsc->partial_migration_target->details->uname, rsc->partial_migration_source->details->uname); } else { pe_proc_err("Resource %s (%s::%s) is active on %d nodes %s", rsc->id, class, type, num_active_nodes, recovery2text(rsc->recovery_type)); crm_warn("See %s for more information.", "http://clusterlabs.org/wiki/FAQ#Resource_is_Too_Active"); } if (rsc->recovery_type == recovery_stop_start) { need_stop = TRUE; } /* If by chance a partial migration is in process, * but the migration target is not chosen still, clear all * partial migration data. */ rsc->partial_migration_source = rsc->partial_migration_target = NULL; allow_migrate = FALSE; } } if (is_set(rsc->flags, pe_rsc_start_pending)) { start = start_action(rsc, chosen, TRUE); set_bit(start->flags, pe_action_print_always); } if (current && chosen && current->details != chosen->details) { pe_rsc_trace(rsc, "Moving %s", rsc->id); is_moving = TRUE; need_stop = TRUE; } else if (is_set(rsc->flags, pe_rsc_failed)) { pe_rsc_trace(rsc, "Recovering %s", rsc->id); need_stop = TRUE; } else if (is_set(rsc->flags, pe_rsc_block)) { pe_rsc_trace(rsc, "Block %s", rsc->id); need_stop = TRUE; } else if (rsc->role > RSC_ROLE_STARTED && current != NULL && chosen != NULL) { /* Recovery of a promoted resource */ start = start_action(rsc, chosen, TRUE); if (is_set(start->flags, pe_action_optional) == FALSE) { pe_rsc_trace(rsc, "Forced start %s", rsc->id); need_stop = TRUE; } } pe_rsc_trace(rsc, "Creating actions for %s: %s->%s", rsc->id, role2text(rsc->role), role2text(rsc->next_role)); role = rsc->role; /* Potentiall optional steps on brining the resource down and back up to the same level */ while (role != RSC_ROLE_STOPPED) { next_role = rsc_state_matrix[role][RSC_ROLE_STOPPED]; pe_rsc_trace(rsc, "Down: Executing: %s->%s (%s)%s", role2text(role), role2text(next_role), rsc->id, need_stop ? " required" : ""); if (rsc_action_matrix[role][next_role] (rsc, current, !need_stop, data_set) == FALSE) { break; } role = next_role; } while (rsc->role <= rsc->next_role && role != rsc->role && is_not_set(rsc->flags, pe_rsc_block)) { next_role = rsc_state_matrix[role][rsc->role]; pe_rsc_trace(rsc, "Up: Executing: %s->%s (%s)%s", role2text(role), role2text(next_role), rsc->id, need_stop ? " required" : ""); if (rsc_action_matrix[role][next_role] (rsc, chosen, !need_stop, data_set) == FALSE) { break; } role = next_role; } role = rsc->role; /* Required steps from this role to the next */ while (role != rsc->next_role) { next_role = rsc_state_matrix[role][rsc->next_role]; pe_rsc_trace(rsc, "Role: Executing: %s->%s = (%s)", role2text(role), role2text(rsc->next_role), role2text(next_role), rsc->id); if (rsc_action_matrix[role][next_role] (rsc, chosen, FALSE, data_set) == FALSE) { break; } role = next_role; } if(is_set(rsc->flags, pe_rsc_block)) { pe_rsc_trace(rsc, "No monitor additional ops for blocked resource"); } else if (rsc->next_role != RSC_ROLE_STOPPED || is_set(rsc->flags, pe_rsc_managed) == FALSE) { pe_rsc_trace(rsc, "Monitor ops for active resource"); start = start_action(rsc, chosen, TRUE); Recurring(rsc, start, chosen, data_set); Recurring_Stopped(rsc, start, chosen, data_set); } else { pe_rsc_trace(rsc, "Monitor ops for in-active resource"); Recurring_Stopped(rsc, NULL, NULL, data_set); } /* if we are stuck in a partial migration, where the target * of the partial migration no longer matches the chosen target. * A full stop/start is required */ if (rsc->partial_migration_target && (chosen == NULL || rsc->partial_migration_target->details != chosen->details)) { pe_rsc_trace(rsc, "Not allowing partial migration to continue. %s", rsc->id); allow_migrate = FALSE; } else if (is_moving == FALSE || is_not_set(rsc->flags, pe_rsc_managed) || is_set(rsc->flags, pe_rsc_failed) || is_set(rsc->flags, pe_rsc_start_pending) || (current->details->unclean == TRUE) || rsc->next_role < RSC_ROLE_STARTED) { allow_migrate = FALSE; } if (allow_migrate) { handle_migration_actions(rsc, current, chosen, data_set); } } static void rsc_avoids_remote_nodes(resource_t *rsc) { GHashTableIter iter; node_t *node = NULL; g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { if (node->details->remote_rsc) { node->weight = -INFINITY; } } } void native_internal_constraints(resource_t * rsc, pe_working_set_t * data_set) { /* This function is on the critical path and worth optimizing as much as possible */ resource_t *top = uber_parent(rsc); int type = pe_order_optional | pe_order_implies_then | pe_order_restart; gboolean is_stonith = is_set(rsc->flags, pe_rsc_fence_device); custom_action_order(rsc, generate_op_key(rsc->id, RSC_STOP, 0), NULL, rsc, generate_op_key(rsc->id, RSC_START, 0), NULL, type, data_set); if (top->variant == pe_master || rsc->role > RSC_ROLE_SLAVE) { custom_action_order(rsc, generate_op_key(rsc->id, RSC_DEMOTE, 0), NULL, rsc, generate_op_key(rsc->id, RSC_STOP, 0), NULL, pe_order_implies_first_master, data_set); custom_action_order(rsc, generate_op_key(rsc->id, RSC_START, 0), NULL, rsc, generate_op_key(rsc->id, RSC_PROMOTE, 0), NULL, pe_order_runnable_left, data_set); } if (is_stonith == FALSE && is_set(data_set->flags, pe_flag_enable_unfencing) && is_set(rsc->flags, pe_rsc_needs_unfencing) && is_not_set(rsc->flags, pe_rsc_have_unfencing)) { /* Check if the node needs to be unfenced first */ node_t *node = NULL; GHashTableIter iter; if(rsc != top) { /* Only create these constraints once, rsc is almost certainly cloned */ clear_bit_recursive(top, pe_rsc_have_unfencing); } g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { action_t *unfence = pe_fence_op(node, "on", TRUE, data_set); custom_action_order(top, generate_op_key(top->id, top == rsc?RSC_STOP:RSC_STOPPED, 0), NULL, NULL, strdup(unfence->uuid), unfence, pe_order_optional, data_set); crm_debug("Stopping %s prior to unfencing %s", top->id, unfence->uuid); custom_action_order(NULL, strdup(unfence->uuid), unfence, top, generate_op_key(top->id, RSC_START, 0), NULL, pe_order_implies_then_on_node, data_set); } } if (is_not_set(rsc->flags, pe_rsc_managed)) { pe_rsc_trace(rsc, "Skipping fencing constraints for unmanaged resource: %s", rsc->id); return; } { action_t *all_stopped = get_pseudo_op(ALL_STOPPED, data_set); custom_action_order(rsc, stop_key(rsc), NULL, NULL, strdup(all_stopped->task), all_stopped, pe_order_implies_then | pe_order_runnable_left, data_set); } if (g_hash_table_size(rsc->utilization) > 0 && safe_str_neq(data_set->placement_strategy, "default")) { GHashTableIter iter; node_t *next = NULL; GListPtr gIter = NULL; pe_rsc_trace(rsc, "Creating utilization constraints for %s - strategy: %s", rsc->id, data_set->placement_strategy); for (gIter = rsc->running_on; gIter != NULL; gIter = gIter->next) { node_t *current = (node_t *) gIter->data; char *load_stopped_task = crm_concat(LOAD_STOPPED, current->details->uname, '_'); action_t *load_stopped = get_pseudo_op(load_stopped_task, data_set); if (load_stopped->node == NULL) { load_stopped->node = node_copy(current); update_action_flags(load_stopped, pe_action_optional | pe_action_clear); } custom_action_order(rsc, stop_key(rsc), NULL, NULL, load_stopped_task, load_stopped, pe_order_load, data_set); } g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&next)) { char *load_stopped_task = crm_concat(LOAD_STOPPED, next->details->uname, '_'); action_t *load_stopped = get_pseudo_op(load_stopped_task, data_set); if (load_stopped->node == NULL) { load_stopped->node = node_copy(next); update_action_flags(load_stopped, pe_action_optional | pe_action_clear); } custom_action_order(NULL, strdup(load_stopped_task), load_stopped, rsc, start_key(rsc), NULL, pe_order_load, data_set); custom_action_order(NULL, strdup(load_stopped_task), load_stopped, rsc, generate_op_key(rsc->id, RSC_MIGRATE, 0), NULL, pe_order_load, data_set); free(load_stopped_task); } } if (rsc->container) { resource_t *remote_rsc = NULL; /* find out if the container is associated with remote node connection resource */ if (rsc->container->is_remote_node) { remote_rsc = rsc->container; } else if (rsc->is_remote_node == FALSE) { remote_rsc = rsc_contains_remote_node(data_set, rsc->container); } /* if the container is a remote-node, force the resource within the container * instead of colocating the resource with the container. */ if (remote_rsc) { GHashTableIter iter; node_t *node = NULL; g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { if (node->details->remote_rsc != remote_rsc) { node->weight = -INFINITY; } } } else { crm_trace("Generating order and colocation rules for rsc %s with container %s", rsc->id, rsc->container->id); custom_action_order(rsc->container, generate_op_key(rsc->container->id, RSC_START, 0), NULL, rsc, generate_op_key(rsc->id, RSC_START, 0), NULL, pe_order_implies_then | pe_order_runnable_left, data_set); custom_action_order(rsc, generate_op_key(rsc->id, RSC_STOP, 0), NULL, rsc->container, generate_op_key(rsc->container->id, RSC_STOP, 0), NULL, pe_order_implies_first, data_set); rsc_colocation_new("resource-with-containter", NULL, INFINITY, rsc, rsc->container, NULL, NULL, data_set); } } if (rsc->is_remote_node || is_stonith) { /* don't allow remote nodes to run stonith devices * or remote connection resources.*/ rsc_avoids_remote_nodes(rsc); } /* If this rsc is a remote connection resource associated * with a container ( which will most likely be a virtual guest ) * do not allow the container to live on any remote-nodes. * remote-nodes managing nested remote-nodes should not be allowed. */ if (rsc->is_remote_node && rsc->container) { rsc_avoids_remote_nodes(rsc->container); } } void native_rsc_colocation_lh(resource_t * rsc_lh, resource_t * rsc_rh, rsc_colocation_t * constraint) { if (rsc_lh == NULL) { pe_err("rsc_lh was NULL for %s", constraint->id); return; } else if (constraint->rsc_rh == NULL) { pe_err("rsc_rh was NULL for %s", constraint->id); return; } pe_rsc_trace(rsc_lh, "Processing colocation constraint between %s and %s", rsc_lh->id, rsc_rh->id); rsc_rh->cmds->rsc_colocation_rh(rsc_lh, rsc_rh, constraint); } enum filter_colocation_res { influence_nothing = 0, influence_rsc_location, influence_rsc_priority, }; static enum filter_colocation_res filter_colocation_constraint(resource_t * rsc_lh, resource_t * rsc_rh, rsc_colocation_t * constraint) { if (constraint->score == 0) { return influence_nothing; } /* rh side must be allocated before we can process constraint */ if (is_set(rsc_rh->flags, pe_rsc_provisional)) { return influence_nothing; } if ((constraint->role_lh >= RSC_ROLE_SLAVE) && rsc_lh->parent && rsc_lh->parent->variant == pe_master && is_not_set(rsc_lh->flags, pe_rsc_provisional)) { /* LH and RH resources have already been allocated, place the correct * priority oh LH rsc for the given multistate resource role */ return influence_rsc_priority; } if (is_not_set(rsc_lh->flags, pe_rsc_provisional)) { /* error check */ struct node_shared_s *details_lh; struct node_shared_s *details_rh; if ((constraint->score > -INFINITY) && (constraint->score < INFINITY)) { return influence_nothing; } details_rh = rsc_rh->allocated_to ? rsc_rh->allocated_to->details : NULL; details_lh = rsc_lh->allocated_to ? rsc_lh->allocated_to->details : NULL; if (constraint->score == INFINITY && details_lh != details_rh) { crm_err("%s and %s are both allocated" " but to different nodes: %s vs. %s", rsc_lh->id, rsc_rh->id, details_lh ? details_lh->uname : "n/a", details_rh ? details_rh->uname : "n/a"); } else if (constraint->score == -INFINITY && details_lh == details_rh) { crm_err("%s and %s are both allocated" " but to the SAME node: %s", rsc_lh->id, rsc_rh->id, details_rh ? details_rh->uname : "n/a"); } return influence_nothing; } if (constraint->score > 0 && constraint->role_lh != RSC_ROLE_UNKNOWN && constraint->role_lh != rsc_lh->next_role) { crm_trace("LH: Skipping constraint: \"%s\" state filter nextrole is %s", role2text(constraint->role_lh), role2text(rsc_lh->next_role)); return influence_nothing; } if (constraint->score > 0 && constraint->role_rh != RSC_ROLE_UNKNOWN && constraint->role_rh != rsc_rh->next_role) { crm_trace("RH: Skipping constraint: \"%s\" state filter", role2text(constraint->role_rh)); return FALSE; } if (constraint->score < 0 && constraint->role_lh != RSC_ROLE_UNKNOWN && constraint->role_lh == rsc_lh->next_role) { crm_trace("LH: Skipping -ve constraint: \"%s\" state filter", role2text(constraint->role_lh)); return influence_nothing; } if (constraint->score < 0 && constraint->role_rh != RSC_ROLE_UNKNOWN && constraint->role_rh == rsc_rh->next_role) { crm_trace("RH: Skipping -ve constraint: \"%s\" state filter", role2text(constraint->role_rh)); return influence_nothing; } return influence_rsc_location; } static void influence_priority(resource_t * rsc_lh, resource_t * rsc_rh, rsc_colocation_t * constraint) { const char *rh_value = NULL; const char *lh_value = NULL; const char *attribute = "#id"; int score_multiplier = 1; if (constraint->node_attribute != NULL) { attribute = constraint->node_attribute; } if (!rsc_rh->allocated_to || !rsc_lh->allocated_to) { return; } lh_value = g_hash_table_lookup(rsc_lh->allocated_to->details->attrs, attribute); rh_value = g_hash_table_lookup(rsc_rh->allocated_to->details->attrs, attribute); if (!safe_str_eq(lh_value, rh_value)) { if(constraint->score == INFINITY && constraint->role_lh == RSC_ROLE_MASTER) { rsc_lh->priority = -INFINITY; } return; } if (constraint->role_rh && (constraint->role_rh != rsc_rh->next_role)) { return; } if (constraint->role_lh == RSC_ROLE_SLAVE) { score_multiplier = -1; } rsc_lh->priority = merge_weights(score_multiplier * constraint->score, rsc_lh->priority); } static void colocation_match(resource_t * rsc_lh, resource_t * rsc_rh, rsc_colocation_t * constraint) { const char *tmp = NULL; const char *value = NULL; const char *attribute = "#id"; GHashTable *work = NULL; gboolean do_check = FALSE; GHashTableIter iter; node_t *node = NULL; if (constraint->node_attribute != NULL) { attribute = constraint->node_attribute; } if (rsc_rh->allocated_to) { value = g_hash_table_lookup(rsc_rh->allocated_to->details->attrs, attribute); do_check = TRUE; } else if (constraint->score < 0) { /* nothing to do: * anti-colocation with something thats not running */ return; } work = node_hash_dup(rsc_lh->allowed_nodes); g_hash_table_iter_init(&iter, work); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { tmp = g_hash_table_lookup(node->details->attrs, attribute); if (do_check && safe_str_eq(tmp, value)) { if (constraint->score < INFINITY) { pe_rsc_trace(rsc_lh, "%s: %s.%s += %d", constraint->id, rsc_lh->id, node->details->uname, constraint->score); node->weight = merge_weights(constraint->score, node->weight); } } else if (do_check == FALSE || constraint->score >= INFINITY) { pe_rsc_trace(rsc_lh, "%s: %s.%s -= %d (%s)", constraint->id, rsc_lh->id, node->details->uname, constraint->score, do_check ? "failed" : "unallocated"); node->weight = merge_weights(-constraint->score, node->weight); } } if (can_run_any(work) || constraint->score <= -INFINITY || constraint->score >= INFINITY) { g_hash_table_destroy(rsc_lh->allowed_nodes); rsc_lh->allowed_nodes = work; work = NULL; } else { static char score[33]; score2char_stack(constraint->score, score, sizeof(score)); pe_rsc_info(rsc_lh, "%s: Rolling back scores from %s (%d, %s)", rsc_lh->id, rsc_rh->id, do_check, score); } if (work) { g_hash_table_destroy(work); } } void native_rsc_colocation_rh(resource_t * rsc_lh, resource_t * rsc_rh, rsc_colocation_t * constraint) { enum filter_colocation_res filter_results; CRM_ASSERT(rsc_lh); CRM_ASSERT(rsc_rh); filter_results = filter_colocation_constraint(rsc_lh, rsc_rh, constraint); pe_rsc_trace(rsc_lh, "%sColocating %s with %s (%s, weight=%d, filter=%d)", constraint->score >= 0 ? "" : "Anti-", rsc_lh->id, rsc_rh->id, constraint->id, constraint->score, filter_results); switch (filter_results) { case influence_rsc_priority: influence_priority(rsc_lh, rsc_rh, constraint); break; case influence_rsc_location: pe_rsc_trace(rsc_lh, "%sColocating %s with %s (%s, weight=%d)", constraint->score >= 0 ? "" : "Anti-", rsc_lh->id, rsc_rh->id, constraint->id, constraint->score); colocation_match(rsc_lh, rsc_rh, constraint); break; case influence_nothing: default: return; } } static gboolean filter_rsc_ticket(resource_t * rsc_lh, rsc_ticket_t * rsc_ticket) { if (rsc_ticket->role_lh != RSC_ROLE_UNKNOWN && rsc_ticket->role_lh != rsc_lh->role) { pe_rsc_trace(rsc_lh, "LH: Skipping constraint: \"%s\" state filter", role2text(rsc_ticket->role_lh)); return FALSE; } return TRUE; } void rsc_ticket_constraint(resource_t * rsc_lh, rsc_ticket_t * rsc_ticket, pe_working_set_t * data_set) { if (rsc_ticket == NULL) { pe_err("rsc_ticket was NULL"); return; } if (rsc_lh == NULL) { pe_err("rsc_lh was NULL for %s", rsc_ticket->id); return; } if (rsc_ticket->ticket->granted && rsc_ticket->ticket->standby == FALSE) { return; } if (rsc_lh->children) { GListPtr gIter = rsc_lh->children; pe_rsc_trace(rsc_lh, "Processing ticket dependencies from %s", rsc_lh->id); for (; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; rsc_ticket_constraint(child_rsc, rsc_ticket, data_set); } return; } pe_rsc_trace(rsc_lh, "%s: Processing ticket dependency on %s (%s, %s)", rsc_lh->id, rsc_ticket->ticket->id, rsc_ticket->id, role2text(rsc_ticket->role_lh)); if (rsc_ticket->ticket->granted == FALSE && g_list_length(rsc_lh->running_on) > 0) { GListPtr gIter = NULL; switch (rsc_ticket->loss_policy) { case loss_ticket_stop: resource_location(rsc_lh, NULL, -INFINITY, "__loss_of_ticket__", data_set); break; case loss_ticket_demote: /*Promotion score will be set to -INFINITY in master_promotion_order() */ if (rsc_ticket->role_lh != RSC_ROLE_MASTER) { resource_location(rsc_lh, NULL, -INFINITY, "__loss_of_ticket__", data_set); } break; case loss_ticket_fence: if (filter_rsc_ticket(rsc_lh, rsc_ticket) == FALSE) { return; } resource_location(rsc_lh, NULL, -INFINITY, "__loss_of_ticket__", data_set); for (gIter = rsc_lh->running_on; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; pe_fence_node(data_set, node, "deadman ticket lost"); } break; case loss_ticket_freeze: if (filter_rsc_ticket(rsc_lh, rsc_ticket) == FALSE) { return; } if (g_list_length(rsc_lh->running_on) > 0) { clear_bit(rsc_lh->flags, pe_rsc_managed); set_bit(rsc_lh->flags, pe_rsc_block); } break; } } else if (rsc_ticket->ticket->granted == FALSE) { if (rsc_ticket->role_lh != RSC_ROLE_MASTER || rsc_ticket->loss_policy == loss_ticket_stop) { resource_location(rsc_lh, NULL, -INFINITY, "__no_ticket__", data_set); } } else if (rsc_ticket->ticket->standby) { if (rsc_ticket->role_lh != RSC_ROLE_MASTER || rsc_ticket->loss_policy == loss_ticket_stop) { resource_location(rsc_lh, NULL, -INFINITY, "__ticket_standby__", data_set); } } } enum pe_action_flags native_action_flags(action_t * action, node_t * node) { return action->flags; } enum pe_graph_flags native_update_actions(action_t * first, action_t * then, node_t * node, enum pe_action_flags flags, enum pe_action_flags filter, enum pe_ordering type) { /* flags == get_action_flags(first, then_node) called from update_action() */ enum pe_graph_flags changed = pe_graph_none; enum pe_action_flags then_flags = then->flags; enum pe_action_flags first_flags = first->flags; crm_trace( "Testing %s on %s (0x%.6x) with %s 0x%.6x %x %x", first->uuid, first->node ? first->node->details->uname : "[none]", first->flags, then->uuid, then->flags); if (type & pe_order_asymmetrical) { resource_t *then_rsc = then->rsc; enum rsc_role_e then_rsc_role = then_rsc ? then_rsc->fns->state(then_rsc, TRUE) : 0; if (!then_rsc) { /* ignore */ } else if ((then_rsc_role == RSC_ROLE_STOPPED) && safe_str_eq(then->task, RSC_STOP)) { /* ignore... if 'then' is supposed to be stopped after 'first', but * then is already stopped, there is nothing to be done when non-symmetrical. */ } else if ((then_rsc_role >= RSC_ROLE_STARTED) && safe_str_eq(then->task, RSC_START)) { /* ignore... if 'then' is supposed to be started after 'first', but * then is already started, there is nothing to be done when non-symmetrical. */ } else if (!(first->flags & pe_action_runnable)) { /* prevent 'then' action from happening if 'first' is not runnable and * 'then' has not yet occurred. */ pe_clear_action_bit(then, pe_action_runnable); pe_clear_action_bit(then, pe_action_optional); pe_rsc_trace(then->rsc, "Unset optional and runnable on %s", then->uuid); } else { /* ignore... then is allowed to start/stop if it wants to. */ } } if (type & pe_order_implies_first) { if ((filter & pe_action_optional) && (flags & pe_action_optional) == 0) { pe_rsc_trace(first->rsc, "Unset optional on %s because of %s", first->uuid, then->uuid); pe_clear_action_bit(first, pe_action_optional); } if (is_set(flags, pe_action_migrate_runnable) && is_set(then->flags, pe_action_migrate_runnable) == FALSE && is_set(then->flags, pe_action_optional) == FALSE) { pe_rsc_trace(first->rsc, "Unset migrate runnable on %s because of %s", first->uuid, then->uuid); pe_clear_action_bit(first, pe_action_migrate_runnable); } } if (type & pe_order_implies_first_master) { if ((filter & pe_action_optional) && ((then->flags & pe_action_optional) == FALSE) && then->rsc && (then->rsc->role == RSC_ROLE_MASTER)) { pe_clear_action_bit(first, pe_action_optional); if (is_set(first->flags, pe_action_migrate_runnable) && is_set(then->flags, pe_action_migrate_runnable) == FALSE) { pe_rsc_trace(first->rsc, "Unset migrate runnable on %s because of %s", first->uuid, then->uuid); pe_clear_action_bit(first, pe_action_migrate_runnable); } pe_rsc_trace(then->rsc, "Unset optional on %s because of %s", first->uuid, then->uuid); } } if ((type & pe_order_implies_first_migratable) && is_set(filter, pe_action_optional)) { if (((then->flags & pe_action_migrate_runnable) == FALSE) || ((then->flags & pe_action_runnable) == FALSE)) { pe_rsc_trace(then->rsc, "Unset runnable on %s because %s is neither runnable or migratable", first->uuid, then->uuid); pe_clear_action_bit(first, pe_action_runnable); } if ((then->flags & pe_action_optional) == 0) { pe_rsc_trace(then->rsc, "Unset optional on %s because %s is not optional", first->uuid, then->uuid); pe_clear_action_bit(first, pe_action_optional); } } if ((type & pe_order_pseudo_left) && is_set(filter, pe_action_optional)) { if ((first->flags & pe_action_runnable) == FALSE) { pe_clear_action_bit(then, pe_action_migrate_runnable); pe_clear_action_bit(then, pe_action_pseudo); pe_rsc_trace(then->rsc, "Unset pseudo on %s because %s is not runnable", then->uuid, first->uuid); } } if (is_set(type, pe_order_runnable_left) && is_set(filter, pe_action_runnable) && is_set(then->flags, pe_action_runnable) && is_set(flags, pe_action_runnable) == FALSE) { pe_rsc_trace(then->rsc, "Unset runnable on %s because of %s", then->uuid, first->uuid); pe_clear_action_bit(then, pe_action_runnable); pe_clear_action_bit(then, pe_action_migrate_runnable); } if (is_set(type, pe_order_implies_then) && is_set(filter, pe_action_optional) && is_set(then->flags, pe_action_optional) && is_set(flags, pe_action_optional) == FALSE) { /* in this case, treat migrate_runnable as if first is optional */ if (is_set(first->flags, pe_action_migrate_runnable) == FALSE) { pe_rsc_trace(then->rsc, "Unset optional on %s because of %s", then->uuid, first->uuid); pe_clear_action_bit(then, pe_action_optional); } } if (is_set(type, pe_order_restart)) { const char *reason = NULL; CRM_ASSERT(first->rsc && first->rsc->variant == pe_native); CRM_ASSERT(then->rsc && then->rsc->variant == pe_native); if ((filter & pe_action_runnable) && (then->flags & pe_action_runnable) == 0 && (then->rsc->flags & pe_rsc_managed)) { reason = "shutdown"; } if ((filter & pe_action_optional) && (then->flags & pe_action_optional) == 0) { reason = "recover"; } if (reason && is_set(first->flags, pe_action_optional)) { if (is_set(first->flags, pe_action_runnable) || is_not_set(then->flags, pe_action_optional)) { pe_rsc_trace(first->rsc, "Handling %s: %s -> %s", reason, first->uuid, then->uuid); pe_clear_action_bit(first, pe_action_optional); } } if (reason && is_not_set(first->flags, pe_action_optional) && is_not_set(first->flags, pe_action_runnable)) { pe_rsc_trace(then->rsc, "Handling %s: %s -> %s", reason, first->uuid, then->uuid); pe_clear_action_bit(then, pe_action_runnable); } if (reason && is_not_set(first->flags, pe_action_optional) && is_set(first->flags, pe_action_migrate_runnable) && is_not_set(then->flags, pe_action_migrate_runnable)) { pe_clear_action_bit(first, pe_action_migrate_runnable); } } if (then_flags != then->flags) { changed |= pe_graph_updated_then; pe_rsc_trace(then->rsc, "Then: Flags for %s on %s are now 0x%.6x (was 0x%.6x) because of %s 0x%.6x", then->uuid, then->node ? then->node->details->uname : "[none]", then->flags, then_flags, first->uuid, first->flags); if(then->rsc && then->rsc->parent) { /* "X_stop then X_start" doesn't get handled for cloned groups unless we do this */ update_action(then); } } if (first_flags != first->flags) { changed |= pe_graph_updated_first; pe_rsc_trace(first->rsc, "First: Flags for %s on %s are now 0x%.6x (was 0x%.6x) because of %s 0x%.6x", first->uuid, first->node ? first->node->details->uname : "[none]", first->flags, first_flags, then->uuid, then->flags); } return changed; } void native_rsc_location(resource_t * rsc, rsc_to_node_t * constraint) { GListPtr gIter = NULL; GHashTableIter iter; node_t *node = NULL; if (constraint == NULL) { pe_err("Constraint is NULL"); return; } else if (rsc == NULL) { pe_err("LHS of rsc_to_node (%s) is NULL", constraint->id); return; } pe_rsc_trace(rsc, "Applying %s (%s) to %s", constraint->id, role2text(constraint->role_filter), rsc->id); /* take "lifetime" into account */ if (constraint->role_filter > RSC_ROLE_UNKNOWN && constraint->role_filter != rsc->next_role) { pe_rsc_debug(rsc, "Constraint (%s) is not active (role : %s vs. %s)", constraint->id, role2text(constraint->role_filter), role2text(rsc->next_role)); return; } else if (is_active(constraint) == FALSE) { pe_rsc_trace(rsc, "Constraint (%s) is not active", constraint->id); return; } if (constraint->node_list_rh == NULL) { pe_rsc_trace(rsc, "RHS of constraint %s is NULL", constraint->id); return; } for (gIter = constraint->node_list_rh; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; node_t *other_node = NULL; other_node = (node_t *) pe_hash_table_lookup(rsc->allowed_nodes, node->details->id); if (other_node != NULL) { pe_rsc_trace(rsc, "%s + %s: %d + %d", node->details->uname, other_node->details->uname, node->weight, other_node->weight); other_node->weight = merge_weights(other_node->weight, node->weight); } else { other_node = node_copy(node); g_hash_table_insert(rsc->allowed_nodes, (gpointer) other_node->details->id, other_node); } if (other_node->rsc_discover_mode < constraint->discover_mode) { if (constraint->discover_mode == discover_exclusive) { rsc->exclusive_discover = TRUE; } /* exclusive > never > always... always is default */ other_node->rsc_discover_mode = constraint->discover_mode; } } g_hash_table_iter_init(&iter, rsc->allowed_nodes); while (g_hash_table_iter_next(&iter, NULL, (void **)&node)) { pe_rsc_trace(rsc, "%s + %s : %d", rsc->id, node->details->uname, node->weight); } } void native_expand(resource_t * rsc, pe_working_set_t * data_set) { GListPtr gIter = NULL; CRM_ASSERT(rsc); pe_rsc_trace(rsc, "Processing actions from %s", rsc->id); for (gIter = rsc->actions; gIter != NULL; gIter = gIter->next) { action_t *action = (action_t *) gIter->data; crm_trace("processing action %d for rsc=%s", action->id, rsc->id); graph_element_from_action(action, data_set); } for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; child_rsc->cmds->expand(child_rsc, data_set); } } #define log_change(fmt, args...) do { \ if(terminal) { \ printf(" * "fmt"\n", ##args); \ } else { \ crm_notice(fmt, ##args); \ } \ } while(0) #define STOP_SANITY_ASSERT(lineno) do { \ if(current && current->details->unclean) { \ /* It will be a pseduo op */ \ } else if(stop == NULL) { \ crm_err("%s:%d: No stop action exists for %s", __FUNCTION__, lineno, rsc->id); \ CRM_ASSERT(stop != NULL); \ } else if(is_set(stop->flags, pe_action_optional)) { \ crm_err("%s:%d: Action %s is still optional", __FUNCTION__, lineno, stop->uuid); \ CRM_ASSERT(is_not_set(stop->flags, pe_action_optional)); \ } \ } while(0) void LogActions(resource_t * rsc, pe_working_set_t * data_set, gboolean terminal) { node_t *next = NULL; node_t *current = NULL; action_t *stop = NULL; action_t *start = NULL; action_t *demote = NULL; action_t *promote = NULL; char *key = NULL; gboolean moving = FALSE; GListPtr possible_matches = NULL; if (rsc->children) { GListPtr gIter = NULL; for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; LogActions(child_rsc, data_set, terminal); } return; } next = rsc->allocated_to; if (rsc->running_on) { if (g_list_length(rsc->running_on) > 1 && rsc->partial_migration_source) { current = rsc->partial_migration_source; } else { current = rsc->running_on->data; } if (rsc->role == RSC_ROLE_STOPPED) { /* * This can occur when resources are being recovered * We fiddle with the current role in native_create_actions() */ rsc->role = RSC_ROLE_STARTED; } } if (current == NULL && is_set(rsc->flags, pe_rsc_orphan)) { /* Don't log stopped orphans */ return; } if (is_not_set(rsc->flags, pe_rsc_managed) || (current == NULL && next == NULL)) { pe_rsc_info(rsc, "Leave %s\t(%s%s)", rsc->id, role2text(rsc->role), is_not_set(rsc->flags, pe_rsc_managed) ? " unmanaged" : ""); return; } if (current != NULL && next != NULL && safe_str_neq(current->details->id, next->details->id)) { moving = TRUE; } key = start_key(rsc); possible_matches = find_actions(rsc->actions, key, next); free(key); if (possible_matches) { start = possible_matches->data; g_list_free(possible_matches); } key = stop_key(rsc); if(start == NULL || is_set(start->flags, pe_action_runnable) == FALSE) { possible_matches = find_actions(rsc->actions, key, NULL); } else { possible_matches = find_actions(rsc->actions, key, current); } free(key); if (possible_matches) { stop = possible_matches->data; g_list_free(possible_matches); } key = promote_key(rsc); possible_matches = find_actions(rsc->actions, key, next); free(key); if (possible_matches) { promote = possible_matches->data; g_list_free(possible_matches); } key = demote_key(rsc); possible_matches = find_actions(rsc->actions, key, next); free(key); if (possible_matches) { demote = possible_matches->data; g_list_free(possible_matches); } if (rsc->role == rsc->next_role) { action_t *migrate_to = NULL; key = generate_op_key(rsc->id, RSC_MIGRATED, 0); possible_matches = find_actions(rsc->actions, key, next); free(key); if (possible_matches) { migrate_to = possible_matches->data; } CRM_CHECK(next != NULL,); if (next == NULL) { } else if (migrate_to && is_set(migrate_to->flags, pe_action_runnable) && current) { log_change("Migrate %s\t(%s %s -> %s)", rsc->id, role2text(rsc->role), current->details->uname, next->details->uname); } else if (is_set(rsc->flags, pe_rsc_reload)) { log_change("Reload %s\t(%s %s)", rsc->id, role2text(rsc->role), next->details->uname); } else if (start == NULL || is_set(start->flags, pe_action_optional)) { pe_rsc_info(rsc, "Leave %s\t(%s %s)", rsc->id, role2text(rsc->role), next->details->uname); } else if (start && is_set(start->flags, pe_action_runnable) == FALSE) { log_change("Stop %s\t(%s %s%s)", rsc->id, role2text(rsc->role), current?current->details->uname:"N/A", stop && is_not_set(stop->flags, pe_action_runnable) ? " - blocked" : ""); STOP_SANITY_ASSERT(__LINE__); } else if (moving && current) { log_change("%s %s\t(%s %s -> %s)", is_set(rsc->flags, pe_rsc_failed) ? "Recover" : "Move ", rsc->id, role2text(rsc->role), current->details->uname, next->details->uname); } else if (is_set(rsc->flags, pe_rsc_failed)) { log_change("Recover %s\t(%s %s)", rsc->id, role2text(rsc->role), next->details->uname); STOP_SANITY_ASSERT(__LINE__); } else { log_change("Restart %s\t(%s %s)", rsc->id, role2text(rsc->role), next->details->uname); /* STOP_SANITY_ASSERT(__LINE__); False positive for migrate-fail-7 */ } g_list_free(possible_matches); return; } if (rsc->role > RSC_ROLE_SLAVE && rsc->role > rsc->next_role) { CRM_CHECK(current != NULL,); if (current != NULL) { gboolean allowed = FALSE; if (demote != NULL && (demote->flags & pe_action_runnable)) { allowed = TRUE; } log_change("Demote %s\t(%s -> %s %s%s)", rsc->id, role2text(rsc->role), role2text(rsc->next_role), current->details->uname, allowed ? "" : " - blocked"); if (stop != NULL && is_not_set(stop->flags, pe_action_optional) && rsc->next_role > RSC_ROLE_STOPPED && moving == FALSE) { if (is_set(rsc->flags, pe_rsc_failed)) { log_change("Recover %s\t(%s %s)", rsc->id, role2text(rsc->role), next->details->uname); STOP_SANITY_ASSERT(__LINE__); } else if (is_set(rsc->flags, pe_rsc_reload)) { log_change("Reload %s\t(%s %s)", rsc->id, role2text(rsc->role), next->details->uname); } else { log_change("Restart %s\t(%s %s)", rsc->id, role2text(rsc->next_role), next->details->uname); STOP_SANITY_ASSERT(__LINE__); } } } } else if (rsc->next_role == RSC_ROLE_STOPPED) { GListPtr gIter = NULL; CRM_CHECK(current != NULL,); key = stop_key(rsc); for (gIter = rsc->running_on; gIter != NULL; gIter = gIter->next) { node_t *node = (node_t *) gIter->data; action_t *stop_op = NULL; gboolean allowed = FALSE; possible_matches = find_actions(rsc->actions, key, node); if (possible_matches) { stop_op = possible_matches->data; g_list_free(possible_matches); } if (stop_op && (stop_op->flags & pe_action_runnable)) { STOP_SANITY_ASSERT(__LINE__); allowed = TRUE; } log_change("Stop %s\t(%s%s)", rsc->id, node->details->uname, allowed ? "" : " - blocked"); } free(key); } if (moving) { log_change("Move %s\t(%s %s -> %s)", rsc->id, role2text(rsc->next_role), current->details->uname, next->details->uname); STOP_SANITY_ASSERT(__LINE__); } if (rsc->role == RSC_ROLE_STOPPED) { gboolean allowed = FALSE; if (start && (start->flags & pe_action_runnable)) { allowed = TRUE; } CRM_CHECK(next != NULL,); if (next != NULL) { log_change("Start %s\t(%s%s)", rsc->id, next->details->uname, allowed ? "" : " - blocked"); } if (allowed == FALSE) { return; } } if (rsc->next_role > RSC_ROLE_SLAVE && rsc->role < rsc->next_role) { gboolean allowed = FALSE; CRM_LOG_ASSERT(next); if (stop != NULL && is_not_set(stop->flags, pe_action_optional) && rsc->role > RSC_ROLE_STOPPED) { if (is_set(rsc->flags, pe_rsc_failed)) { log_change("Recover %s\t(%s %s)", rsc->id, role2text(rsc->role), next?next->details->uname:NULL); STOP_SANITY_ASSERT(__LINE__); } else if (is_set(rsc->flags, pe_rsc_reload)) { log_change("Reload %s\t(%s %s)", rsc->id, role2text(rsc->role), next?next->details->uname:NULL); STOP_SANITY_ASSERT(__LINE__); } else { log_change("Restart %s\t(%s %s)", rsc->id, role2text(rsc->role), next?next->details->uname:NULL); STOP_SANITY_ASSERT(__LINE__); } } if (promote && (promote->flags & pe_action_runnable)) { allowed = TRUE; } log_change("Promote %s\t(%s -> %s %s%s)", rsc->id, role2text(rsc->role), role2text(rsc->next_role), next?next->details->uname:NULL, allowed ? "" : " - blocked"); } } gboolean StopRsc(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set) { GListPtr gIter = NULL; CRM_ASSERT(rsc); pe_rsc_trace(rsc, "%s", rsc->id); for (gIter = rsc->running_on; gIter != NULL; gIter = gIter->next) { node_t *current = (node_t *) gIter->data; action_t *stop; if (rsc->partial_migration_target) { if (rsc->partial_migration_target->details == current->details) { pe_rsc_trace(rsc, "Filtered %s -> %s %s", current->details->uname, next->details->uname, rsc->id); continue; } else { pe_rsc_trace(rsc, "Forced on %s %s", current->details->uname, rsc->id); optional = FALSE; } } pe_rsc_trace(rsc, "%s on %s", rsc->id, current->details->uname); stop = stop_action(rsc, current, optional); if (is_not_set(rsc->flags, pe_rsc_managed)) { update_action_flags(stop, pe_action_runnable | pe_action_clear); } if (is_set(data_set->flags, pe_flag_remove_after_stop)) { DeleteRsc(rsc, current, optional, data_set); } } return TRUE; } gboolean StartRsc(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set) { action_t *start = NULL; CRM_ASSERT(rsc); pe_rsc_trace(rsc, "%s on %s %d", rsc->id, next ? next->details->uname : "N/A", optional); start = start_action(rsc, next, TRUE); if (is_set(start->flags, pe_action_runnable) && optional == FALSE) { update_action_flags(start, pe_action_optional | pe_action_clear); } return TRUE; } gboolean PromoteRsc(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set) { char *key = NULL; GListPtr gIter = NULL; gboolean runnable = TRUE; GListPtr action_list = NULL; CRM_ASSERT(rsc); CRM_CHECK(next != NULL, return FALSE); pe_rsc_trace(rsc, "%s on %s", rsc->id, next->details->uname); key = start_key(rsc); action_list = find_actions_exact(rsc->actions, key, next); free(key); for (gIter = action_list; gIter != NULL; gIter = gIter->next) { action_t *start = (action_t *) gIter->data; if (is_set(start->flags, pe_action_runnable) == FALSE) { runnable = FALSE; } } g_list_free(action_list); if (runnable) { promote_action(rsc, next, optional); return TRUE; } pe_rsc_debug(rsc, "%s\tPromote %s (canceled)", next->details->uname, rsc->id); key = promote_key(rsc); action_list = find_actions_exact(rsc->actions, key, next); free(key); for (gIter = action_list; gIter != NULL; gIter = gIter->next) { action_t *promote = (action_t *) gIter->data; update_action_flags(promote, pe_action_runnable | pe_action_clear); } g_list_free(action_list); return TRUE; } gboolean DemoteRsc(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set) { GListPtr gIter = NULL; CRM_ASSERT(rsc); pe_rsc_trace(rsc, "%s", rsc->id); /* CRM_CHECK(rsc->next_role == RSC_ROLE_SLAVE, return FALSE); */ for (gIter = rsc->running_on; gIter != NULL; gIter = gIter->next) { node_t *current = (node_t *) gIter->data; pe_rsc_trace(rsc, "%s on %s", rsc->id, next ? next->details->uname : "N/A"); demote_action(rsc, current, optional); } return TRUE; } gboolean RoleError(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set) { CRM_ASSERT(rsc); crm_err("%s on %s", rsc->id, next ? next->details->uname : "N/A"); CRM_CHECK(FALSE, return FALSE); return FALSE; } gboolean NullOp(resource_t * rsc, node_t * next, gboolean optional, pe_working_set_t * data_set) { CRM_ASSERT(rsc); pe_rsc_trace(rsc, "%s", rsc->id); return FALSE; } gboolean DeleteRsc(resource_t * rsc, node_t * node, gboolean optional, pe_working_set_t * data_set) { if (is_set(rsc->flags, pe_rsc_failed)) { pe_rsc_trace(rsc, "Resource %s not deleted from %s: failed", rsc->id, node->details->uname); return FALSE; } else if (node == NULL) { pe_rsc_trace(rsc, "Resource %s not deleted: NULL node", rsc->id); return FALSE; } else if (node->details->unclean || node->details->online == FALSE) { pe_rsc_trace(rsc, "Resource %s not deleted from %s: unrunnable", rsc->id, node->details->uname); return FALSE; } crm_notice("Removing %s from %s", rsc->id, node->details->uname); delete_action(rsc, node, optional); new_rsc_order(rsc, RSC_STOP, rsc, RSC_DELETE, optional ? pe_order_implies_then : pe_order_optional, data_set); new_rsc_order(rsc, RSC_DELETE, rsc, RSC_START, optional ? pe_order_implies_then : pe_order_optional, data_set); return TRUE; } #include <../lib/pengine/unpack.h> #define set_char(x) last_rsc_id[lpc] = x; complete = TRUE; static char * increment_clone(char *last_rsc_id) { int lpc = 0; int len = 0; char *tmp = NULL; gboolean complete = FALSE; CRM_CHECK(last_rsc_id != NULL, return NULL); if (last_rsc_id != NULL) { len = strlen(last_rsc_id); } lpc = len - 1; while (complete == FALSE && lpc > 0) { switch (last_rsc_id[lpc]) { case 0: lpc--; break; case '0': set_char('1'); break; case '1': set_char('2'); break; case '2': set_char('3'); break; case '3': set_char('4'); break; case '4': set_char('5'); break; case '5': set_char('6'); break; case '6': set_char('7'); break; case '7': set_char('8'); break; case '8': set_char('9'); break; case '9': last_rsc_id[lpc] = '0'; lpc--; break; case ':': tmp = last_rsc_id; last_rsc_id = calloc(1, len + 2); memcpy(last_rsc_id, tmp, len); last_rsc_id[++lpc] = '1'; last_rsc_id[len] = '0'; last_rsc_id[len + 1] = 0; complete = TRUE; free(tmp); break; default: crm_err("Unexpected char: %c (%d)", last_rsc_id[lpc], lpc); return NULL; break; } } return last_rsc_id; } static node_t * probe_grouped_clone(resource_t * rsc, node_t * node, pe_working_set_t * data_set) { node_t *running = NULL; resource_t *top = uber_parent(rsc); if (running == NULL && is_set(top->flags, pe_rsc_unique) == FALSE) { /* Annoyingly we also need to check any other clone instances * Clumsy, but it will work. * * An alternative would be to update known_on for every peer * during process_rsc_state() * * This code desperately needs optimization * ptest -x with 100 nodes, 100 clones and clone-max=10: * No probes O(25s) * Detection without clone loop O(3m) * Detection with clone loop O(8m) ptest[32211]: 2010/02/18_14:27:55 CRIT: stage5: Probing for unknown resources ptest[32211]: 2010/02/18_14:33:39 CRIT: stage5: Done ptest[32211]: 2010/02/18_14:35:05 CRIT: stage7: Updating action states ptest[32211]: 2010/02/18_14:35:05 CRIT: stage7: Done */ char *clone_id = clone_zero(rsc->id); resource_t *peer = pe_find_resource(top->children, clone_id); while (peer && running == NULL) { running = pe_hash_table_lookup(peer->known_on, node->details->id); if (running != NULL) { /* we already know the status of the resource on this node */ pe_rsc_trace(rsc, "Skipping active clone: %s", rsc->id); free(clone_id); return running; } clone_id = increment_clone(clone_id); peer = pe_find_resource(data_set->resources, clone_id); } free(clone_id); } return running; } gboolean native_create_probe(resource_t * rsc, node_t * node, action_t * complete, gboolean force, pe_working_set_t * data_set) { char *key = NULL; action_t *probe = NULL; node_t *running = NULL; node_t *allowed = NULL; resource_t *top = uber_parent(rsc); static const char *rc_master = NULL; static const char *rc_inactive = NULL; if (rc_inactive == NULL) { rc_inactive = crm_itoa(PCMK_OCF_NOT_RUNNING); rc_master = crm_itoa(PCMK_OCF_RUNNING_MASTER); } CRM_CHECK(node != NULL, return FALSE); if (force == FALSE && is_not_set(data_set->flags, pe_flag_startup_probes)) { pe_rsc_trace(rsc, "Skipping active resource detection for %s", rsc->id); return FALSE; } else if (force == FALSE && is_container_remote_node(node)) { pe_rsc_trace(rsc, "Skipping active resource detection for %s on container %s", rsc->id, node->details->id); return FALSE; } if (is_remote_node(node)) { const char *class = crm_element_value(rsc->xml, XML_AGENT_ATTR_CLASS); if (safe_str_eq(class, "stonith")) { pe_rsc_trace(rsc, "Skipping probe for %s on node %s, remote-nodes do not run stonith agents.", rsc->id, node->details->id); return FALSE; } else if (rsc_contains_remote_node(data_set, rsc)) { pe_rsc_trace(rsc, "Skipping probe for %s on node %s, remote-nodes can not run resources that contain connection resources.", rsc->id, node->details->id); return FALSE; } else if (rsc->is_remote_node) { pe_rsc_trace(rsc, "Skipping probe for %s on node %s, remote-nodes can not run connection resources", rsc->id, node->details->id); return FALSE; } } if (rsc->children) { GListPtr gIter = NULL; gboolean any_created = FALSE; for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; any_created = child_rsc->cmds->create_probe(child_rsc, node, complete, force, data_set) || any_created; } return any_created; } else if (rsc->container) { pe_rsc_trace(rsc, "Skipping %s: it is within container %s", rsc->id, rsc->container->id); return FALSE; } if (is_set(rsc->flags, pe_rsc_orphan)) { pe_rsc_trace(rsc, "Skipping orphan: %s", rsc->id); return FALSE; } running = g_hash_table_lookup(rsc->known_on, node->details->id); if (running == NULL && is_set(rsc->flags, pe_rsc_unique) == FALSE) { /* Anonymous clones */ if (rsc->parent == top) { running = g_hash_table_lookup(rsc->parent->known_on, node->details->id); } else { /* Grouped anonymous clones need extra special handling */ running = probe_grouped_clone(rsc, node, data_set); } } if (force == FALSE && running != NULL) { /* we already know the status of the resource on this node */ pe_rsc_trace(rsc, "Skipping active: %s on %s", rsc->id, node->details->uname); return FALSE; } allowed = g_hash_table_lookup(rsc->allowed_nodes, node->details->id); if (rsc->exclusive_discover || top->exclusive_discover) { if (allowed == NULL) { /* exclusive discover is enabled and this node is not in the allowed list. */ return FALSE; } else if (allowed->rsc_discover_mode != discover_exclusive) { /* exclusive discover is enabled and this node is not marked * as a node this resource should be discovered on */ return FALSE; } } if (allowed && allowed->rsc_discover_mode == discover_never) { /* this resource is marked as not needing to be discovered on this node */ return FALSE; } key = generate_op_key(rsc->id, RSC_STATUS, 0); probe = custom_action(rsc, key, RSC_STATUS, node, FALSE, TRUE, data_set); update_action_flags(probe, pe_action_optional | pe_action_clear); /* If enabled, require unfencing before probing any fence devices * but ensure it happens after any resources that require * unfencing have been probed. * * Doing it the other way (requiring unfencing after probing * resources that need it) would result in the node being * unfenced, and all its resources being stopped, whenever a new * resource is added. Which would be highly suboptimal. * * So essentially, at the point the fencing device(s) have been * probed, we know the state of all resources that require * unfencing and that unfencing occurred. */ if(is_set(rsc->flags, pe_rsc_fence_device) && is_set(data_set->flags, pe_flag_enable_unfencing)) { trigger_unfencing(NULL, node, "node discovery", probe, data_set); probe->priority = INFINITY; /* Ensure this runs if unfencing succeeds */ } else if(is_set(rsc->flags, pe_rsc_needs_unfencing)) { action_t *unfence = pe_fence_op(node, "on", TRUE, data_set); order_actions(probe, unfence, pe_order_optional); } /* * We need to know if it's running_on (not just known_on) this node * to correctly determine the target rc. */ running = pe_find_node_id(rsc->running_on, node->details->id); if (running == NULL) { add_hash_param(probe->meta, XML_ATTR_TE_TARGET_RC, rc_inactive); } else if (rsc->role == RSC_ROLE_MASTER) { add_hash_param(probe->meta, XML_ATTR_TE_TARGET_RC, rc_master); } pe_rsc_debug(rsc, "Probing %s on %s (%s)", rsc->id, node->details->uname, role2text(rsc->role)); if(is_set(rsc->flags, pe_rsc_fence_device) && is_set(data_set->flags, pe_flag_enable_unfencing)) { /* Normally rsc.start depends on probe complete which depends * on rsc.probe. But this can't be the case in this scenario as * it would create graph loops. * * So instead we explicitly order 'rsc.probe then rsc.start' */ custom_action_order(rsc, NULL, probe, rsc, generate_op_key(rsc->id, RSC_START, 0), NULL, pe_order_optional, data_set); } else { order_actions(probe, complete, pe_order_implies_then); } return TRUE; } static void native_start_constraints(resource_t * rsc, action_t * stonith_op, gboolean is_stonith, pe_working_set_t * data_set) { node_t *target = stonith_op ? stonith_op->node : NULL; GListPtr gIter = NULL; action_t *all_stopped = get_pseudo_op(ALL_STOPPED, data_set); action_t *stonith_done = get_pseudo_op(STONITH_DONE, data_set); for (gIter = rsc->actions; gIter != NULL; gIter = gIter->next) { action_t *action = (action_t *) gIter->data; if(action->needs == rsc_req_nothing) { } else if (action->needs == rsc_req_stonith) { order_actions(stonith_done, action, pe_order_optional); } else if (target != NULL && safe_str_eq(action->task, RSC_START) && NULL == pe_hash_table_lookup(rsc->known_on, target->details->id)) { /* if known == NULL, then we dont know if * the resource is active on the node * we're about to shoot * * in this case, regardless of action->needs, * the only safe option is to wait until * the node is shot before doing anything * to with the resource * * its analogous to waiting for all the probes * for rscX to complete before starting rscX * * the most likely explaination is that the * DC died and took its status with it */ pe_rsc_debug(rsc, "Ordering %s after %s recovery", action->uuid, target->details->uname); order_actions(all_stopped, action, pe_order_optional | pe_order_runnable_left); } } } static GListPtr find_fence_target_node_actions(GListPtr search_list, const char *key, node_t *fence_target, pe_working_set_t *data_set) { GListPtr gIter = NULL; GListPtr result_list = find_actions(search_list, key, fence_target); /* find stop actions for this rsc on any container nodes running on * the fencing target node */ for (gIter = fence_target->details->running_rsc; gIter != NULL; gIter = gIter->next) { GListPtr iter = NULL; GListPtr tmp_list = NULL; resource_t *tmp_rsc = (resource_t *) gIter->data; node_t *container_node = NULL; /* found a container node that lives on the host node * that is getting fenced. Find stop for our rsc that live on * the container node as well. These stop operations are also * implied by fencing of the host cluster node. */ if (tmp_rsc->is_remote_node && tmp_rsc->container != NULL) { container_node = pe_find_node(data_set->nodes, tmp_rsc->id); } if (container_node) { tmp_list = find_actions(search_list, key, container_node); } for (iter = tmp_list; iter != NULL; iter = iter->next) { result_list = g_list_prepend(result_list, (action_t *) iter->data); } g_list_free(tmp_list); } return result_list; } static void native_stop_constraints(resource_t * rsc, action_t * stonith_op, gboolean is_stonith, pe_working_set_t * data_set) { char *key = NULL; GListPtr gIter = NULL; GListPtr action_list = NULL; resource_t *top = uber_parent(rsc); key = stop_key(rsc); action_list = find_fence_target_node_actions(rsc->actions, key, stonith_op->node, data_set); free(key); /* add the stonith OP as a stop pre-req and the mark the stop * as a pseudo op - since its now redundant */ for (gIter = action_list; gIter != NULL; gIter = gIter->next) { action_t *action = (action_t *) gIter->data; if (action->node->details->online && action->node->details->unclean == FALSE && is_set(rsc->flags, pe_rsc_failed)) { continue; } if (is_set(rsc->flags, pe_rsc_failed)) { crm_notice("Stop of failed resource %s is" " implicit after %s is fenced", rsc->id, action->node->details->uname); } else { crm_info("%s is implicit after %s is fenced", action->uuid, action->node->details->uname); } /* the stop would never complete and is * now implied by the stonith operation */ update_action_flags(action, pe_action_pseudo); update_action_flags(action, pe_action_runnable); update_action_flags(action, pe_action_implied_by_stonith); { enum pe_ordering flags = pe_order_optional; action_t *parent_stop = find_first_action(top->actions, NULL, RSC_STOP, NULL); if(stonith_op->node->details->remote_rsc) { flags |= pe_order_preserve; } order_actions(stonith_op, action, flags); order_actions(stonith_op, parent_stop, flags); } if (is_set(rsc->flags, pe_rsc_notify)) { /* Create a second notification that will be delivered * immediately after the node is fenced * * Basic problem: * - C is a clone active on the node to be shot and stopping on another * - R is a resource that depends on C * * + C.stop depends on R.stop * + C.stopped depends on STONITH * + C.notify depends on C.stopped * + C.healthy depends on C.notify * + R.stop depends on C.healthy * * The extra notification here changes * + C.healthy depends on C.notify * into: * + C.healthy depends on C.notify' * + C.notify' depends on STONITH' * thus breaking the loop */ notify_data_t *n_data = create_notification_boundaries(rsc, RSC_STOP, NULL, stonith_op, data_set); crm_info("Creating secondary notification for %s", action->uuid); collect_notification_data(rsc, TRUE, FALSE, n_data); g_hash_table_insert(n_data->keys, strdup("notify_stop_resource"), strdup(rsc->id)); g_hash_table_insert(n_data->keys, strdup("notify_stop_uname"), strdup(action->node->details->uname)); create_notifications(uber_parent(rsc), n_data, data_set); free_notification_data(n_data); } /* From Bug #1601, successful fencing must be an input to a failed resources stop action. However given group(rA, rB) running on nodeX and B.stop has failed, A := stop healthy resource (rA.stop) B := stop failed resource (pseudo operation B.stop) C := stonith nodeX A requires B, B requires C, C requires A This loop would prevent the cluster from making progress. This block creates the "C requires A" dependency and therefore must (at least for now) be disabled. Instead, run the block above and treat all resources on nodeX as B would be (marked as a pseudo op depending on the STONITH). TODO: Break the "A requires B" dependency in update_action() and re-enable this block } else if(is_stonith == FALSE) { crm_info("Moving healthy resource %s" " off %s before fencing", rsc->id, node->details->uname); * stop healthy resources before the * stonith op * custom_action_order( rsc, stop_key(rsc), NULL, NULL,strdup(CRM_OP_FENCE),stonith_op, pe_order_optional, data_set); */ } g_list_free(action_list); key = demote_key(rsc); action_list = find_fence_target_node_actions(rsc->actions, key, stonith_op->node, data_set); free(key); for (gIter = action_list; gIter != NULL; gIter = gIter->next) { action_t *action = (action_t *) gIter->data; if (action->node->details->online == FALSE || action->node->details->unclean == TRUE || is_set(rsc->flags, pe_rsc_failed)) { if (is_set(rsc->flags, pe_rsc_failed)) { pe_rsc_info(rsc, "Demote of failed resource %s is" " implict after %s is fenced", rsc->id, action->node->details->uname); } else { pe_rsc_info(rsc, "%s is implicit after %s is fenced", action->uuid, action->node->details->uname); } /* the stop would never complete and is * now implied by the stonith operation */ crm_trace("here - 1"); update_action_flags(action, pe_action_pseudo); update_action_flags(action, pe_action_runnable); if (is_stonith == FALSE) { order_actions(stonith_op, action, pe_order_preserve|pe_order_optional); } } } g_list_free(action_list); } void rsc_stonith_ordering(resource_t * rsc, action_t * stonith_op, pe_working_set_t * data_set) { gboolean is_stonith = FALSE; if (rsc->children) { GListPtr gIter = NULL; for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; rsc_stonith_ordering(child_rsc, stonith_op, data_set); } return; } if (is_not_set(rsc->flags, pe_rsc_managed)) { pe_rsc_trace(rsc, "Skipping fencing constraints for unmanaged resource: %s", rsc->id); return; } /* Start constraints */ native_start_constraints(rsc, stonith_op, is_stonith, data_set); /* Stop constraints */ if (stonith_op) { native_stop_constraints(rsc, stonith_op, is_stonith, data_set); } } enum stack_activity { stack_stable = 0, stack_starting = 1, stack_stopping = 2, stack_middle = 4, }; static action_t * get_first_named_action(resource_t * rsc, const char *action, gboolean only_valid, node_t * current) { action_t *a = NULL; GListPtr action_list = NULL; char *key = generate_op_key(rsc->id, action, 0); action_list = find_actions(rsc->actions, key, current); if (action_list == NULL || action_list->data == NULL) { crm_trace("%s: no %s action", rsc->id, action); free(key); return NULL; } a = action_list->data; g_list_free(action_list); if (only_valid && is_set(a->flags, pe_action_pseudo)) { crm_trace("%s: pseudo", key); a = NULL; } else if (only_valid && is_not_set(a->flags, pe_action_runnable)) { crm_trace("%s: runnable", key); a = NULL; } free(key); return a; } static void ReloadRsc(resource_t * rsc, action_t * stop, action_t * start, pe_working_set_t * data_set) { action_t *action = NULL; action_t *rewrite = NULL; if (is_not_set(rsc->flags, pe_rsc_try_reload)) { return; } else if (is_not_set(stop->flags, pe_action_optional)) { pe_rsc_trace(rsc, "%s: stop action", rsc->id); return; } else if (is_not_set(start->flags, pe_action_optional)) { pe_rsc_trace(rsc, "%s: start action", rsc->id); return; } pe_rsc_trace(rsc, "%s on %s", rsc->id, stop->node->details->uname); action = get_first_named_action(rsc, RSC_PROMOTE, TRUE, NULL); if (action && is_set(action->flags, pe_action_optional) == FALSE) { update_action_flags(action, pe_action_pseudo); } action = get_first_named_action(rsc, RSC_DEMOTE, TRUE, NULL); if (action && is_set(action->flags, pe_action_optional) == FALSE) { rewrite = action; update_action_flags(stop, pe_action_pseudo); } else { rewrite = start; } pe_rsc_info(rsc, "Rewriting %s of %s on %s as a reload", rewrite->task, rsc->id, stop->node->details->uname); set_bit(rsc->flags, pe_rsc_reload); update_action_flags(rewrite, pe_action_optional | pe_action_clear); free(rewrite->uuid); free(rewrite->task); rewrite->task = strdup("reload"); rewrite->uuid = generate_op_key(rsc->id, rewrite->task, 0); } void rsc_reload(resource_t * rsc, pe_working_set_t * data_set) { GListPtr gIter = NULL; action_t *stop = NULL; action_t *start = NULL; if(is_set(rsc->flags, pe_rsc_munging)) { return; } set_bit(rsc->flags, pe_rsc_munging); if (rsc->children) { for (gIter = rsc->children; gIter != NULL; gIter = gIter->next) { resource_t *child_rsc = (resource_t *) gIter->data; rsc_reload(child_rsc, data_set); } return; } else if (rsc->variant > pe_native) { return; } pe_rsc_trace(rsc, "Processing %s", rsc->id); stop = get_first_named_action(rsc, RSC_STOP, TRUE, rsc->running_on ? rsc->running_on->data : NULL); start = get_first_named_action(rsc, RSC_START, TRUE, NULL); if (is_not_set(rsc->flags, pe_rsc_managed) || is_set(rsc->flags, pe_rsc_failed) || is_set(rsc->flags, pe_rsc_start_pending) || rsc->next_role < RSC_ROLE_STARTED) { pe_rsc_trace(rsc, "%s: general resource state: flags=0x%.16llx", rsc->id, rsc->flags); return; } if (stop != NULL && is_set(stop->flags, pe_action_optional) && is_set(rsc->flags, pe_rsc_try_reload)) { ReloadRsc(rsc, stop, start, data_set); } } void native_append_meta(resource_t * rsc, xmlNode * xml) { char *value = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_INCARNATION); - resource_t *parent, *last_parent; + resource_t *iso_parent, *last_parent; if (value) { char *name = NULL; name = crm_meta_name(XML_RSC_ATTR_INCARNATION); crm_xml_add(xml, name, value); free(name); } value = g_hash_table_lookup(rsc->meta, XML_RSC_ATTR_REMOTE_NODE); if (value) { char *name = NULL; name = crm_meta_name(XML_RSC_ATTR_REMOTE_NODE); crm_xml_add(xml, name, value); free(name); } - last_parent = parent = rsc; - while (parent != NULL) { + last_parent = iso_parent = rsc; + while (iso_parent != NULL) { char *name = NULL; + char *iso = NULL; - if (parent->isolation_wrapper == NULL) { - last_parent = parent; - parent = parent->parent; + if (iso_parent->isolation_wrapper == NULL) { + last_parent = iso_parent; + iso_parent = iso_parent->parent; continue; } /* name of wrapper script this resource is routed through. */ name = crm_meta_name(XML_RSC_ATTR_ISOLATION_WRAPPER); - crm_xml_add(xml, name, parent->isolation_wrapper); + crm_xml_add(xml, name, iso_parent->isolation_wrapper); free(name); /* instance name for isolated environment */ name = crm_meta_name(XML_RSC_ATTR_ISOLATION_INSTANCE); - if (parent->variant < pe_clone) { - crm_xml_add(xml, name, parent->id); - } else { - char *iso = NULL; + if (iso_parent->variant >= pe_clone) { /* if isolation is set at the clone/master level, we have to * give this resource the unique isolation instance associated - * with the matching clone child */ + * with the clone child (last_parent)*/ + + /* Example: cloned group. group is container + * clone myclone - iso_parent + * group mygroup - last_parent (this is the iso environment) + * rsc myrsc1 - rsc + * rsc myrsc2 + * The group is what is isolated in example1. We have to make + * sure myrsc1 and myrsc2 launch in the same isolated environment. + * + * Example: cloned primitives. rsc primitive is container + * clone myclone iso_parent + * rsc myrsc1 - last_parent == rsc (this is the iso environment) + * The individual cloned primitive instances are isolated + */ value = g_hash_table_lookup(last_parent->meta, XML_RSC_ATTR_INCARNATION); CRM_ASSERT(value != NULL); - iso = crm_concat(parent->id, value, '_'); + iso = crm_concat(crm_element_value(last_parent->xml, XML_ATTR_ID), value, '_'); crm_xml_add(xml, name, iso); free(iso); + } else { + /* + * Example: cloned group of containers + * clone myclone + * group mygroup + * rsc myrsc1 - iso_parent (this is the iso environment) + * rsc myrsc2 + * + * Example: group of containers + * group mygroup + * rsc myrsc1 - iso_parent (this is the iso environment) + * rsc myrsc2 + * + * Example: group is container + * group mygroup - iso_parent ( this is iso environment) + * rsc myrsc1 + * rsc myrsc2 + * + * Example: single primitive + * rsc myrsc1 - iso_parent (this is the iso environment) + */ + value = g_hash_table_lookup(iso_parent->meta, XML_RSC_ATTR_INCARNATION); + if (value) { + crm_xml_add(xml, name, iso_parent->id); + iso = crm_concat(crm_element_value(iso_parent->xml, XML_ATTR_ID), value, '_'); + crm_xml_add(xml, name, iso); + free(iso); + } else { + crm_xml_add(xml, name, iso_parent->id); + } } free(name); break; } } diff --git a/pengine/regression.sh b/pengine/regression.sh index 59e511e488..52971cdce3 100755 --- a/pengine/regression.sh +++ b/pengine/regression.sh @@ -1,806 +1,807 @@ #!/bin/bash # Copyright (C) 2004 Andrew Beekhof # # This program is free software; you can redistribute it and/or # modify it under the terms of the GNU General Public # License as published by the Free Software Foundation; either # version 2 of the License, or (at your option) any later version. # # This software is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU # General Public License for more details. # # You should have received a copy of the GNU General Public # License along with this library; if not, write to the Free Software # Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA # core=`dirname $0` . $core/regression.core.sh || exit 1 create_mode="true" info Generating test outputs for these tests... # do_test file description info Done. echo "" info Performing the following tests from $io_dir create_mode="false" echo "" do_test simple1 "Offline " do_test simple2 "Start " do_test simple3 "Start 2 " do_test simple4 "Start Failed" do_test simple6 "Stop Start " do_test simple7 "Shutdown " #do_test simple8 "Stonith " #do_test simple9 "Lower version" #do_test simple10 "Higher version" do_test simple11 "Priority (ne)" do_test simple12 "Priority (eq)" do_test simple8 "Stickiness" echo "" do_test group1 "Group " do_test group2 "Group + Native " do_test group3 "Group + Group " do_test group4 "Group + Native (nothing)" do_test group5 "Group + Native (move) " do_test group6 "Group + Group (move) " do_test group7 "Group colocation" do_test group13 "Group colocation (cant run)" do_test group8 "Group anti-colocation" do_test group9 "Group recovery" do_test group10 "Group partial recovery" do_test group11 "Group target_role" do_test group14 "Group stop (graph terminated)" do_test group15 "-ve group colocation" do_test bug-1573 "Partial stop of a group with two children" do_test bug-1718 "Mandatory group ordering - Stop group_FUN" do_test bug-lf-2613 "Move group on failure" do_test bug-lf-2619 "Move group on clone failure" do_test group-fail "Ensure stop order is preserved for partially active groups" do_test group-unmanaged "No need to restart r115 because r114 is unmanaged" do_test group-unmanaged-stopped "Make sure r115 is stopped when r114 fails" do_test group-dependants "Account for the location preferences of things colocated with a group" echo "" do_test rsc_dep1 "Must not " do_test rsc_dep3 "Must " do_test rsc_dep5 "Must not 3 " do_test rsc_dep7 "Must 3 " do_test rsc_dep10 "Must (but cant)" do_test rsc_dep2 "Must (running) " do_test rsc_dep8 "Must (running : alt) " do_test rsc_dep4 "Must (running + move)" do_test asymmetric "Asymmetric - require explicit location constraints" echo "" do_test orphan-0 "Orphan ignore" do_test orphan-1 "Orphan stop" do_test orphan-2 "Orphan stop, remove failcount" echo "" do_test params-0 "Params: No change" do_test params-1 "Params: Changed" do_test params-2 "Params: Resource definition" do_test params-4 "Params: Reload" do_test params-5 "Params: Restart based on probe digest" do_test novell-251689 "Resource definition change + target_role=stopped" do_test bug-lf-2106 "Restart all anonymous clone instances after config change" do_test params-6 "Params: Detect reload in previously migrated resource" do_test nvpair-id-ref "Support id-ref in nvpair with optional name" do_test not-reschedule-unneeded-monitor "Do not reschedule unneeded monitors while resource definitions have changed" echo "" do_test target-0 "Target Role : baseline" do_test target-1 "Target Role : master" do_test target-2 "Target Role : invalid" echo "" do_test base-score "Set a node's default score for all nodes" echo "" do_test date-1 "Dates" -t "2005-020" do_test date-2 "Date Spec - Pass" -t "2005-020T12:30" do_test date-3 "Date Spec - Fail" -t "2005-020T11:30" do_test origin "Timing of recurring operations" -t "2014-05-07 00:28:00" do_test probe-0 "Probe (anon clone)" do_test probe-1 "Pending Probe" do_test probe-2 "Correctly re-probe cloned groups" do_test probe-3 "Probe (pending node)" do_test probe-4 "Probe (pending node + stopped resource)" --rc 4 do_test standby "Standby" do_test comments "Comments" echo "" do_test one-or-more-0 "Everything starts" do_test one-or-more-1 "Nothing starts because of A" do_test one-or-more-2 "D can start because of C" do_test one-or-more-3 "D cannot start because of B and C" do_test one-or-more-4 "D cannot start because of target-role" do_test one-or-more-5 "Start A and F even though C and D are stopped" do_test one-or-more-6 "Leave A running even though B is stopped" do_test one-or-more-7 "Leave A running even though C is stopped" do_test bug-5140-require-all-false "Allow basegrp:0 to stop" do_test clone-require-all-1 "clone B starts node 3 and 4" do_test clone-require-all-2 "clone B remains stopped everywhere" do_test clone-require-all-3 "clone B stops everywhere because A stops everywhere" do_test clone-require-all-4 "clone B remains on node 3 and 4 with only one instance of A remaining." do_test clone-require-all-5 "clone B starts on node 1 3 and 4" do_test clone-require-all-6 "clone B remains active after shutting down instances of A" do_test clone-require-all-7 "clone A and B both start at the same time. all instances of A start before B." do_test clone-require-all-no-interleave-1 "C starts everywhere after A and B" do_test clone-require-all-no-interleave-2 "C starts on nodes 1, 2, and 4 with only one active instance of B" do_test clone-require-all-no-interleave-3 "C remains active when instance of B is stopped on one node and started on another." echo "" do_test order1 "Order start 1 " do_test order2 "Order start 2 " do_test order3 "Order stop " do_test order4 "Order (multiple) " do_test order5 "Order (move) " do_test order6 "Order (move w/ restart) " do_test order7 "Order (manditory) " do_test order-optional "Order (score=0) " do_test order-required "Order (score=INFINITY) " do_test bug-lf-2171 "Prevent group start when clone is stopped" do_test order-clone "Clone ordering should be able to prevent startup of dependant clones" do_test order-sets "Ordering for resource sets" do_test order-serialize "Serialize resources without inhibiting migration" do_test order-serialize-set "Serialize a set of resources without inhibiting migration" do_test clone-order-primitive "Order clone start after a primitive" do_test clone-order-16instances "Verify ordering of 16 cloned resources" do_test order-optional-keyword "Order (optional keyword)" do_test order-mandatory "Order (mandatory keyword)" do_test bug-lf-2493 "Don't imply colocation requirements when applying ordering constraints with clones" do_test ordered-set-basic-startup "Constraint set with default order settings." do_test ordered-set-natural "Allow natural set ordering" do_test order-wrong-kind "Order (error)" echo "" do_test coloc-loop "Colocation - loop" do_test coloc-many-one "Colocation - many-to-one" do_test coloc-list "Colocation - many-to-one with list" do_test coloc-group "Colocation - groups" do_test coloc-slave-anti "Anti-colocation with slave shouldn't prevent master colocation" do_test coloc-attr "Colocation based on node attributes" do_test coloc-negative-group "Negative colocation with a group" do_test coloc-intra-set "Intra-set colocation" do_test bug-lf-2435 "Colocation sets with a negative score" do_test coloc-clone-stays-active "Ensure clones don't get stopped/demoted because a dependant must stop" do_test coloc_fp_logic "Verify floating point calculations in colocation are working" do_test colo_master_w_native "cl#5070 - Verify promotion order is affected when colocating master to native rsc." do_test colo_slave_w_native "cl#5070 - Verify promotion order is affected when colocating slave to native rsc." do_test anti-colocation-order "cl#5187 - Prevent resources in an anti-colocation from even temporarily running on a same node" do_test enforce-colo1 "Always enforce B with A INFINITY." do_test complex_enforce_colo "Always enforce B with A INFINITY. (make sure heat-engine stops)" echo "" do_test rsc-sets-seq-true "Resource Sets - sequential=false" do_test rsc-sets-seq-false "Resource Sets - sequential=true" do_test rsc-sets-clone "Resource Sets - Clone" do_test rsc-sets-master "Resource Sets - Master" do_test rsc-sets-clone-1 "Resource Sets - Clone (lf#2404)" #echo "" #do_test agent1 "version: lt (empty)" #do_test agent2 "version: eq " #do_test agent3 "version: gt " echo "" do_test attrs1 "string: eq (and) " do_test attrs2 "string: lt / gt (and)" do_test attrs3 "string: ne (or) " do_test attrs4 "string: exists " do_test attrs5 "string: not_exists " do_test attrs6 "is_dc: true " do_test attrs7 "is_dc: false " do_test attrs8 "score_attribute " do_test per-node-attrs "Per node resource parameters" echo "" do_test mon-rsc-1 "Schedule Monitor - start" do_test mon-rsc-2 "Schedule Monitor - move " do_test mon-rsc-3 "Schedule Monitor - pending start " do_test mon-rsc-4 "Schedule Monitor - move/pending start" echo "" do_test rec-rsc-0 "Resource Recover - no start " do_test rec-rsc-1 "Resource Recover - start " do_test rec-rsc-2 "Resource Recover - monitor " do_test rec-rsc-3 "Resource Recover - stop - ignore" do_test rec-rsc-4 "Resource Recover - stop - block " do_test rec-rsc-5 "Resource Recover - stop - fence " do_test rec-rsc-6 "Resource Recover - multiple - restart" do_test rec-rsc-7 "Resource Recover - multiple - stop " do_test rec-rsc-8 "Resource Recover - multiple - block " do_test rec-rsc-9 "Resource Recover - group/group" do_test monitor-recovery "on-fail=block + resource recovery detected by recurring monitor" do_test stop-failure-no-quorum "Stop failure without quorum" do_test stop-failure-no-fencing "Stop failure without fencing available" do_test stop-failure-with-fencing "Stop failure with fencing available" echo "" do_test quorum-1 "No quorum - ignore" do_test quorum-2 "No quorum - freeze" do_test quorum-3 "No quorum - stop " do_test quorum-4 "No quorum - start anyway" do_test quorum-5 "No quorum - start anyway (group)" do_test quorum-6 "No quorum - start anyway (clone)" do_test bug-cl-5212 "No promotion with no-quorum-policy=freeze" echo "" do_test rec-node-1 "Node Recover - Startup - no fence" do_test rec-node-2 "Node Recover - Startup - fence " do_test rec-node-3 "Node Recover - HA down - no fence" do_test rec-node-4 "Node Recover - HA down - fence " do_test rec-node-5 "Node Recover - CRM down - no fence" do_test rec-node-6 "Node Recover - CRM down - fence " do_test rec-node-7 "Node Recover - no quorum - ignore " do_test rec-node-8 "Node Recover - no quorum - freeze " do_test rec-node-9 "Node Recover - no quorum - stop " do_test rec-node-10 "Node Recover - no quorum - stop w/fence" do_test rec-node-11 "Node Recover - CRM down w/ group - fence " do_test rec-node-12 "Node Recover - nothing active - fence " do_test rec-node-13 "Node Recover - failed resource + shutdown - fence " do_test rec-node-15 "Node Recover - unknown lrm section" do_test rec-node-14 "Serialize all stonith's" echo "" do_test multi1 "Multiple Active (stop/start)" echo "" do_test migrate-begin "Normal migration" do_test migrate-success "Completed migration" do_test migrate-partial-1 "Completed migration, missing stop on source" do_test migrate-partial-2 "Successful migrate_to only" do_test migrate-partial-3 "Successful migrate_to only, target down" do_test migrate-partial-4 "Migrate from the correct host after migrate_to+migrate_from" do_test bug-5186-partial-migrate "Handle partial migration when src node loses membership" do_test migrate-fail-2 "Failed migrate_from" do_test migrate-fail-3 "Failed migrate_from + stop on source" do_test migrate-fail-4 "Failed migrate_from + stop on target - ideally we wouldn't need to re-stop on target" do_test migrate-fail-5 "Failed migrate_from + stop on source and target" do_test migrate-fail-6 "Failed migrate_to" do_test migrate-fail-7 "Failed migrate_to + stop on source" do_test migrate-fail-8 "Failed migrate_to + stop on target - ideally we wouldn't need to re-stop on target" do_test migrate-fail-9 "Failed migrate_to + stop on source and target" do_test migrate-stop "Migration in a stopping stack" do_test migrate-start "Migration in a starting stack" do_test migrate-stop_start "Migration in a restarting stack" do_test migrate-stop-complex "Migration in a complex stopping stack" do_test migrate-start-complex "Migration in a complex starting stack" do_test migrate-stop-start-complex "Migration in a complex moving stack" do_test migrate-shutdown "Order the post-migration 'stop' before node shutdown" do_test migrate-1 "Migrate (migrate)" do_test migrate-2 "Migrate (stable)" do_test migrate-3 "Migrate (failed migrate_to)" do_test migrate-4 "Migrate (failed migrate_from)" do_test novell-252693 "Migration in a stopping stack" do_test novell-252693-2 "Migration in a starting stack" do_test novell-252693-3 "Non-Migration in a starting and stopping stack" do_test bug-1820 "Migration in a group" do_test bug-1820-1 "Non-migration in a group" do_test migrate-5 "Primitive migration with a clone" do_test migrate-fencing "Migration after Fencing" do_test migrate-both-vms "Migrate two VMs that have no colocation" do_test 1-a-then-bm-move-b "Advanced migrate logic. A then B. migrate B." do_test 2-am-then-b-move-a "Advanced migrate logic, A then B, migrate A without stopping B" do_test 3-am-then-bm-both-migrate "Advanced migrate logic. A then B. migrate both" do_test 4-am-then-bm-b-not-migratable "Advanced migrate logic, A then B, B not migratable" do_test 5-am-then-bm-a-not-migratable "Advanced migrate logic. A then B. move both, a not migratable" do_test 6-migrate-group "Advanced migrate logic, migrate a group" do_test 7-migrate-group-one-unmigratable "Advanced migrate logic, migrate group mixed with allow-migrate true/false" do_test 8-am-then-bm-a-migrating-b-stopping "Advanced migrate logic, A then B, A migrating, B stopping" do_test 9-am-then-bm-b-migrating-a-stopping "Advanced migrate logic, A then B, B migrate, A stopping" do_test 10-a-then-bm-b-move-a-clone "Advanced migrate logic, A clone then B, migrate B while stopping A" do_test 11-a-then-bm-b-move-a-clone-starting "Advanced migrate logic, A clone then B, B moving while A is start/stopping" do_test a-promote-then-b-migrate "A promote then B start. migrate B" do_test a-demote-then-b-migrate "A demote then B stop. migrate B" #echo "" #do_test complex1 "Complex " do_test bug-lf-2422 "Dependancy on partially active group - stop ocfs:*" echo "" do_test clone-anon-probe-1 "Probe the correct (anonymous) clone instance for each node" do_test clone-anon-probe-2 "Avoid needless re-probing of anonymous clones" do_test clone-anon-failcount "Merge failcounts for anonymous clones" do_test inc0 "Incarnation start" do_test inc1 "Incarnation start order" do_test inc2 "Incarnation silent restart, stop, move" do_test inc3 "Inter-incarnation ordering, silent restart, stop, move" do_test inc4 "Inter-incarnation ordering, silent restart, stop, move (ordered)" do_test inc5 "Inter-incarnation ordering, silent restart, stop, move (restart 1)" do_test inc6 "Inter-incarnation ordering, silent restart, stop, move (restart 2)" do_test inc7 "Clone colocation" do_test inc8 "Clone anti-colocation" do_test inc9 "Non-unique clone" do_test inc10 "Non-unique clone (stop)" do_test inc11 "Primitive colocation with clones" do_test inc12 "Clone shutdown" do_test cloned-group "Make sure only the correct number of cloned groups are started" do_test cloned-group-stop "Ensure stopping qpidd also stops glance and cinder" do_test clone-no-shuffle "Dont prioritize allocation of instances that must be moved" do_test clone-max-zero "Orphan processing with clone-max=0" do_test clone-anon-dup "Bug LF#2087 - Correctly parse the state of anonymous clones that are active more than once per node" do_test bug-lf-2160 "Dont shuffle clones due to colocation" do_test bug-lf-2213 "clone-node-max enforcement for cloned groups" do_test bug-lf-2153 "Clone ordering constraints" do_test bug-lf-2361 "Ensure clones observe mandatory ordering constraints if the LHS is unrunnable" do_test bug-lf-2317 "Avoid needless restart of primitive depending on a clone" do_test clone-colocate-instance-1 "Colocation with a specific clone instance (negative example)" do_test clone-colocate-instance-2 "Colocation with a specific clone instance" do_test clone-order-instance "Ordering with specific clone instances" do_test bug-lf-2453 "Enforce mandatory clone ordering without colocation" do_test bug-lf-2508 "Correctly reconstruct the status of anonymous cloned groups" do_test bug-lf-2544 "Balanced clone placement" do_test bug-lf-2445 "Redistribute clones with node-max > 1 and stickiness = 0" do_test bug-lf-2574 "Avoid clone shuffle" do_test bug-lf-2581 "Avoid group restart due to unrelated clone (re)start" do_test bug-cl-5168 "Don't shuffle clones" do_test bug-cl-5170 "Prevent clone from starting with on-fail=block" do_test clone-fail-block-colocation "Move colocated group when failed clone has on-fail=block" do_test clone-interleave-1 "Clone-3 cannot start on pcmk-1 due to interleaved ordering (no colocation)" do_test clone-interleave-2 "Clone-3 must stop on pcmk-1 due to interleaved ordering (no colocation)" do_test clone-interleave-3 "Clone-3 must be recovered on pcmk-1 due to interleaved ordering (no colocation)" echo "" do_test unfence-startup "Clean unfencing" do_test unfence-definition "Unfencing when the agent changes" do_test unfence-parameters "Unfencing when the agent parameters changes" echo "" do_test master-0 "Stopped -> Slave" do_test master-1 "Stopped -> Promote" do_test master-2 "Stopped -> Promote : notify" do_test master-3 "Stopped -> Promote : master location" do_test master-4 "Started -> Promote : master location" do_test master-5 "Promoted -> Promoted" do_test master-6 "Promoted -> Promoted (2)" do_test master-7 "Promoted -> Fenced" do_test master-8 "Promoted -> Fenced -> Moved" do_test master-9 "Stopped + Promotable + No quorum" do_test master-10 "Stopped -> Promotable : notify with monitor" do_test master-11 "Stopped -> Promote : colocation" do_test novell-239082 "Demote/Promote ordering" do_test novell-239087 "Stable master placement" do_test master-12 "Promotion based solely on rsc_location constraints" do_test master-13 "Include preferences of colocated resources when placing master" do_test master-demote "Ordering when actions depends on demoting a slave resource" do_test master-ordering "Prevent resources from starting that need a master" do_test bug-1765 "Master-Master Colocation (dont stop the slaves)" do_test master-group "Promotion of cloned groups" do_test bug-lf-1852 "Don't shuffle master/slave instances unnecessarily" do_test master-failed-demote "Dont retry failed demote actions" do_test master-failed-demote-2 "Dont retry failed demote actions (notify=false)" do_test master-depend "Ensure resources that depend on the master don't get allocated until the master does" do_test master-reattach "Re-attach to a running master" do_test master-allow-start "Don't include master score if it would prevent allocation" do_test master-colocation "Allow master instances placemaker to be influenced by colocation constraints" do_test master-pseudo "Make sure promote/demote pseudo actions are created correctly" do_test master-role "Prevent target-role from promoting more than master-max instances" do_test bug-lf-2358 "Master-Master anti-colocation" do_test master-promotion-constraint "Mandatory master colocation constraints" do_test unmanaged-master "Ensure role is preserved for unmanaged resources" do_test master-unmanaged-monitor "Start the correct monitor operation for unmanaged masters" do_test master-demote-2 "Demote does not clear past failure" do_test master-move "Move master based on failure of colocated group" do_test master-probed-score "Observe the promotion score of probed resources" do_test colocation_constraint_stops_master "cl#5054 - Ensure master is demoted when stopped by colocation constraint" do_test colocation_constraint_stops_slave "cl#5054 - Ensure slave is not demoted when stopped by colocation constraint" do_test order_constraint_stops_master "cl#5054 - Ensure master is demoted when stopped by order constraint" do_test order_constraint_stops_slave "cl#5054 - Ensure slave is not demoted when stopped by order constraint" do_test master_monitor_restart "cl#5072 - Ensure master monitor operation will start after promotion." do_test bug-rh-880249 "Handle replacement of an m/s resource with a primitive" do_test bug-5143-ms-shuffle "Prevent master shuffling due to promotion score" do_test master-demote-block "Block promotion if demote fails with on-fail=block" do_test master-dependant-ban "Don't stop instances from being active because a dependant is banned from that host" do_test master-stop "Stop instances due to location constraint with role=Started" do_test master-partially-demoted-group "Allow partially demoted group to finish demoting" do_test bug-cl-5213 "Ensure role colocation with -INFINITY is enforced" do_test bug-cl-5219 "Allow unrelated resources with a common colocation target to remain promoted" do_test master-asymmetrical-order "Fix the behaviors of multi-state resources with asymmetrical ordering" do_test master-notify "Master promotion with notifies" echo "" do_test history-1 "Correctly parse stateful-1 resource state" echo "" do_test managed-0 "Managed (reference)" do_test managed-1 "Not managed - down " do_test managed-2 "Not managed - up " do_test bug-5028 "Shutdown should block if anything depends on an unmanaged resource" do_test bug-5028-detach "Ensure detach still works" do_test bug-5028-bottom "Ensure shutdown still blocks if the blocked resource is at the bottom of the stack" do_test unmanaged-stop-1 "cl#5155 - Block the stop of resources if any depending resource is unmanaged " do_test unmanaged-stop-2 "cl#5155 - Block the stop of resources if the first resource in a mandatory stop order is unmanaged " do_test unmanaged-stop-3 "cl#5155 - Block the stop of resources if any depending resource in a group is unmanaged " do_test unmanaged-stop-4 "cl#5155 - Block the stop of resources if any depending resource in the middle of a group is unmanaged " do_test unmanaged-block-restart "Block restart of resources if any dependent resource in a group is unmanaged" echo "" do_test interleave-0 "Interleave (reference)" do_test interleave-1 "coloc - not interleaved" do_test interleave-2 "coloc - interleaved " do_test interleave-3 "coloc - interleaved (2)" do_test interleave-pseudo-stop "Interleaved clone during stonith" do_test interleave-stop "Interleaved clone during stop" do_test interleave-restart "Interleaved clone during dependancy restart" echo "" do_test notify-0 "Notify reference" do_test notify-1 "Notify simple" do_test notify-2 "Notify simple, confirm" do_test notify-3 "Notify move, confirm" do_test novell-239079 "Notification priority" #do_test notify-2 "Notify - 764" echo "" do_test 594 "OSDL #594 - Unrunnable actions scheduled in transition" do_test 662 "OSDL #662 - Two resources start on one node when incarnation_node_max = 1" do_test 696 "OSDL #696 - CRM starts stonith RA without monitor" do_test 726 "OSDL #726 - Attempting to schedule rsc_posic041_monitor_5000 _after_ a stop" do_test 735 "OSDL #735 - Correctly detect that rsc_hadev1 is stopped on hadev3" do_test 764 "OSDL #764 - Missing monitor op for DoFencing:child_DoFencing:1" do_test 797 "OSDL #797 - Assert triggered: task_id_i > max_call_id" do_test 829 "OSDL #829" do_test 994 "OSDL #994 - Stopping the last resource in a resource group causes the entire group to be restarted" do_test 994-2 "OSDL #994 - with a dependant resource" do_test 1360 "OSDL #1360 - Clone stickiness" do_test 1484 "OSDL #1484 - on_fail=stop" do_test 1494 "OSDL #1494 - Clone stability" do_test unrunnable-1 "Unrunnable" do_test stonith-0 "Stonith loop - 1" do_test stonith-1 "Stonith loop - 2" do_test stonith-2 "Stonith loop - 3" do_test stonith-3 "Stonith startup" do_test stonith-4 "Stonith node state" --rc 4 do_test bug-1572-1 "Recovery of groups depending on master/slave" do_test bug-1572-2 "Recovery of groups depending on master/slave when the master is never re-promoted" do_test bug-1685 "Depends-on-master ordering" do_test bug-1822 "Dont promote partially active groups" do_test bug-pm-11 "New resource added to a m/s group" do_test bug-pm-12 "Recover only the failed portion of a cloned group" do_test bug-n-387749 "Don't shuffle clone instances" do_test bug-n-385265 "Don't ignore the failure stickiness of group children - resource_idvscommon should stay stopped" do_test bug-n-385265-2 "Ensure groups are migrated instead of remaining partially active on the current node" do_test bug-lf-1920 "Correctly handle probes that find active resources" do_test bnc-515172 "Location constraint with multiple expressions" do_test colocate-primitive-with-clone "Optional colocation with a clone" do_test use-after-free-merge "Use-after-free in native_merge_weights" do_test bug-lf-2551 "STONITH ordering for stop" do_test bug-lf-2606 "Stonith implies demote" do_test bug-lf-2474 "Ensure resource op timeout takes precedence over op_defaults" do_test bug-suse-707150 "Prevent vm-01 from starting due to colocation/ordering" do_test bug-5014-A-start-B-start "Verify when A starts B starts using symmetrical=false" do_test bug-5014-A-stop-B-started "Verify when A stops B does not stop if it has already started using symmetric=false" do_test bug-5014-A-stopped-B-stopped "Verify when A is stopped and B has not started, B does not start before A using symmetric=false" do_test bug-5014-CthenAthenB-C-stopped "Verify when C then A is symmetrical=true, A then B is symmetric=false, and C is stopped that nothing starts." do_test bug-5014-CLONE-A-start-B-start "Verify when A starts B starts using clone resources with symmetric=false" do_test bug-5014-CLONE-A-stop-B-started "Verify when A stops B does not stop if it has already started using clone resources with symmetric=false." do_test bug-5014-GROUP-A-start-B-start "Verify when A starts B starts when using group resources with symmetric=false." do_test bug-5014-GROUP-A-stopped-B-started "Verify when A stops B does not stop if it has already started using group resources with symmetric=false." do_test bug-5014-GROUP-A-stopped-B-stopped "Verify when A is stopped and B has not started, B does not start before A using group resources with symmetric=false." do_test bug-5014-ordered-set-symmetrical-false "Verify ordered sets work with symmetrical=false" do_test bug-5014-ordered-set-symmetrical-true "Verify ordered sets work with symmetrical=true" do_test bug-5007-masterslave_colocation "Verify use of colocation scores other than INFINITY and -INFINITY work on multi-state resources." do_test bug-5038 "Prevent restart of anonymous clones when clone-max decreases" do_test bug-5025-1 "Automatically clean up failcount after resource config change with reload" do_test bug-5025-2 "Make sure clear failcount action isn't set when config does not change." do_test bug-5025-3 "Automatically clean up failcount after resource config change with restart" do_test bug-5025-4 "Clear failcount when last failure is a start op and rsc attributes changed." do_test failcount "Ensure failcounts are correctly expired" do_test failcount-block "Ensure failcounts are not expired when on-fail=block is present" do_test monitor-onfail-restart "bug-5058 - Monitor failure with on-fail set to restart" do_test monitor-onfail-stop "bug-5058 - Monitor failure wiht on-fail set to stop" do_test bug-5059 "No need to restart p_stateful1:*" do_test bug-5069-op-enabled "Test on-fail=ignore with failure when monitor is enabled." do_test bug-5069-op-disabled "Test on-fail-ignore with failure when monitor is disabled." do_test obsolete-lrm-resource "cl#5115 - Do not use obsolete lrm_resource sections" do_test expire-non-blocked-failure "Ignore failure-timeout only if the failed operation has on-fail=block" do_test ignore_stonith_rsc_order1 "cl#5056- Ignore order constraint between stonith and non-stonith rsc." do_test ignore_stonith_rsc_order2 "cl#5056- Ignore order constraint with group rsc containing mixed stonith and non-stonith." do_test ignore_stonith_rsc_order3 "cl#5056- Ignore order constraint, stonith clone and mixed group" do_test ignore_stonith_rsc_order4 "cl#5056- Ignore order constraint, stonith clone and clone with nested mixed group" do_test honor_stonith_rsc_order1 "cl#5056- Honor order constraint, stonith clone and pure stonith group(single rsc)." do_test honor_stonith_rsc_order2 "cl#5056- Honor order constraint, stonith clone and pure stonith group(multiple rsc)" do_test honor_stonith_rsc_order3 "cl#5056- Honor order constraint, stonith clones with nested pure stonith group." do_test honor_stonith_rsc_order4 "cl#5056- Honor order constraint, between two native stonith rscs." do_test probe-timeout "cl#5099 - Default probe timeout" echo "" do_test systemhealth1 "System Health () #1" do_test systemhealth2 "System Health () #2" do_test systemhealth3 "System Health () #3" do_test systemhealthn1 "System Health (None) #1" do_test systemhealthn2 "System Health (None) #2" do_test systemhealthn3 "System Health (None) #3" do_test systemhealthm1 "System Health (Migrate On Red) #1" do_test systemhealthm2 "System Health (Migrate On Red) #2" do_test systemhealthm3 "System Health (Migrate On Red) #3" do_test systemhealtho1 "System Health (Only Green) #1" do_test systemhealtho2 "System Health (Only Green) #2" do_test systemhealtho3 "System Health (Only Green) #3" do_test systemhealthp1 "System Health (Progessive) #1" do_test systemhealthp2 "System Health (Progessive) #2" do_test systemhealthp3 "System Health (Progessive) #3" echo "" do_test utilization "Placement Strategy - utilization" do_test minimal "Placement Strategy - minimal" do_test balanced "Placement Strategy - balanced" echo "" do_test placement-stickiness "Optimized Placement Strategy - stickiness" do_test placement-priority "Optimized Placement Strategy - priority" do_test placement-location "Optimized Placement Strategy - location" do_test placement-capacity "Optimized Placement Strategy - capacity" echo "" do_test utilization-order1 "Utilization Order - Simple" do_test utilization-order2 "Utilization Order - Complex" do_test utilization-order3 "Utilization Order - Migrate" do_test utilization-order4 "Utilization Order - Live Mirgration (bnc#695440)" do_test utilization-shuffle "Don't displace prmExPostgreSQLDB2 on act2, Start prmExPostgreSQLDB1 on act3" do_test load-stopped-loop "Avoid transition loop due to load_stopped (cl#5044)" echo "" do_test reprobe-target_rc "Ensure correct target_rc for reprobe of inactive resources" do_test node-maintenance-1 "cl#5128 - Node maintenance" do_test node-maintenance-2 "cl#5128 - Node maintenance (coming out of maintenance mode)" do_test rsc-maintenance "Per-resource maintenance" echo "" do_test not-installed-agent "The resource agent is missing" do_test not-installed-tools "Something the resource agent needs is missing" echo "" do_test stopped-monitor-00 "Stopped Monitor - initial start" do_test stopped-monitor-01 "Stopped Monitor - failed started" do_test stopped-monitor-02 "Stopped Monitor - started multi-up" do_test stopped-monitor-03 "Stopped Monitor - stop started" do_test stopped-monitor-04 "Stopped Monitor - failed stop" do_test stopped-monitor-05 "Stopped Monitor - start unmanaged" do_test stopped-monitor-06 "Stopped Monitor - unmanaged multi-up" do_test stopped-monitor-07 "Stopped Monitor - start unmanaged multi-up" do_test stopped-monitor-08 "Stopped Monitor - migrate" do_test stopped-monitor-09 "Stopped Monitor - unmanage started" do_test stopped-monitor-10 "Stopped Monitor - unmanaged started multi-up" do_test stopped-monitor-11 "Stopped Monitor - stop unmanaged started" do_test stopped-monitor-12 "Stopped Monitor - unmanaged started multi-up (targer-role="Stopped")" do_test stopped-monitor-20 "Stopped Monitor - initial stop" do_test stopped-monitor-21 "Stopped Monitor - stopped single-up" do_test stopped-monitor-22 "Stopped Monitor - stopped multi-up" do_test stopped-monitor-23 "Stopped Monitor - start stopped" do_test stopped-monitor-24 "Stopped Monitor - unmanage stopped" do_test stopped-monitor-25 "Stopped Monitor - unmanaged stopped multi-up" do_test stopped-monitor-26 "Stopped Monitor - start unmanaged stopped" do_test stopped-monitor-27 "Stopped Monitor - unmanaged stopped multi-up (target-role="Started")" do_test stopped-monitor-30 "Stopped Monitor - new node started" do_test stopped-monitor-31 "Stopped Monitor - new node stopped" echo"" do_test ticket-primitive-1 "Ticket - Primitive (loss-policy=stop, initial)" do_test ticket-primitive-2 "Ticket - Primitive (loss-policy=stop, granted)" do_test ticket-primitive-3 "Ticket - Primitive (loss-policy-stop, revoked)" do_test ticket-primitive-4 "Ticket - Primitive (loss-policy=demote, initial)" do_test ticket-primitive-5 "Ticket - Primitive (loss-policy=demote, granted)" do_test ticket-primitive-6 "Ticket - Primitive (loss-policy=demote, revoked)" do_test ticket-primitive-7 "Ticket - Primitive (loss-policy=fence, initial)" do_test ticket-primitive-8 "Ticket - Primitive (loss-policy=fence, granted)" do_test ticket-primitive-9 "Ticket - Primitive (loss-policy=fence, revoked)" do_test ticket-primitive-10 "Ticket - Primitive (loss-policy=freeze, initial)" do_test ticket-primitive-11 "Ticket - Primitive (loss-policy=freeze, granted)" do_test ticket-primitive-12 "Ticket - Primitive (loss-policy=freeze, revoked)" do_test ticket-primitive-13 "Ticket - Primitive (loss-policy=stop, standby, granted)" do_test ticket-primitive-14 "Ticket - Primitive (loss-policy=stop, granted, standby)" do_test ticket-primitive-15 "Ticket - Primitive (loss-policy=stop, standby, revoked)" do_test ticket-primitive-16 "Ticket - Primitive (loss-policy=demote, standby, granted)" do_test ticket-primitive-17 "Ticket - Primitive (loss-policy=demote, granted, standby)" do_test ticket-primitive-18 "Ticket - Primitive (loss-policy=demote, standby, revoked)" do_test ticket-primitive-19 "Ticket - Primitive (loss-policy=fence, standby, granted)" do_test ticket-primitive-20 "Ticket - Primitive (loss-policy=fence, granted, standby)" do_test ticket-primitive-21 "Ticket - Primitive (loss-policy=fence, standby, revoked)" do_test ticket-primitive-22 "Ticket - Primitive (loss-policy=freeze, standby, granted)" do_test ticket-primitive-23 "Ticket - Primitive (loss-policy=freeze, granted, standby)" do_test ticket-primitive-24 "Ticket - Primitive (loss-policy=freeze, standby, revoked)" echo"" do_test ticket-group-1 "Ticket - Group (loss-policy=stop, initial)" do_test ticket-group-2 "Ticket - Group (loss-policy=stop, granted)" do_test ticket-group-3 "Ticket - Group (loss-policy-stop, revoked)" do_test ticket-group-4 "Ticket - Group (loss-policy=demote, initial)" do_test ticket-group-5 "Ticket - Group (loss-policy=demote, granted)" do_test ticket-group-6 "Ticket - Group (loss-policy=demote, revoked)" do_test ticket-group-7 "Ticket - Group (loss-policy=fence, initial)" do_test ticket-group-8 "Ticket - Group (loss-policy=fence, granted)" do_test ticket-group-9 "Ticket - Group (loss-policy=fence, revoked)" do_test ticket-group-10 "Ticket - Group (loss-policy=freeze, initial)" do_test ticket-group-11 "Ticket - Group (loss-policy=freeze, granted)" do_test ticket-group-12 "Ticket - Group (loss-policy=freeze, revoked)" do_test ticket-group-13 "Ticket - Group (loss-policy=stop, standby, granted)" do_test ticket-group-14 "Ticket - Group (loss-policy=stop, granted, standby)" do_test ticket-group-15 "Ticket - Group (loss-policy=stop, standby, revoked)" do_test ticket-group-16 "Ticket - Group (loss-policy=demote, standby, granted)" do_test ticket-group-17 "Ticket - Group (loss-policy=demote, granted, standby)" do_test ticket-group-18 "Ticket - Group (loss-policy=demote, standby, revoked)" do_test ticket-group-19 "Ticket - Group (loss-policy=fence, standby, granted)" do_test ticket-group-20 "Ticket - Group (loss-policy=fence, granted, standby)" do_test ticket-group-21 "Ticket - Group (loss-policy=fence, standby, revoked)" do_test ticket-group-22 "Ticket - Group (loss-policy=freeze, standby, granted)" do_test ticket-group-23 "Ticket - Group (loss-policy=freeze, granted, standby)" do_test ticket-group-24 "Ticket - Group (loss-policy=freeze, standby, revoked)" echo"" do_test ticket-clone-1 "Ticket - Clone (loss-policy=stop, initial)" do_test ticket-clone-2 "Ticket - Clone (loss-policy=stop, granted)" do_test ticket-clone-3 "Ticket - Clone (loss-policy-stop, revoked)" do_test ticket-clone-4 "Ticket - Clone (loss-policy=demote, initial)" do_test ticket-clone-5 "Ticket - Clone (loss-policy=demote, granted)" do_test ticket-clone-6 "Ticket - Clone (loss-policy=demote, revoked)" do_test ticket-clone-7 "Ticket - Clone (loss-policy=fence, initial)" do_test ticket-clone-8 "Ticket - Clone (loss-policy=fence, granted)" do_test ticket-clone-9 "Ticket - Clone (loss-policy=fence, revoked)" do_test ticket-clone-10 "Ticket - Clone (loss-policy=freeze, initial)" do_test ticket-clone-11 "Ticket - Clone (loss-policy=freeze, granted)" do_test ticket-clone-12 "Ticket - Clone (loss-policy=freeze, revoked)" do_test ticket-clone-13 "Ticket - Clone (loss-policy=stop, standby, granted)" do_test ticket-clone-14 "Ticket - Clone (loss-policy=stop, granted, standby)" do_test ticket-clone-15 "Ticket - Clone (loss-policy=stop, standby, revoked)" do_test ticket-clone-16 "Ticket - Clone (loss-policy=demote, standby, granted)" do_test ticket-clone-17 "Ticket - Clone (loss-policy=demote, granted, standby)" do_test ticket-clone-18 "Ticket - Clone (loss-policy=demote, standby, revoked)" do_test ticket-clone-19 "Ticket - Clone (loss-policy=fence, standby, granted)" do_test ticket-clone-20 "Ticket - Clone (loss-policy=fence, granted, standby)" do_test ticket-clone-21 "Ticket - Clone (loss-policy=fence, standby, revoked)" do_test ticket-clone-22 "Ticket - Clone (loss-policy=freeze, standby, granted)" do_test ticket-clone-23 "Ticket - Clone (loss-policy=freeze, granted, standby)" do_test ticket-clone-24 "Ticket - Clone (loss-policy=freeze, standby, revoked)" echo"" do_test ticket-master-1 "Ticket - Master (loss-policy=stop, initial)" do_test ticket-master-2 "Ticket - Master (loss-policy=stop, granted)" do_test ticket-master-3 "Ticket - Master (loss-policy-stop, revoked)" do_test ticket-master-4 "Ticket - Master (loss-policy=demote, initial)" do_test ticket-master-5 "Ticket - Master (loss-policy=demote, granted)" do_test ticket-master-6 "Ticket - Master (loss-policy=demote, revoked)" do_test ticket-master-7 "Ticket - Master (loss-policy=fence, initial)" do_test ticket-master-8 "Ticket - Master (loss-policy=fence, granted)" do_test ticket-master-9 "Ticket - Master (loss-policy=fence, revoked)" do_test ticket-master-10 "Ticket - Master (loss-policy=freeze, initial)" do_test ticket-master-11 "Ticket - Master (loss-policy=freeze, granted)" do_test ticket-master-12 "Ticket - Master (loss-policy=freeze, revoked)" do_test ticket-master-13 "Ticket - Master (loss-policy=stop, standby, granted)" do_test ticket-master-14 "Ticket - Master (loss-policy=stop, granted, standby)" do_test ticket-master-15 "Ticket - Master (loss-policy=stop, standby, revoked)" do_test ticket-master-16 "Ticket - Master (loss-policy=demote, standby, granted)" do_test ticket-master-17 "Ticket - Master (loss-policy=demote, granted, standby)" do_test ticket-master-18 "Ticket - Master (loss-policy=demote, standby, revoked)" do_test ticket-master-19 "Ticket - Master (loss-policy=fence, standby, granted)" do_test ticket-master-20 "Ticket - Master (loss-policy=fence, granted, standby)" do_test ticket-master-21 "Ticket - Master (loss-policy=fence, standby, revoked)" do_test ticket-master-22 "Ticket - Master (loss-policy=freeze, standby, granted)" do_test ticket-master-23 "Ticket - Master (loss-policy=freeze, granted, standby)" do_test ticket-master-24 "Ticket - Master (loss-policy=freeze, standby, revoked)" echo "" do_test ticket-rsc-sets-1 "Ticket - Resource sets (1 ticket, initial)" do_test ticket-rsc-sets-2 "Ticket - Resource sets (1 ticket, granted)" do_test ticket-rsc-sets-3 "Ticket - Resource sets (1 ticket, revoked)" do_test ticket-rsc-sets-4 "Ticket - Resource sets (2 tickets, initial)" do_test ticket-rsc-sets-5 "Ticket - Resource sets (2 tickets, granted)" do_test ticket-rsc-sets-6 "Ticket - Resource sets (2 tickets, granted)" do_test ticket-rsc-sets-7 "Ticket - Resource sets (2 tickets, revoked)" do_test ticket-rsc-sets-8 "Ticket - Resource sets (1 ticket, standby, granted)" do_test ticket-rsc-sets-9 "Ticket - Resource sets (1 ticket, granted, standby)" do_test ticket-rsc-sets-10 "Ticket - Resource sets (1 ticket, standby, revoked)" do_test ticket-rsc-sets-11 "Ticket - Resource sets (2 tickets, standby, granted)" do_test ticket-rsc-sets-12 "Ticket - Resource sets (2 tickets, standby, granted)" do_test ticket-rsc-sets-13 "Ticket - Resource sets (2 tickets, granted, standby)" do_test ticket-rsc-sets-14 "Ticket - Resource sets (2 tickets, standby, revoked)" do_test cluster-specific-params "Cluster-specific instance attributes based on rules" do_test site-specific-params "Site-specific instance attributes based on rules" echo "" do_test template-1 "Template - 1" do_test template-2 "Template - 2" do_test template-3 "Template - 3 (merge operations)" do_test template-coloc-1 "Template - Colocation 1" do_test template-coloc-2 "Template - Colocation 2" do_test template-coloc-3 "Template - Colocation 3" do_test template-order-1 "Template - Order 1" do_test template-order-2 "Template - Order 2" do_test template-order-3 "Template - Order 3" do_test template-ticket "Template - Ticket" do_test template-rsc-sets-1 "Template - Resource Sets 1" do_test template-rsc-sets-2 "Template - Resource Sets 2" do_test template-rsc-sets-3 "Template - Resource Sets 3" do_test template-rsc-sets-4 "Template - Resource Sets 4" do_test template-clone-primitive "Cloned primitive from template" do_test template-clone-group "Cloned group from template" do_test location-sets-templates "Resource sets and templates - Location" do_test tags-coloc-order-1 "Tags - Colocation and Order (Simple)" do_test tags-coloc-order-2 "Tags - Colocation and Order (Resource Sets with Templates)" do_test tags-location "Tags - Location" do_test tags-ticket "Tags - Ticket" echo "" do_test container-1 "Container - initial" do_test container-2 "Container - monitor failed" do_test container-3 "Container - stop failed" do_test container-4 "Container - reached migration-threshold" do_test container-group-1 "Container in group - initial" do_test container-group-2 "Container in group - monitor failed" do_test container-group-3 "Container in group - stop failed" do_test container-group-4 "Container in group - reached migration-threshold" do_test container-is-remote-node "Place resource within container when container is remote-node" do_test bug-rh-1097457 "Kill user defined container/contents ordering" echo "" do_test whitebox-fail1 "Fail whitebox container rsc." do_test whitebox-fail2 "Fail whitebox container rsc lrmd connection." do_test whitebox-fail3 "Failed containers should not run nested on remote nodes." do_test whitebox-start "Start whitebox container with resources assigned to it" do_test whitebox-stop "Stop whitebox container with resources assigned to it" do_test whitebox-move "Move whitebox container with resources assigned to it" do_test whitebox-asymmetric "Verify connection rsc opts-in based on container resource" do_test whitebox-ms-ordering "Verify promote/demote can not occur before connection is established" do_test whitebox-orphaned "Properly shutdown orphaned whitebox container" do_test whitebox-orphan-ms "Properly tear down orphan ms resources on remote-nodes" do_test whitebox-unexpectedly-running "Recover container nodes the cluster did not start." do_test whitebox-migrate1 "Migrate both container and connection resource" do_test whitebox-imply-stop-on-fence "imply stop action on container node rsc when host node is fenced" echo "" do_test remote-startup-probes "Baremetal remote-node startup probes" do_test remote-startup "Startup a newly discovered remote-nodes with no status." do_test remote-fence-unclean "Fence unclean baremetal remote-node" do_test remote-fence-unclean2 "Fence baremetal remote-node after cluster node fails and connection can not be recovered" do_test remote-move "Move remote-node connection resource" do_test remote-disable "Disable a baremetal remote-node" do_test remote-orphaned "Properly shutdown orphaned connection resource" do_test remote-orphaned2 "verify we can handle orphaned remote connections with active resources on the remote" do_test remote-recover "Recover connection resource after cluster-node fails." do_test remote-stale-node-entry "Make sure we properly handle leftover remote-node entries in the node section" do_test remote-partial-migrate "Make sure partial migrations are handled before ops on the remote node." do_test remote-partial-migrate2 "Make sure partial migration target is prefered for remote connection." do_test remote-recover-fail "Make sure start failure causes fencing if rsc are active on remote." do_test remote-start-fail "Make sure a start failure does not result in fencing if no active resources are on remote." do_test remote-unclean2 "Make monitor failure always results in fencing, even if no rsc are active on remote." echo "" do_test resource-discovery "Exercises resource-discovery location constraint option." do_test rsc-discovery-per-node "Disable resource discovery per node" echo "" -do_test isolation-start-all "Start docker isolated resources." +do_test isolation-start-all "Start docker isolated resources." do_test isolation-restart-all "Restart docker isolated resources." +do_test isolation-clone "Cloned isolated primitive." echo "" test_results diff --git a/pengine/test10/829.summary b/pengine/test10/829.summary index cf23cf0ca3..44300c3f44 100644 --- a/pengine/test10/829.summary +++ b/pengine/test10/829.summary @@ -1,64 +1,64 @@ Current cluster status: Node c001n02 (e9bdfde9-01b0-421f-acd8-8a65a53e775f): UNCLEAN (offline) Online: [ c001n01 c001n03 c001n08 ] DcIPaddr (ocf::heartbeat:IPaddr): Started c001n08 rsc_c001n08 (ocf::heartbeat:IPaddr): Started c001n08 - rsc_c001n02 (ocf::heartbeat:IPaddr): Started c001n02 + rsc_c001n02 (ocf::heartbeat:IPaddr): Started c001n02 (UNCLEAN) rsc_c001n03 (ocf::heartbeat:IPaddr): Started c001n03 rsc_c001n01 (ocf::heartbeat:IPaddr): Started c001n01 Clone Set: DoFencing [child_DoFencing] (unique) - child_DoFencing:0 (stonith:ssh): Started c001n02 + child_DoFencing:0 (stonith:ssh): Started c001n02 (UNCLEAN) child_DoFencing:1 (stonith:ssh): Started c001n03 child_DoFencing:2 (stonith:ssh): Started c001n01 child_DoFencing:3 (stonith:ssh): Started c001n08 Transition Summary: * Move rsc_c001n02 (Started c001n02 -> c001n01) * Stop child_DoFencing:0 (c001n02) Executing cluster transition: * Resource action: DcIPaddr monitor on c001n03 * Resource action: DcIPaddr monitor on c001n01 * Resource action: rsc_c001n08 monitor on c001n03 * Resource action: rsc_c001n08 monitor on c001n01 * Resource action: rsc_c001n02 monitor on c001n08 * Resource action: rsc_c001n02 monitor on c001n03 * Resource action: rsc_c001n02 monitor on c001n01 * Resource action: rsc_c001n03 monitor on c001n08 * Resource action: rsc_c001n03 monitor on c001n01 * Resource action: rsc_c001n01 monitor on c001n08 * Resource action: rsc_c001n01 monitor on c001n03 * Resource action: child_DoFencing:0 monitor on c001n01 * Resource action: child_DoFencing:1 monitor on c001n01 * Resource action: child_DoFencing:2 monitor on c001n08 * Resource action: child_DoFencing:2 monitor on c001n03 * Resource action: child_DoFencing:3 monitor on c001n03 * Resource action: child_DoFencing:3 monitor on c001n01 * Fencing c001n02 (reboot) * Pseudo action: stonith_complete * Pseudo action: probe_complete * Pseudo action: rsc_c001n02_stop_0 * Pseudo action: DoFencing_stop_0 * Resource action: rsc_c001n02 start on c001n01 * Pseudo action: child_DoFencing:0_stop_0 * Pseudo action: DoFencing_stopped_0 * Pseudo action: all_stopped * Resource action: rsc_c001n02 monitor=5000 on c001n01 Revised cluster status: Online: [ c001n01 c001n03 c001n08 ] OFFLINE: [ c001n02 ] DcIPaddr (ocf::heartbeat:IPaddr): Started c001n08 rsc_c001n08 (ocf::heartbeat:IPaddr): Started c001n08 rsc_c001n02 (ocf::heartbeat:IPaddr): Started c001n01 rsc_c001n03 (ocf::heartbeat:IPaddr): Started c001n03 rsc_c001n01 (ocf::heartbeat:IPaddr): Started c001n01 Clone Set: DoFencing [child_DoFencing] (unique) child_DoFencing:0 (stonith:ssh): Stopped child_DoFencing:1 (stonith:ssh): Started c001n03 child_DoFencing:2 (stonith:ssh): Started c001n01 child_DoFencing:3 (stonith:ssh): Started c001n08 diff --git a/pengine/test10/bug-5186-partial-migrate.summary b/pengine/test10/bug-5186-partial-migrate.summary index 1d3f1eb46a..f848c97bf1 100644 --- a/pengine/test10/bug-5186-partial-migrate.summary +++ b/pengine/test10/bug-5186-partial-migrate.summary @@ -1,87 +1,90 @@ Current cluster status: Node bl460g1n7 (3232261593): UNCLEAN (offline) Online: [ bl460g1n6 bl460g1n8 ] - prmDummy (ocf::pacemaker:Dummy): Started bl460g1n7 - prmVM2 (ocf::heartbeat:VirtualDomain): Started bl460g1n7 + prmDummy (ocf::pacemaker:Dummy): Started bl460g1n7 (UNCLEAN) + prmVM2 (ocf::heartbeat:VirtualDomain): Started bl460g1n7 (UNCLEAN) Resource Group: grpStonith6 prmStonith6-1 (stonith:external/stonith-helper): Started bl460g1n8 prmStonith6-2 (stonith:external/ipmi): Started bl460g1n8 Resource Group: grpStonith7 prmStonith7-1 (stonith:external/stonith-helper): Started bl460g1n6 prmStonith7-2 (stonith:external/ipmi): Started bl460g1n6 Resource Group: grpStonith8 - prmStonith8-1 (stonith:external/stonith-helper): Started bl460g1n7 - prmStonith8-2 (stonith:external/ipmi): Started bl460g1n7 + prmStonith8-1 (stonith:external/stonith-helper): Started bl460g1n7 (UNCLEAN) + prmStonith8-2 (stonith:external/ipmi): Started bl460g1n7 (UNCLEAN) Clone Set: clnDiskd1 [prmDiskd1] - Started: [ bl460g1n6 bl460g1n7 bl460g1n8 ] + prmDiskd1 (ocf::pacemaker:diskd): Started bl460g1n7 (UNCLEAN) + Started: [ bl460g1n6 bl460g1n8 ] Clone Set: clnDiskd2 [prmDiskd2] - Started: [ bl460g1n6 bl460g1n7 bl460g1n8 ] + prmDiskd2 (ocf::pacemaker:diskd): Started bl460g1n7 (UNCLEAN) + Started: [ bl460g1n6 bl460g1n8 ] Clone Set: clnPing [prmPing] - Started: [ bl460g1n6 bl460g1n7 bl460g1n8 ] + prmPing (ocf::pacemaker:ping): Started bl460g1n7 (UNCLEAN) + Started: [ bl460g1n6 bl460g1n8 ] Transition Summary: * Move prmDummy (Started bl460g1n7 -> bl460g1n6) * Move prmVM2 (Started bl460g1n7 -> bl460g1n8) * Move prmStonith8-1 (Started bl460g1n7 -> bl460g1n6) * Move prmStonith8-2 (Started bl460g1n7 -> bl460g1n6) * Stop prmDiskd1:0 (bl460g1n7) * Stop prmDiskd2:0 (bl460g1n7) * Stop prmPing:0 (bl460g1n7) Executing cluster transition: * Resource action: prmVM2 stop on bl460g1n6 * Fencing bl460g1n7 (reboot) * Pseudo action: stonith_complete * Pseudo action: prmDummy_stop_0 * Pseudo action: prmVM2_stop_0 * Pseudo action: grpStonith8_stop_0 * Pseudo action: prmStonith8-2_stop_0 * Pseudo action: clnDiskd1_stop_0 * Pseudo action: clnDiskd2_stop_0 * Pseudo action: clnPing_stop_0 * Resource action: prmDummy start on bl460g1n6 * Resource action: prmVM2 start on bl460g1n8 * Pseudo action: prmStonith8-1_stop_0 * Pseudo action: prmDiskd1_stop_0 * Pseudo action: clnDiskd1_stopped_0 * Pseudo action: prmDiskd2_stop_0 * Pseudo action: clnDiskd2_stopped_0 * Pseudo action: prmPing_stop_0 * Pseudo action: clnPing_stopped_0 * Pseudo action: all_stopped * Resource action: prmVM2 monitor=10000 on bl460g1n8 * Pseudo action: grpStonith8_stopped_0 * Pseudo action: grpStonith8_start_0 * Resource action: prmStonith8-1 start on bl460g1n6 * Resource action: prmStonith8-2 start on bl460g1n6 * Pseudo action: grpStonith8_running_0 * Resource action: prmStonith8-1 monitor=10000 on bl460g1n6 * Resource action: prmStonith8-2 monitor=3600000 on bl460g1n6 Revised cluster status: Online: [ bl460g1n6 bl460g1n8 ] OFFLINE: [ bl460g1n7 ] prmDummy (ocf::pacemaker:Dummy): Started bl460g1n6 prmVM2 (ocf::heartbeat:VirtualDomain): Started bl460g1n8 Resource Group: grpStonith6 prmStonith6-1 (stonith:external/stonith-helper): Started bl460g1n8 prmStonith6-2 (stonith:external/ipmi): Started bl460g1n8 Resource Group: grpStonith7 prmStonith7-1 (stonith:external/stonith-helper): Started bl460g1n6 prmStonith7-2 (stonith:external/ipmi): Started bl460g1n6 Resource Group: grpStonith8 prmStonith8-1 (stonith:external/stonith-helper): Started bl460g1n6 prmStonith8-2 (stonith:external/ipmi): Started bl460g1n6 Clone Set: clnDiskd1 [prmDiskd1] Started: [ bl460g1n6 bl460g1n8 ] Stopped: [ bl460g1n7 ] Clone Set: clnDiskd2 [prmDiskd2] Started: [ bl460g1n6 bl460g1n8 ] Stopped: [ bl460g1n7 ] Clone Set: clnPing [prmPing] Started: [ bl460g1n6 bl460g1n8 ] Stopped: [ bl460g1n7 ] diff --git a/pengine/test10/bug-cl-5212.summary b/pengine/test10/bug-cl-5212.summary index 22a47577c2..542afbf86d 100644 --- a/pengine/test10/bug-cl-5212.summary +++ b/pengine/test10/bug-cl-5212.summary @@ -1,61 +1,67 @@ Current cluster status: Node srv01 (3232238280): UNCLEAN (offline) Node srv02 (3232238290): UNCLEAN (offline) Online: [ srv03 ] Resource Group: grpStonith1 - prmStonith1-1 (stonith:external/ssh): Started srv02 + prmStonith1-1 (stonith:external/ssh): Started srv02 (UNCLEAN) Resource Group: grpStonith2 - prmStonith2-1 (stonith:external/ssh): Started srv01 + prmStonith2-1 (stonith:external/ssh): Started srv01 (UNCLEAN) Resource Group: grpStonith3 - prmStonith3-1 (stonith:external/ssh): Started srv01 + prmStonith3-1 (stonith:external/ssh): Started srv01 (UNCLEAN) Master/Slave Set: msPostgresql [pgsql] - Masters: [ srv01 ] - Slaves: [ srv02 srv03 ] + pgsql (ocf::pacemaker:Stateful): Started srv02 (UNCLEAN) + pgsql (ocf::pacemaker:Stateful): Master srv01 (UNCLEAN) + Slaves: [ srv03 ] Clone Set: clnPingd [prmPingd] - Started: [ srv01 srv02 srv03 ] + prmPingd (ocf::pacemaker:ping): Started srv02 (UNCLEAN) + prmPingd (ocf::pacemaker:ping): Started srv01 (UNCLEAN) + Started: [ srv03 ] Transition Summary: * Stop prmStonith1-1 (Started srv02 - blocked) * Stop prmStonith2-1 (Started srv01 - blocked) * Stop prmStonith3-1 (srv01 - blocked) * Stop pgsql:0 (srv02 - blocked) * Demote pgsql:1 (Master -> Stopped srv01 - blocked) * Stop prmPingd:0 (srv02 - blocked) * Stop prmPingd:1 (srv01 - blocked) Executing cluster transition: * Pseudo action: grpStonith1_stop_0 * Pseudo action: grpStonith1_start_0 * Pseudo action: grpStonith2_stop_0 * Pseudo action: grpStonith2_start_0 * Pseudo action: grpStonith3_stop_0 * Pseudo action: msPostgresql_pre_notify_stop_0 * Pseudo action: clnPingd_stop_0 * Resource action: pgsql notify on srv03 * Pseudo action: msPostgresql_confirmed-pre_notify_stop_0 * Pseudo action: msPostgresql_stop_0 * Pseudo action: clnPingd_stopped_0 * Pseudo action: msPostgresql_stopped_0 * Pseudo action: msPostgresql_post_notify_stopped_0 * Resource action: pgsql notify on srv03 * Pseudo action: msPostgresql_confirmed-post_notify_stopped_0 Revised cluster status: Node srv01 (3232238280): UNCLEAN (offline) Node srv02 (3232238290): UNCLEAN (offline) Online: [ srv03 ] Resource Group: grpStonith1 - prmStonith1-1 (stonith:external/ssh): Started srv02 + prmStonith1-1 (stonith:external/ssh): Started srv02 (UNCLEAN) Resource Group: grpStonith2 - prmStonith2-1 (stonith:external/ssh): Started srv01 + prmStonith2-1 (stonith:external/ssh): Started srv01 (UNCLEAN) Resource Group: grpStonith3 - prmStonith3-1 (stonith:external/ssh): Started srv01 + prmStonith3-1 (stonith:external/ssh): Started srv01 (UNCLEAN) Master/Slave Set: msPostgresql [pgsql] - Masters: [ srv01 ] - Slaves: [ srv02 srv03 ] + pgsql (ocf::pacemaker:Stateful): Started srv02 (UNCLEAN) + pgsql (ocf::pacemaker:Stateful): Master srv01 (UNCLEAN) + Slaves: [ srv03 ] Clone Set: clnPingd [prmPingd] - Started: [ srv01 srv02 srv03 ] + prmPingd (ocf::pacemaker:ping): Started srv02 (UNCLEAN) + prmPingd (ocf::pacemaker:ping): Started srv01 (UNCLEAN) + Started: [ srv03 ] diff --git a/pengine/test10/bug-lf-2508.summary b/pengine/test10/bug-lf-2508.summary index 7f5aed1493..af5e4e2492 100644 --- a/pengine/test10/bug-lf-2508.summary +++ b/pengine/test10/bug-lf-2508.summary @@ -1,102 +1,111 @@ Current cluster status: Node srv02 (71085d5e-1c63-49e0-8c8c-400d610b4182): UNCLEAN (offline) Online: [ srv01 srv03 srv04 ] Resource Group: Group01 Dummy01 (ocf::heartbeat:Dummy): Stopped Resource Group: Group02 - Dummy02 (ocf::heartbeat:Dummy): Started srv02 + Dummy02 (ocf::heartbeat:Dummy): Started srv02 (UNCLEAN) Resource Group: Group03 Dummy03 (ocf::heartbeat:Dummy): Started srv03 Clone Set: clnStonith1 [grpStonith1] - Started: [ srv02 srv03 srv04 ] + Resource Group: grpStonith1:1 + prmStonith1-1 (stonith:external/stonith-helper): Started srv02 (UNCLEAN) + prmStonith1-3 (stonith:external/ssh): Started srv02 (UNCLEAN) + Started: [ srv03 srv04 ] Stopped: [ srv01 ] Clone Set: clnStonith2 [grpStonith2] Started: [ srv01 srv03 srv04 ] Stopped: [ srv02 ] Clone Set: clnStonith3 [grpStonith3] + Resource Group: grpStonith3:0 + prmStonith3-1 (stonith:external/stonith-helper): Started srv02 (UNCLEAN) + prmStonith3-3 (stonith:external/ssh): Started srv02 (UNCLEAN) Resource Group: grpStonith3:1 prmStonith3-1 (stonith:external/stonith-helper): Started srv01 prmStonith3-3 (stonith:external/ssh): Stopped - Started: [ srv02 srv04 ] + Started: [ srv04 ] Stopped: [ srv03 ] Clone Set: clnStonith4 [grpStonith4] - Started: [ srv01 srv02 srv03 ] + Resource Group: grpStonith4:1 + prmStonith4-1 (stonith:external/stonith-helper): Started srv02 (UNCLEAN) + prmStonith4-3 (stonith:external/ssh): Started srv02 (UNCLEAN) + Started: [ srv01 srv03 ] Stopped: [ srv04 ] Transition Summary: * Start Dummy01 (srv01) * Move Dummy02 (Started srv02 -> srv04) * Stop prmStonith1-1:1 (srv02) * Stop prmStonith1-3:1 (srv02) * Stop prmStonith3-1:0 (srv02) * Stop prmStonith3-3:0 (srv02) * Start prmStonith3-3:1 (srv01) * Stop prmStonith4-1:1 (srv02) * Stop prmStonith4-3:1 (srv02) Executing cluster transition: * Pseudo action: Group01_start_0 * Resource action: prmStonith3-1:1 monitor=3600000 on srv01 * Fencing srv02 (reboot) * Pseudo action: stonith_complete * Resource action: Dummy01 start on srv01 * Pseudo action: Group02_stop_0 * Pseudo action: Dummy02_stop_0 * Pseudo action: clnStonith1_stop_0 * Pseudo action: clnStonith3_stop_0 * Pseudo action: clnStonith4_stop_0 * Pseudo action: Group01_running_0 * Resource action: Dummy01 monitor=10000 on srv01 * Pseudo action: Group02_stopped_0 * Pseudo action: Group02_start_0 * Resource action: Dummy02 start on srv04 * Pseudo action: grpStonith1:1_stop_0 * Pseudo action: prmStonith1-3:1_stop_0 * Pseudo action: grpStonith3:0_stop_0 * Pseudo action: prmStonith3-3:1_stop_0 * Pseudo action: grpStonith4:1_stop_0 * Pseudo action: prmStonith4-3:1_stop_0 * Pseudo action: Group02_running_0 * Resource action: Dummy02 monitor=10000 on srv04 * Pseudo action: prmStonith1-1:1_stop_0 * Pseudo action: prmStonith3-1:1_stop_0 * Pseudo action: prmStonith4-1:1_stop_0 * Pseudo action: all_stopped * Pseudo action: grpStonith1:1_stopped_0 * Pseudo action: clnStonith1_stopped_0 * Pseudo action: grpStonith3:0_stopped_0 * Pseudo action: clnStonith3_stopped_0 * Pseudo action: clnStonith3_start_0 * Pseudo action: grpStonith4:1_stopped_0 * Pseudo action: clnStonith4_stopped_0 * Pseudo action: grpStonith3:1_start_0 * Resource action: prmStonith3-3:1 start on srv01 * Pseudo action: grpStonith3:1_running_0 * Resource action: prmStonith3-3:1 monitor=3600000 on srv01 * Pseudo action: clnStonith3_running_0 Revised cluster status: Online: [ srv01 srv03 srv04 ] OFFLINE: [ srv02 ] Resource Group: Group01 Dummy01 (ocf::heartbeat:Dummy): Started srv01 Resource Group: Group02 Dummy02 (ocf::heartbeat:Dummy): Started srv04 Resource Group: Group03 Dummy03 (ocf::heartbeat:Dummy): Started srv03 Clone Set: clnStonith1 [grpStonith1] Started: [ srv03 srv04 ] Stopped: [ srv01 srv02 ] Clone Set: clnStonith2 [grpStonith2] Started: [ srv01 srv03 srv04 ] Stopped: [ srv02 ] Clone Set: clnStonith3 [grpStonith3] Started: [ srv01 srv04 ] Stopped: [ srv02 srv03 ] Clone Set: clnStonith4 [grpStonith4] Started: [ srv01 srv03 ] Stopped: [ srv02 srv04 ] diff --git a/pengine/test10/bug-lf-2551.summary b/pengine/test10/bug-lf-2551.summary index d362a23889..f8d861ca2f 100644 --- a/pengine/test10/bug-lf-2551.summary +++ b/pengine/test10/bug-lf-2551.summary @@ -1,218 +1,225 @@ Current cluster status: Node hex-9: UNCLEAN (offline) Online: [ hex-0 hex-7 hex-8 ] vm-00 (ocf::heartbeat:Xen): Started hex-0 Clone Set: base-clone [base-group] - Started: [ hex-0 hex-7 hex-8 hex-9 ] + Resource Group: base-group:3 + dlm (ocf::pacemaker:controld): Started hex-9 (UNCLEAN) + o2cb (ocf::ocfs2:o2cb): Started hex-9 (UNCLEAN) + clvm (ocf::lvm2:clvmd): Started hex-9 (UNCLEAN) + cmirrord (ocf::lvm2:cmirrord): Started hex-9 (UNCLEAN) + vg1 (ocf::heartbeat:LVM): Started hex-9 (UNCLEAN) + ocfs2-1 (ocf::heartbeat:Filesystem): Started hex-9 (UNCLEAN) + Started: [ hex-0 hex-7 hex-8 ] vm-01 (ocf::heartbeat:Xen): Started hex-7 vm-02 (ocf::heartbeat:Xen): Started hex-8 - vm-03 (ocf::heartbeat:Xen): Started hex-9 + vm-03 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-04 (ocf::heartbeat:Xen): Started hex-7 vm-05 (ocf::heartbeat:Xen): Started hex-8 - fencing-sbd (stonith:external/sbd): Started hex-9 - vm-06 (ocf::heartbeat:Xen): Started hex-9 + fencing-sbd (stonith:external/sbd): Started hex-9 (UNCLEAN) + vm-06 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-07 (ocf::heartbeat:Xen): Started hex-7 vm-08 (ocf::heartbeat:Xen): Started hex-8 - vm-09 (ocf::heartbeat:Xen): Started hex-9 + vm-09 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-10 (ocf::heartbeat:Xen): Started hex-0 vm-11 (ocf::heartbeat:Xen): Started hex-7 vm-12 (ocf::heartbeat:Xen): Started hex-8 - vm-13 (ocf::heartbeat:Xen): Started hex-9 + vm-13 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-14 (ocf::heartbeat:Xen): Started hex-0 vm-15 (ocf::heartbeat:Xen): Started hex-7 vm-16 (ocf::heartbeat:Xen): Started hex-8 - vm-17 (ocf::heartbeat:Xen): Started hex-9 + vm-17 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-18 (ocf::heartbeat:Xen): Started hex-0 vm-19 (ocf::heartbeat:Xen): Started hex-7 vm-20 (ocf::heartbeat:Xen): Started hex-8 - vm-21 (ocf::heartbeat:Xen): Started hex-9 + vm-21 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-22 (ocf::heartbeat:Xen): Started hex-0 vm-23 (ocf::heartbeat:Xen): Started hex-7 vm-24 (ocf::heartbeat:Xen): Started hex-8 - vm-25 (ocf::heartbeat:Xen): Started hex-9 + vm-25 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-26 (ocf::heartbeat:Xen): Started hex-0 vm-27 (ocf::heartbeat:Xen): Started hex-7 vm-28 (ocf::heartbeat:Xen): Started hex-8 - vm-29 (ocf::heartbeat:Xen): Started hex-9 + vm-29 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-30 (ocf::heartbeat:Xen): Started hex-0 vm-31 (ocf::heartbeat:Xen): Started hex-7 vm-32 (ocf::heartbeat:Xen): Started hex-8 - dummy1 (ocf::heartbeat:Dummy): Started hex-9 - vm-33 (ocf::heartbeat:Xen): Started hex-9 + dummy1 (ocf::heartbeat:Dummy): Started hex-9 (UNCLEAN) + vm-33 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-34 (ocf::heartbeat:Xen): Started hex-0 vm-35 (ocf::heartbeat:Xen): Started hex-7 vm-36 (ocf::heartbeat:Xen): Started hex-8 - vm-37 (ocf::heartbeat:Xen): Started hex-9 + vm-37 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-38 (ocf::heartbeat:Xen): Started hex-0 vm-39 (ocf::heartbeat:Xen): Started hex-7 vm-40 (ocf::heartbeat:Xen): Started hex-8 - vm-41 (ocf::heartbeat:Xen): Started hex-9 + vm-41 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-42 (ocf::heartbeat:Xen): Started hex-0 vm-43 (ocf::heartbeat:Xen): Started hex-7 vm-44 (ocf::heartbeat:Xen): Started hex-8 - vm-45 (ocf::heartbeat:Xen): Started hex-9 + vm-45 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-46 (ocf::heartbeat:Xen): Started hex-0 vm-47 (ocf::heartbeat:Xen): Started hex-7 vm-48 (ocf::heartbeat:Xen): Started hex-8 - vm-49 (ocf::heartbeat:Xen): Started hex-9 + vm-49 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-50 (ocf::heartbeat:Xen): Started hex-0 vm-51 (ocf::heartbeat:Xen): Started hex-7 vm-52 (ocf::heartbeat:Xen): Started hex-8 - vm-53 (ocf::heartbeat:Xen): Started hex-9 + vm-53 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-54 (ocf::heartbeat:Xen): Started hex-0 vm-55 (ocf::heartbeat:Xen): Started hex-7 vm-56 (ocf::heartbeat:Xen): Started hex-8 - vm-57 (ocf::heartbeat:Xen): Started hex-9 + vm-57 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-58 (ocf::heartbeat:Xen): Started hex-0 vm-59 (ocf::heartbeat:Xen): Started hex-7 vm-60 (ocf::heartbeat:Xen): Started hex-8 - vm-61 (ocf::heartbeat:Xen): Started hex-9 + vm-61 (ocf::heartbeat:Xen): Started hex-9 (UNCLEAN) vm-62 (ocf::heartbeat:Xen): Stopped vm-63 (ocf::heartbeat:Xen): Stopped vm-64 (ocf::heartbeat:Xen): Stopped Transition Summary: * Move fencing-sbd (Started hex-9 -> hex-0) * Move dummy1 (Started hex-9 -> hex-0) * Stop dlm:3 (hex-9) * Stop o2cb:3 (hex-9) * Stop clvm:3 (hex-9) * Stop cmirrord:3 (hex-9) * Stop vg1:3 (hex-9) * Stop ocfs2-1:3 (hex-9) * Stop vm-03 (hex-9) * Stop vm-06 (hex-9) * Stop vm-09 (hex-9) * Stop vm-13 (hex-9) * Stop vm-17 (hex-9) * Stop vm-21 (hex-9) * Stop vm-25 (hex-9) * Stop vm-29 (hex-9) * Stop vm-33 (hex-9) * Stop vm-37 (hex-9) * Stop vm-41 (hex-9) * Stop vm-45 (hex-9) * Stop vm-49 (hex-9) * Stop vm-53 (hex-9) * Stop vm-57 (hex-9) * Stop vm-61 (hex-9) Executing cluster transition: * Resource action: dummy1 monitor=300000 on hex-8 * Resource action: dummy1 monitor=300000 on hex-7 * Fencing hex-9 (reboot) * Pseudo action: stonith_complete * Pseudo action: load_stopped_hex-8 * Pseudo action: load_stopped_hex-7 * Pseudo action: load_stopped_hex-0 * Pseudo action: fencing-sbd_stop_0 * Pseudo action: dummy1_stop_0 * Pseudo action: vm-03_stop_0 * Pseudo action: vm-06_stop_0 * Pseudo action: vm-09_stop_0 * Pseudo action: vm-13_stop_0 * Pseudo action: vm-17_stop_0 * Pseudo action: vm-21_stop_0 * Pseudo action: vm-25_stop_0 * Pseudo action: vm-29_stop_0 * Pseudo action: vm-33_stop_0 * Pseudo action: vm-37_stop_0 * Pseudo action: vm-41_stop_0 * Pseudo action: vm-45_stop_0 * Pseudo action: vm-49_stop_0 * Pseudo action: vm-53_stop_0 * Pseudo action: vm-57_stop_0 * Pseudo action: vm-61_stop_0 * Pseudo action: load_stopped_hex-9 * Resource action: fencing-sbd start on hex-0 * Resource action: dummy1 start on hex-0 * Pseudo action: base-clone_stop_0 * Resource action: dummy1 monitor=30000 on hex-0 * Pseudo action: base-group:3_stop_0 * Pseudo action: ocfs2-1:3_stop_0 * Pseudo action: vg1:3_stop_0 * Pseudo action: cmirrord:3_stop_0 * Pseudo action: clvm:3_stop_0 * Pseudo action: o2cb:3_stop_0 * Pseudo action: dlm:3_stop_0 * Pseudo action: all_stopped * Pseudo action: base-group:3_stopped_0 * Pseudo action: base-clone_stopped_0 Revised cluster status: Online: [ hex-0 hex-7 hex-8 ] OFFLINE: [ hex-9 ] vm-00 (ocf::heartbeat:Xen): Started hex-0 Clone Set: base-clone [base-group] Started: [ hex-0 hex-7 hex-8 ] Stopped: [ hex-9 ] vm-01 (ocf::heartbeat:Xen): Started hex-7 vm-02 (ocf::heartbeat:Xen): Started hex-8 vm-03 (ocf::heartbeat:Xen): Stopped vm-04 (ocf::heartbeat:Xen): Started hex-7 vm-05 (ocf::heartbeat:Xen): Started hex-8 fencing-sbd (stonith:external/sbd): Started hex-0 vm-06 (ocf::heartbeat:Xen): Stopped vm-07 (ocf::heartbeat:Xen): Started hex-7 vm-08 (ocf::heartbeat:Xen): Started hex-8 vm-09 (ocf::heartbeat:Xen): Stopped vm-10 (ocf::heartbeat:Xen): Started hex-0 vm-11 (ocf::heartbeat:Xen): Started hex-7 vm-12 (ocf::heartbeat:Xen): Started hex-8 vm-13 (ocf::heartbeat:Xen): Stopped vm-14 (ocf::heartbeat:Xen): Started hex-0 vm-15 (ocf::heartbeat:Xen): Started hex-7 vm-16 (ocf::heartbeat:Xen): Started hex-8 vm-17 (ocf::heartbeat:Xen): Stopped vm-18 (ocf::heartbeat:Xen): Started hex-0 vm-19 (ocf::heartbeat:Xen): Started hex-7 vm-20 (ocf::heartbeat:Xen): Started hex-8 vm-21 (ocf::heartbeat:Xen): Stopped vm-22 (ocf::heartbeat:Xen): Started hex-0 vm-23 (ocf::heartbeat:Xen): Started hex-7 vm-24 (ocf::heartbeat:Xen): Started hex-8 vm-25 (ocf::heartbeat:Xen): Stopped vm-26 (ocf::heartbeat:Xen): Started hex-0 vm-27 (ocf::heartbeat:Xen): Started hex-7 vm-28 (ocf::heartbeat:Xen): Started hex-8 vm-29 (ocf::heartbeat:Xen): Stopped vm-30 (ocf::heartbeat:Xen): Started hex-0 vm-31 (ocf::heartbeat:Xen): Started hex-7 vm-32 (ocf::heartbeat:Xen): Started hex-8 dummy1 (ocf::heartbeat:Dummy): Started hex-0 vm-33 (ocf::heartbeat:Xen): Stopped vm-34 (ocf::heartbeat:Xen): Started hex-0 vm-35 (ocf::heartbeat:Xen): Started hex-7 vm-36 (ocf::heartbeat:Xen): Started hex-8 vm-37 (ocf::heartbeat:Xen): Stopped vm-38 (ocf::heartbeat:Xen): Started hex-0 vm-39 (ocf::heartbeat:Xen): Started hex-7 vm-40 (ocf::heartbeat:Xen): Started hex-8 vm-41 (ocf::heartbeat:Xen): Stopped vm-42 (ocf::heartbeat:Xen): Started hex-0 vm-43 (ocf::heartbeat:Xen): Started hex-7 vm-44 (ocf::heartbeat:Xen): Started hex-8 vm-45 (ocf::heartbeat:Xen): Stopped vm-46 (ocf::heartbeat:Xen): Started hex-0 vm-47 (ocf::heartbeat:Xen): Started hex-7 vm-48 (ocf::heartbeat:Xen): Started hex-8 vm-49 (ocf::heartbeat:Xen): Stopped vm-50 (ocf::heartbeat:Xen): Started hex-0 vm-51 (ocf::heartbeat:Xen): Started hex-7 vm-52 (ocf::heartbeat:Xen): Started hex-8 vm-53 (ocf::heartbeat:Xen): Stopped vm-54 (ocf::heartbeat:Xen): Started hex-0 vm-55 (ocf::heartbeat:Xen): Started hex-7 vm-56 (ocf::heartbeat:Xen): Started hex-8 vm-57 (ocf::heartbeat:Xen): Stopped vm-58 (ocf::heartbeat:Xen): Started hex-0 vm-59 (ocf::heartbeat:Xen): Started hex-7 vm-60 (ocf::heartbeat:Xen): Started hex-8 vm-61 (ocf::heartbeat:Xen): Stopped vm-62 (ocf::heartbeat:Xen): Stopped vm-63 (ocf::heartbeat:Xen): Stopped vm-64 (ocf::heartbeat:Xen): Stopped diff --git a/pengine/test10/interleave-pseudo-stop.summary b/pengine/test10/interleave-pseudo-stop.summary index ea86c4faef..613d0f3b0c 100644 --- a/pengine/test10/interleave-pseudo-stop.summary +++ b/pengine/test10/interleave-pseudo-stop.summary @@ -1,78 +1,82 @@ Current cluster status: Node node1 (f6d93040-a9ad-4745-a647-57ed32444ca8): UNCLEAN (offline) Online: [ node2 ] Clone Set: stonithcloneset [stonithclone] - Started: [ node1 node2 ] + stonithclone (stonith:external/ssh): Started node1 (UNCLEAN) + Started: [ node2 ] Clone Set: evmscloneset [evmsclone] - Started: [ node1 node2 ] + evmsclone (ocf::heartbeat:EvmsSCC): Started node1 (UNCLEAN) + Started: [ node2 ] Clone Set: imagestorecloneset [imagestoreclone] - Started: [ node1 node2 ] + imagestoreclone (ocf::heartbeat:Filesystem): Started node1 (UNCLEAN) + Started: [ node2 ] Clone Set: configstorecloneset [configstoreclone] - Started: [ node1 node2 ] + configstoreclone (ocf::heartbeat:Filesystem): Started node1 (UNCLEAN) + Started: [ node2 ] Transition Summary: * Stop stonithclone:1 (node1) * Stop evmsclone:1 (node1) * Stop imagestoreclone:1 (node1) * Stop configstoreclone:1 (node1) Executing cluster transition: * Pseudo action: evmscloneset_pre_notify_stop_0 * Pseudo action: imagestorecloneset_pre_notify_stop_0 * Pseudo action: configstorecloneset_pre_notify_stop_0 * Fencing node1 (reboot) * Pseudo action: stonith_complete * Pseudo action: stonithcloneset_stop_0 * Resource action: evmsclone:1 notify on node2 * Pseudo action: evmsclone:0_post_notify_stop_0 * Pseudo action: evmscloneset_confirmed-pre_notify_stop_0 * Resource action: imagestoreclone:1 notify on node2 * Pseudo action: imagestoreclone:0_post_notify_stop_0 * Pseudo action: imagestorecloneset_confirmed-pre_notify_stop_0 * Pseudo action: imagestorecloneset_stop_0 * Resource action: configstoreclone:1 notify on node2 * Pseudo action: configstoreclone:0_post_notify_stop_0 * Pseudo action: configstorecloneset_confirmed-pre_notify_stop_0 * Pseudo action: configstorecloneset_stop_0 * Pseudo action: stonithclone:0_stop_0 * Pseudo action: stonithcloneset_stopped_0 * Pseudo action: imagestoreclone:0_stop_0 * Pseudo action: imagestorecloneset_stopped_0 * Pseudo action: configstoreclone:0_stop_0 * Pseudo action: configstorecloneset_stopped_0 * Pseudo action: imagestorecloneset_post_notify_stopped_0 * Pseudo action: configstorecloneset_post_notify_stopped_0 * Resource action: imagestoreclone:1 notify on node2 * Pseudo action: imagestoreclone:0_notified_0 * Pseudo action: imagestorecloneset_confirmed-post_notify_stopped_0 * Resource action: configstoreclone:1 notify on node2 * Pseudo action: configstoreclone:0_notified_0 * Pseudo action: configstorecloneset_confirmed-post_notify_stopped_0 * Pseudo action: evmscloneset_stop_0 * Pseudo action: evmsclone:0_stop_0 * Pseudo action: evmscloneset_stopped_0 * Pseudo action: evmscloneset_post_notify_stopped_0 * Resource action: evmsclone:1 notify on node2 * Pseudo action: evmsclone:0_notified_0 * Pseudo action: evmscloneset_confirmed-post_notify_stopped_0 * Pseudo action: all_stopped Revised cluster status: Online: [ node2 ] OFFLINE: [ node1 ] Clone Set: stonithcloneset [stonithclone] Started: [ node2 ] Stopped: [ node1 ] Clone Set: evmscloneset [evmsclone] Started: [ node2 ] Stopped: [ node1 ] Clone Set: imagestorecloneset [imagestoreclone] Started: [ node2 ] Stopped: [ node1 ] Clone Set: configstorecloneset [configstoreclone] Started: [ node2 ] Stopped: [ node1 ] diff --git a/pengine/test10/isolation-clone.dot b/pengine/test10/isolation-clone.dot new file mode 100644 index 0000000000..3c12d68b07 --- /dev/null +++ b/pengine/test10/isolation-clone.dot @@ -0,0 +1,30 @@ + digraph "g" { +"probe_complete rhel7-auto1" -> "probe_complete" [ style = bold] +"probe_complete rhel7-auto1" [ style=bold color="green" fontcolor="black"] +"probe_complete rhel7-auto2" -> "probe_complete" [ style = bold] +"probe_complete rhel7-auto2" [ style=bold color="green" fontcolor="black"] +"probe_complete" -> "single:0_start_0 rhel7-auto2" [ style = bold] +"probe_complete" -> "single:1_start_0 rhel7-auto1" [ style = bold] +"probe_complete" [ style=bold color="green" fontcolor="orange"] +"single-clone_running_0" [ style=bold color="green" fontcolor="orange"] +"single-clone_start_0" -> "single-clone_running_0" [ style = bold] +"single-clone_start_0" -> "single:0_start_0 rhel7-auto2" [ style = bold] +"single-clone_start_0" -> "single:1_start_0 rhel7-auto1" [ style = bold] +"single-clone_start_0" [ style=bold color="green" fontcolor="orange"] +"single:0_monitor_0 rhel7-auto1" -> "probe_complete rhel7-auto1" [ style = bold] +"single:0_monitor_0 rhel7-auto1" [ style=bold color="green" fontcolor="black"] +"single:0_monitor_0 rhel7-auto2" -> "probe_complete rhel7-auto2" [ style = bold] +"single:0_monitor_0 rhel7-auto2" [ style=bold color="green" fontcolor="black"] +"single:0_monitor_60000 rhel7-auto2" [ style=bold color="green" fontcolor="black"] +"single:0_start_0 rhel7-auto2" -> "single-clone_running_0" [ style = bold] +"single:0_start_0 rhel7-auto2" -> "single:0_monitor_60000 rhel7-auto2" [ style = bold] +"single:0_start_0 rhel7-auto2" [ style=bold color="green" fontcolor="black"] +"single:1_monitor_0 rhel7-auto1" -> "probe_complete rhel7-auto1" [ style = bold] +"single:1_monitor_0 rhel7-auto1" [ style=bold color="green" fontcolor="black"] +"single:1_monitor_0 rhel7-auto2" -> "probe_complete rhel7-auto2" [ style = bold] +"single:1_monitor_0 rhel7-auto2" [ style=bold color="green" fontcolor="black"] +"single:1_monitor_60000 rhel7-auto1" [ style=bold color="green" fontcolor="black"] +"single:1_start_0 rhel7-auto1" -> "single-clone_running_0" [ style = bold] +"single:1_start_0 rhel7-auto1" -> "single:1_monitor_60000 rhel7-auto1" [ style = bold] +"single:1_start_0 rhel7-auto1" [ style=bold color="green" fontcolor="black"] +} diff --git a/pengine/test10/isolation-clone.exp b/pengine/test10/isolation-clone.exp new file mode 100644 index 0000000000..1041b47bcb --- /dev/null +++ b/pengine/test10/isolation-clone.exp @@ -0,0 +1,167 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/pengine/test10/isolation-clone.scores b/pengine/test10/isolation-clone.scores new file mode 100644 index 0000000000..83cc9893cf --- /dev/null +++ b/pengine/test10/isolation-clone.scores @@ -0,0 +1,13 @@ +Allocation scores: +clone_color: single-clone allocation score on rhel7-auto1: 0 +clone_color: single-clone allocation score on rhel7-auto2: 0 +clone_color: single:0 allocation score on rhel7-auto1: 0 +clone_color: single:0 allocation score on rhel7-auto2: 0 +clone_color: single:1 allocation score on rhel7-auto1: 0 +clone_color: single:1 allocation score on rhel7-auto2: 0 +native_color: shooter allocation score on rhel7-auto1: 0 +native_color: shooter allocation score on rhel7-auto2: 0 +native_color: single:0 allocation score on rhel7-auto1: 0 +native_color: single:0 allocation score on rhel7-auto2: 0 +native_color: single:1 allocation score on rhel7-auto1: 0 +native_color: single:1 allocation score on rhel7-auto2: -INFINITY diff --git a/pengine/test10/isolation-clone.summary b/pengine/test10/isolation-clone.summary new file mode 100644 index 0000000000..2684ee87ba --- /dev/null +++ b/pengine/test10/isolation-clone.summary @@ -0,0 +1,34 @@ + +Current cluster status: +Online: [ rhel7-auto1 rhel7-auto2 ] + + shooter (stonith:fence_xvm): Started rhel7-auto1 + Clone Set: single-clone [single] (unique) + single:0 (ocf::heartbeat:Dummy): Stopped + single:1 (ocf::heartbeat:Dummy): Stopped + +Transition Summary: + * Start single:0 (rhel7-auto2) + * Start single:1 (rhel7-auto1) + +Executing cluster transition: + * Resource action: single:0 monitor on rhel7-auto2 + * Resource action: single:0 monitor on rhel7-auto1 + * Resource action: single:1 monitor on rhel7-auto2 + * Resource action: single:1 monitor on rhel7-auto1 + * Pseudo action: single-clone_start_0 + * Pseudo action: probe_complete + * Resource action: single:0 start on rhel7-auto2 + * Resource action: single:1 start on rhel7-auto1 + * Pseudo action: single-clone_running_0 + * Resource action: single:0 monitor=60000 on rhel7-auto2 + * Resource action: single:1 monitor=60000 on rhel7-auto1 + +Revised cluster status: +Online: [ rhel7-auto1 rhel7-auto2 ] + + shooter (stonith:fence_xvm): Started rhel7-auto1 + Clone Set: single-clone [single] (unique) + single:0 (ocf::heartbeat:Dummy): Started rhel7-auto2 + single:1 (ocf::heartbeat:Dummy): Started rhel7-auto1 + diff --git a/pengine/test10/isolation-clone.xml b/pengine/test10/isolation-clone.xml new file mode 100644 index 0000000000..dd800b512a --- /dev/null +++ b/pengine/test10/isolation-clone.xml @@ -0,0 +1,74 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/pengine/test10/isolation-restart-all.exp b/pengine/test10/isolation-restart-all.exp index ba2ccca211..ee8a287217 100644 --- a/pengine/test10/isolation-restart-all.exp +++ b/pengine/test10/isolation-restart-all.exp @@ -1,873 +1,873 @@ - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + diff --git a/pengine/test10/isolation-start-all.exp b/pengine/test10/isolation-start-all.exp index 8b059da1a9..9ceb351ae3 100644 --- a/pengine/test10/isolation-start-all.exp +++ b/pengine/test10/isolation-start-all.exp @@ -1,460 +1,460 @@ - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + - + diff --git a/pengine/test10/master-7.summary b/pengine/test10/master-7.summary index 81aeed7d58..0d786d79ad 100644 --- a/pengine/test10/master-7.summary +++ b/pengine/test10/master-7.summary @@ -1,121 +1,121 @@ Current cluster status: Node c001n01 (de937e3d-0309-4b5d-b85c-f96edc1ed8e3): UNCLEAN (offline) Online: [ c001n02 c001n03 c001n08 ] - DcIPaddr (ocf::heartbeat:IPaddr): Started c001n01 + DcIPaddr (ocf::heartbeat:IPaddr): Started c001n01 (UNCLEAN) Resource Group: group-1 ocf_192.168.100.181 (ocf::heartbeat:IPaddr): Started c001n03 heartbeat_192.168.100.182 (heartbeat:IPaddr): Started c001n03 ocf_192.168.100.183 (ocf::heartbeat:IPaddr): Started c001n03 lsb_dummy (lsb:/usr/lib/heartbeat/cts/LSBDummy): Started c001n02 - rsc_c001n01 (ocf::heartbeat:IPaddr): Started c001n01 + rsc_c001n01 (ocf::heartbeat:IPaddr): Started c001n01 (UNCLEAN) rsc_c001n08 (ocf::heartbeat:IPaddr): Started c001n08 rsc_c001n02 (ocf::heartbeat:IPaddr): Started c001n02 rsc_c001n03 (ocf::heartbeat:IPaddr): Started c001n03 Clone Set: DoFencing [child_DoFencing] (unique) - child_DoFencing:0 (stonith:ssh): Started c001n01 + child_DoFencing:0 (stonith:ssh): Started c001n01 (UNCLEAN) child_DoFencing:1 (stonith:ssh): Started c001n03 child_DoFencing:2 (stonith:ssh): Started c001n02 child_DoFencing:3 (stonith:ssh): Started c001n08 Master/Slave Set: master_rsc_1 [ocf_msdummy] (unique) - ocf_msdummy:0 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Master c001n01 + ocf_msdummy:0 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Master c001n01 (UNCLEAN) ocf_msdummy:1 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n03 ocf_msdummy:2 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n02 ocf_msdummy:3 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n08 - ocf_msdummy:4 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n01 + ocf_msdummy:4 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n01 (UNCLEAN) ocf_msdummy:5 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n03 ocf_msdummy:6 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n02 ocf_msdummy:7 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n08 Transition Summary: * Move DcIPaddr (Started c001n01 -> c001n03) * Move ocf_192.168.100.181 (Started c001n03 -> c001n02) * Move heartbeat_192.168.100.182 (Started c001n03 -> c001n02) * Move ocf_192.168.100.183 (Started c001n03 -> c001n02) * Move lsb_dummy (Started c001n02 -> c001n08) * Move rsc_c001n01 (Started c001n01 -> c001n03) * Stop child_DoFencing:0 (c001n01) * Demote ocf_msdummy:0 (Master -> Stopped c001n01) * Stop ocf_msdummy:4 (c001n01) Executing cluster transition: * Pseudo action: group-1_stop_0 * Resource action: child_DoFencing:2 monitor on c001n08 * Resource action: child_DoFencing:2 monitor on c001n03 * Resource action: child_DoFencing:3 monitor on c001n03 * Resource action: child_DoFencing:3 monitor on c001n02 * Resource action: ocf_msdummy:4 monitor on c001n08 * Resource action: ocf_msdummy:4 monitor on c001n03 * Resource action: ocf_msdummy:4 monitor on c001n02 * Resource action: ocf_msdummy:5 monitor on c001n08 * Resource action: ocf_msdummy:5 monitor on c001n02 * Resource action: ocf_msdummy:6 monitor on c001n08 * Resource action: ocf_msdummy:6 monitor on c001n03 * Resource action: ocf_msdummy:7 monitor on c001n03 * Resource action: ocf_msdummy:7 monitor on c001n02 * Pseudo action: master_rsc_1_demote_0 * Fencing c001n01 (reboot) * Pseudo action: stonith_complete * Pseudo action: probe_complete * Pseudo action: DcIPaddr_stop_0 * Resource action: ocf_192.168.100.183 stop on c001n03 * Resource action: lsb_dummy stop on c001n02 * Pseudo action: rsc_c001n01_stop_0 * Pseudo action: DoFencing_stop_0 * Pseudo action: ocf_msdummy:0_demote_0 * Pseudo action: master_rsc_1_demoted_0 * Pseudo action: master_rsc_1_stop_0 * Resource action: DcIPaddr start on c001n03 * Resource action: heartbeat_192.168.100.182 stop on c001n03 * Resource action: lsb_dummy start on c001n08 * Resource action: rsc_c001n01 start on c001n03 * Pseudo action: child_DoFencing:0_stop_0 * Pseudo action: DoFencing_stopped_0 * Pseudo action: ocf_msdummy:0_stop_0 * Pseudo action: ocf_msdummy:4_stop_0 * Pseudo action: master_rsc_1_stopped_0 * Resource action: DcIPaddr monitor=5000 on c001n03 * Resource action: ocf_192.168.100.181 stop on c001n03 * Resource action: lsb_dummy monitor=5000 on c001n08 * Resource action: rsc_c001n01 monitor=5000 on c001n03 * Pseudo action: all_stopped * Pseudo action: group-1_stopped_0 * Pseudo action: group-1_start_0 * Resource action: ocf_192.168.100.181 start on c001n02 * Resource action: heartbeat_192.168.100.182 start on c001n02 * Resource action: ocf_192.168.100.183 start on c001n02 * Pseudo action: group-1_running_0 * Resource action: ocf_192.168.100.181 monitor=5000 on c001n02 * Resource action: heartbeat_192.168.100.182 monitor=5000 on c001n02 * Resource action: ocf_192.168.100.183 monitor=5000 on c001n02 Revised cluster status: Online: [ c001n02 c001n03 c001n08 ] OFFLINE: [ c001n01 ] DcIPaddr (ocf::heartbeat:IPaddr): Started c001n03 Resource Group: group-1 ocf_192.168.100.181 (ocf::heartbeat:IPaddr): Started c001n02 heartbeat_192.168.100.182 (heartbeat:IPaddr): Started c001n02 ocf_192.168.100.183 (ocf::heartbeat:IPaddr): Started c001n02 lsb_dummy (lsb:/usr/lib/heartbeat/cts/LSBDummy): Started c001n08 rsc_c001n01 (ocf::heartbeat:IPaddr): Started c001n03 rsc_c001n08 (ocf::heartbeat:IPaddr): Started c001n08 rsc_c001n02 (ocf::heartbeat:IPaddr): Started c001n02 rsc_c001n03 (ocf::heartbeat:IPaddr): Started c001n03 Clone Set: DoFencing [child_DoFencing] (unique) child_DoFencing:0 (stonith:ssh): Stopped child_DoFencing:1 (stonith:ssh): Started c001n03 child_DoFencing:2 (stonith:ssh): Started c001n02 child_DoFencing:3 (stonith:ssh): Started c001n08 Master/Slave Set: master_rsc_1 [ocf_msdummy] (unique) ocf_msdummy:0 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Stopped ocf_msdummy:1 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n03 ocf_msdummy:2 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n02 ocf_msdummy:3 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n08 ocf_msdummy:4 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Stopped ocf_msdummy:5 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n03 ocf_msdummy:6 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n02 ocf_msdummy:7 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n08 diff --git a/pengine/test10/master-8.summary b/pengine/test10/master-8.summary index e12f7911ae..225be94c22 100644 --- a/pengine/test10/master-8.summary +++ b/pengine/test10/master-8.summary @@ -1,125 +1,125 @@ Current cluster status: Node c001n01 (de937e3d-0309-4b5d-b85c-f96edc1ed8e3): UNCLEAN (offline) Online: [ c001n02 c001n03 c001n08 ] - DcIPaddr (ocf::heartbeat:IPaddr): Started c001n01 + DcIPaddr (ocf::heartbeat:IPaddr): Started c001n01 (UNCLEAN) Resource Group: group-1 ocf_192.168.100.181 (ocf::heartbeat:IPaddr): Started c001n03 heartbeat_192.168.100.182 (heartbeat:IPaddr): Started c001n03 ocf_192.168.100.183 (ocf::heartbeat:IPaddr): Started c001n03 lsb_dummy (lsb:/usr/lib/heartbeat/cts/LSBDummy): Started c001n02 - rsc_c001n01 (ocf::heartbeat:IPaddr): Started c001n01 + rsc_c001n01 (ocf::heartbeat:IPaddr): Started c001n01 (UNCLEAN) rsc_c001n08 (ocf::heartbeat:IPaddr): Started c001n08 rsc_c001n02 (ocf::heartbeat:IPaddr): Started c001n02 rsc_c001n03 (ocf::heartbeat:IPaddr): Started c001n03 Clone Set: DoFencing [child_DoFencing] (unique) - child_DoFencing:0 (stonith:ssh): Started c001n01 + child_DoFencing:0 (stonith:ssh): Started c001n01 (UNCLEAN) child_DoFencing:1 (stonith:ssh): Started c001n03 child_DoFencing:2 (stonith:ssh): Started c001n02 child_DoFencing:3 (stonith:ssh): Started c001n08 Master/Slave Set: master_rsc_1 [ocf_msdummy] (unique) - ocf_msdummy:0 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Master c001n01 + ocf_msdummy:0 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Master c001n01 (UNCLEAN) ocf_msdummy:1 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n03 ocf_msdummy:2 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n02 ocf_msdummy:3 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n08 ocf_msdummy:4 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Stopped ocf_msdummy:5 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Stopped ocf_msdummy:6 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n02 ocf_msdummy:7 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n08 Transition Summary: * Move DcIPaddr (Started c001n01 -> c001n03) * Move ocf_192.168.100.181 (Started c001n03 -> c001n02) * Move heartbeat_192.168.100.182 (Started c001n03 -> c001n02) * Move ocf_192.168.100.183 (Started c001n03 -> c001n02) * Move lsb_dummy (Started c001n02 -> c001n08) * Move rsc_c001n01 (Started c001n01 -> c001n03) * Stop child_DoFencing:0 (c001n01) * Demote ocf_msdummy:0 (Master -> Slave c001n01 - blocked) * Move ocf_msdummy:0 (Slave c001n01 -> c001n03) Executing cluster transition: * Pseudo action: group-1_stop_0 * Resource action: child_DoFencing:2 monitor on c001n08 * Resource action: child_DoFencing:2 monitor on c001n03 * Resource action: child_DoFencing:3 monitor on c001n03 * Resource action: child_DoFencing:3 monitor on c001n02 * Resource action: ocf_msdummy:4 monitor on c001n08 * Resource action: ocf_msdummy:4 monitor on c001n03 * Resource action: ocf_msdummy:4 monitor on c001n02 * Resource action: ocf_msdummy:5 monitor on c001n08 * Resource action: ocf_msdummy:5 monitor on c001n03 * Resource action: ocf_msdummy:5 monitor on c001n02 * Resource action: ocf_msdummy:6 monitor on c001n08 * Resource action: ocf_msdummy:6 monitor on c001n03 * Resource action: ocf_msdummy:7 monitor on c001n03 * Resource action: ocf_msdummy:7 monitor on c001n02 * Pseudo action: master_rsc_1_demote_0 * Fencing c001n01 (reboot) * Pseudo action: stonith_complete * Pseudo action: probe_complete * Pseudo action: DcIPaddr_stop_0 * Resource action: ocf_192.168.100.183 stop on c001n03 * Resource action: lsb_dummy stop on c001n02 * Pseudo action: rsc_c001n01_stop_0 * Pseudo action: DoFencing_stop_0 * Pseudo action: ocf_msdummy:0_demote_0 * Pseudo action: master_rsc_1_demoted_0 * Pseudo action: master_rsc_1_stop_0 * Resource action: DcIPaddr start on c001n03 * Resource action: heartbeat_192.168.100.182 stop on c001n03 * Resource action: lsb_dummy start on c001n08 * Resource action: rsc_c001n01 start on c001n03 * Pseudo action: child_DoFencing:0_stop_0 * Pseudo action: DoFencing_stopped_0 * Pseudo action: ocf_msdummy:0_stop_0 * Pseudo action: master_rsc_1_stopped_0 * Pseudo action: master_rsc_1_start_0 * Resource action: DcIPaddr monitor=5000 on c001n03 * Resource action: ocf_192.168.100.181 stop on c001n03 * Resource action: lsb_dummy monitor=5000 on c001n08 * Resource action: rsc_c001n01 monitor=5000 on c001n03 * Resource action: ocf_msdummy:0 start on c001n03 * Pseudo action: master_rsc_1_running_0 * Pseudo action: all_stopped * Pseudo action: group-1_stopped_0 * Pseudo action: group-1_start_0 * Resource action: ocf_192.168.100.181 start on c001n02 * Resource action: heartbeat_192.168.100.182 start on c001n02 * Resource action: ocf_192.168.100.183 start on c001n02 * Resource action: ocf_msdummy:0 monitor=5000 on c001n03 * Pseudo action: group-1_running_0 * Resource action: ocf_192.168.100.181 monitor=5000 on c001n02 * Resource action: heartbeat_192.168.100.182 monitor=5000 on c001n02 * Resource action: ocf_192.168.100.183 monitor=5000 on c001n02 Revised cluster status: Online: [ c001n02 c001n03 c001n08 ] OFFLINE: [ c001n01 ] DcIPaddr (ocf::heartbeat:IPaddr): Started c001n03 Resource Group: group-1 ocf_192.168.100.181 (ocf::heartbeat:IPaddr): Started c001n02 heartbeat_192.168.100.182 (heartbeat:IPaddr): Started c001n02 ocf_192.168.100.183 (ocf::heartbeat:IPaddr): Started c001n02 lsb_dummy (lsb:/usr/lib/heartbeat/cts/LSBDummy): Started c001n08 rsc_c001n01 (ocf::heartbeat:IPaddr): Started c001n03 rsc_c001n08 (ocf::heartbeat:IPaddr): Started c001n08 rsc_c001n02 (ocf::heartbeat:IPaddr): Started c001n02 rsc_c001n03 (ocf::heartbeat:IPaddr): Started c001n03 Clone Set: DoFencing [child_DoFencing] (unique) child_DoFencing:0 (stonith:ssh): Stopped child_DoFencing:1 (stonith:ssh): Started c001n03 child_DoFencing:2 (stonith:ssh): Started c001n02 child_DoFencing:3 (stonith:ssh): Started c001n08 Master/Slave Set: master_rsc_1 [ocf_msdummy] (unique) ocf_msdummy:0 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n03 ocf_msdummy:1 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n03 ocf_msdummy:2 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n02 ocf_msdummy:3 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n08 ocf_msdummy:4 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Stopped ocf_msdummy:5 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Stopped ocf_msdummy:6 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n02 ocf_msdummy:7 (ocf::heartbeat:/usr/lib/heartbeat/cts/OCFMSDummy): Started c001n08 diff --git a/pengine/test10/rec-node-10.summary b/pengine/test10/rec-node-10.summary index 73649c850f..02da5d7e50 100644 --- a/pengine/test10/rec-node-10.summary +++ b/pengine/test10/rec-node-10.summary @@ -1,28 +1,28 @@ Current cluster status: Node node1 (uuid1): UNCLEAN (offline) Online: [ node2 ] stonith-1 (stonith:dummy): Stopped - rsc1 (heartbeat:apache): Started node1 - rsc2 (heartbeat:apache): Started node1 + rsc1 (heartbeat:apache): Started node1 (UNCLEAN) + rsc2 (heartbeat:apache): Started node1 (UNCLEAN) Transition Summary: * Start stonith-1 (node2 - blocked) * Stop rsc1 (Started node1 - blocked) * Stop rsc2 (Started node1 - blocked) Executing cluster transition: * Resource action: stonith-1 monitor on node2 * Resource action: rsc1 monitor on node2 * Resource action: rsc2 monitor on node2 * Pseudo action: probe_complete Revised cluster status: Node node1 (uuid1): UNCLEAN (offline) Online: [ node2 ] stonith-1 (stonith:dummy): Stopped - rsc1 (heartbeat:apache): Started node1 - rsc2 (heartbeat:apache): Started node1 + rsc1 (heartbeat:apache): Started node1 (UNCLEAN) + rsc2 (heartbeat:apache): Started node1 (UNCLEAN) diff --git a/pengine/test10/rec-node-4.summary b/pengine/test10/rec-node-4.summary index 60f70fc209..36ae8a0ce7 100644 --- a/pengine/test10/rec-node-4.summary +++ b/pengine/test10/rec-node-4.summary @@ -1,36 +1,36 @@ Current cluster status: Node node1 (uuid1): UNCLEAN (offline) Online: [ node2 ] stonith-1 (stonith:dummy): Stopped - rsc1 (heartbeat:apache): Started node1 - rsc2 (heartbeat:apache): Started node1 + rsc1 (heartbeat:apache): Started node1 (UNCLEAN) + rsc2 (heartbeat:apache): Started node1 (UNCLEAN) Transition Summary: * Start stonith-1 (node2) * Move rsc1 (Started node1 -> node2) * Move rsc2 (Started node1 -> node2) Executing cluster transition: * Resource action: stonith-1 monitor on node2 * Resource action: rsc1 monitor on node2 * Resource action: rsc2 monitor on node2 * Fencing node1 (reboot) * Pseudo action: stonith_complete * Pseudo action: probe_complete * Pseudo action: rsc1_stop_0 * Pseudo action: rsc2_stop_0 * Pseudo action: all_stopped * Resource action: stonith-1 start on node2 * Resource action: rsc1 start on node2 * Resource action: rsc2 start on node2 Revised cluster status: Online: [ node2 ] OFFLINE: [ node1 ] stonith-1 (stonith:dummy): Started node2 rsc1 (heartbeat:apache): Started node2 rsc2 (heartbeat:apache): Started node2 diff --git a/pengine/test10/rec-node-7.summary b/pengine/test10/rec-node-7.summary index 60f70fc209..36ae8a0ce7 100644 --- a/pengine/test10/rec-node-7.summary +++ b/pengine/test10/rec-node-7.summary @@ -1,36 +1,36 @@ Current cluster status: Node node1 (uuid1): UNCLEAN (offline) Online: [ node2 ] stonith-1 (stonith:dummy): Stopped - rsc1 (heartbeat:apache): Started node1 - rsc2 (heartbeat:apache): Started node1 + rsc1 (heartbeat:apache): Started node1 (UNCLEAN) + rsc2 (heartbeat:apache): Started node1 (UNCLEAN) Transition Summary: * Start stonith-1 (node2) * Move rsc1 (Started node1 -> node2) * Move rsc2 (Started node1 -> node2) Executing cluster transition: * Resource action: stonith-1 monitor on node2 * Resource action: rsc1 monitor on node2 * Resource action: rsc2 monitor on node2 * Fencing node1 (reboot) * Pseudo action: stonith_complete * Pseudo action: probe_complete * Pseudo action: rsc1_stop_0 * Pseudo action: rsc2_stop_0 * Pseudo action: all_stopped * Resource action: stonith-1 start on node2 * Resource action: rsc1 start on node2 * Resource action: rsc2 start on node2 Revised cluster status: Online: [ node2 ] OFFLINE: [ node1 ] stonith-1 (stonith:dummy): Started node2 rsc1 (heartbeat:apache): Started node2 rsc2 (heartbeat:apache): Started node2 diff --git a/pengine/test10/rec-node-8.summary b/pengine/test10/rec-node-8.summary index d1cb30e2e0..b78307586d 100644 --- a/pengine/test10/rec-node-8.summary +++ b/pengine/test10/rec-node-8.summary @@ -1,32 +1,32 @@ Current cluster status: Node node1 (uuid1): UNCLEAN (offline) Online: [ node2 ] stonith-1 (stonith:dummy): Stopped - rsc1 (heartbeat:apache): Started node1 - rsc2 (heartbeat:apache): Started node1 + rsc1 (heartbeat:apache): Started node1 (UNCLEAN) + rsc2 (heartbeat:apache): Started node1 (UNCLEAN) rsc3 (heartbeat:apache): Stopped Transition Summary: * Start stonith-1 (node2 - blocked) * Stop rsc1 (Started node1 - blocked) * Stop rsc2 (Started node1 - blocked) * Start rsc3 (node2 - blocked) Executing cluster transition: * Resource action: stonith-1 monitor on node2 * Resource action: rsc1 monitor on node2 * Resource action: rsc2 monitor on node2 * Resource action: rsc3 monitor on node2 * Pseudo action: probe_complete Revised cluster status: Node node1 (uuid1): UNCLEAN (offline) Online: [ node2 ] stonith-1 (stonith:dummy): Stopped - rsc1 (heartbeat:apache): Started node1 - rsc2 (heartbeat:apache): Started node1 + rsc1 (heartbeat:apache): Started node1 (UNCLEAN) + rsc2 (heartbeat:apache): Started node1 (UNCLEAN) rsc3 (heartbeat:apache): Stopped diff --git a/pengine/test10/remote-fence-unclean.summary b/pengine/test10/remote-fence-unclean.summary index eedc5edc49..60e70a51e8 100644 --- a/pengine/test10/remote-fence-unclean.summary +++ b/pengine/test10/remote-fence-unclean.summary @@ -1,46 +1,46 @@ Current cluster status: RemoteNode remote1: UNCLEAN (offline) Online: [ 18builder 18node1 18node2 ] shooter (stonith:fence_xvm): Started 18builder remote1 (ocf::pacemaker:remote): FAILED 18node1 FAKE1 (ocf::heartbeat:Dummy): Started 18node2 - FAKE2 (ocf::heartbeat:Dummy): Started remote1 + FAKE2 (ocf::heartbeat:Dummy): Started remote1 (UNCLEAN) FAKE3 (ocf::heartbeat:Dummy): Started 18builder FAKE4 (ocf::heartbeat:Dummy): Started 18node1 Transition Summary: * Recover remote1 (Started 18node1) * Move FAKE2 (Started remote1 -> 18builder) * Move FAKE3 (Started 18builder -> 18node1) * Move FAKE4 (Started 18node1 -> 18node2) Executing cluster transition: * Resource action: remote1 stop on 18node1 * Resource action: FAKE3 stop on 18builder * Resource action: FAKE4 stop on 18node1 * Fencing remote1 (reboot) * Pseudo action: stonith_complete * Resource action: remote1 start on 18node1 * Resource action: remote1 monitor=60000 on 18node1 * Pseudo action: FAKE2_stop_0 * Resource action: FAKE3 start on 18node1 * Resource action: FAKE4 start on 18node2 * Pseudo action: all_stopped * Resource action: FAKE2 start on 18builder * Resource action: FAKE3 monitor=60000 on 18node1 * Resource action: FAKE4 monitor=60000 on 18node2 * Resource action: FAKE2 monitor=60000 on 18builder Revised cluster status: Online: [ 18builder 18node1 18node2 ] RemoteOnline: [ remote1 ] shooter (stonith:fence_xvm): Started 18builder remote1 (ocf::pacemaker:remote): Started 18node1 FAKE1 (ocf::heartbeat:Dummy): Started 18node2 FAKE2 (ocf::heartbeat:Dummy): Started 18builder FAKE3 (ocf::heartbeat:Dummy): Started 18node1 FAKE4 (ocf::heartbeat:Dummy): Started 18node2 diff --git a/pengine/test10/remote-fence-unclean2.summary b/pengine/test10/remote-fence-unclean2.summary index bfaf77b5dd..78ff784076 100644 --- a/pengine/test10/remote-fence-unclean2.summary +++ b/pengine/test10/remote-fence-unclean2.summary @@ -1,30 +1,30 @@ Current cluster status: Node rhel7-alt1 (1): standby Node rhel7-alt2 (2): standby RemoteNode rhel7-alt4: UNCLEAN (offline) OFFLINE: [ rhel7-alt3 ] shooter (stonith:fence_xvm): Stopped rhel7-alt4 (ocf::pacemaker:remote): Stopped - fake (ocf::heartbeat:Dummy): Started rhel7-alt4 + fake (ocf::heartbeat:Dummy): Started rhel7-alt4 (UNCLEAN) Transition Summary: * Stop fake (rhel7-alt4) Executing cluster transition: * Fencing rhel7-alt4 (reboot) * Pseudo action: stonith_complete * Pseudo action: fake_stop_0 * Pseudo action: all_stopped Revised cluster status: Node rhel7-alt1 (1): standby Node rhel7-alt2 (2): standby OFFLINE: [ rhel7-alt3 ] RemoteOFFLINE: [ rhel7-alt4 ] shooter (stonith:fence_xvm): Stopped rhel7-alt4 (ocf::pacemaker:remote): Stopped fake (ocf::heartbeat:Dummy): Stopped diff --git a/pengine/test10/remote-recover-fail.summary b/pengine/test10/remote-recover-fail.summary index 2811fdc1ff..85847455bd 100644 --- a/pengine/test10/remote-recover-fail.summary +++ b/pengine/test10/remote-recover-fail.summary @@ -1,53 +1,53 @@ Current cluster status: RemoteNode rhel7-auto4: UNCLEAN (offline) Online: [ rhel7-auto2 rhel7-auto3 ] OFFLINE: [ rhel7-auto1 ] shooter (stonith:fence_xvm): Started rhel7-auto3 rhel7-auto4 (ocf::pacemaker:remote): FAILED rhel7-auto2 FAKE1 (ocf::heartbeat:Dummy): Stopped - FAKE2 (ocf::heartbeat:Dummy): Started rhel7-auto4 + FAKE2 (ocf::heartbeat:Dummy): Started rhel7-auto4 (UNCLEAN) FAKE3 (ocf::heartbeat:Dummy): Started rhel7-auto2 FAKE4 (ocf::heartbeat:Dummy): Started rhel7-auto3 FAKE5 (ocf::heartbeat:Dummy): Started rhel7-auto3 - FAKE6 (ocf::heartbeat:Dummy): Started rhel7-auto4 + FAKE6 (ocf::heartbeat:Dummy): Started rhel7-auto4 (UNCLEAN) Transition Summary: * Recover rhel7-auto4 (Started rhel7-auto2) * Start FAKE1 (rhel7-auto2) * Move FAKE2 (Started rhel7-auto4 -> rhel7-auto3) * Move FAKE6 (Started rhel7-auto4 -> rhel7-auto2) Executing cluster transition: * Resource action: rhel7-auto4 stop on rhel7-auto2 * Resource action: FAKE3 monitor=10000 on rhel7-auto2 * Resource action: FAKE4 monitor=10000 on rhel7-auto3 * Fencing rhel7-auto4 (reboot) * Pseudo action: stonith_complete * Resource action: rhel7-auto4 start on rhel7-auto2 * Resource action: FAKE1 start on rhel7-auto2 * Pseudo action: FAKE2_stop_0 * Pseudo action: FAKE6_stop_0 * Pseudo action: all_stopped * Resource action: rhel7-auto4 monitor=60000 on rhel7-auto2 * Resource action: FAKE1 monitor=10000 on rhel7-auto2 * Resource action: FAKE2 start on rhel7-auto3 * Resource action: FAKE6 start on rhel7-auto2 * Resource action: FAKE2 monitor=10000 on rhel7-auto3 * Resource action: FAKE6 monitor=10000 on rhel7-auto2 Revised cluster status: Online: [ rhel7-auto2 rhel7-auto3 ] OFFLINE: [ rhel7-auto1 ] RemoteOnline: [ rhel7-auto4 ] shooter (stonith:fence_xvm): Started rhel7-auto3 rhel7-auto4 (ocf::pacemaker:remote): Started rhel7-auto2 FAKE1 (ocf::heartbeat:Dummy): Started rhel7-auto2 FAKE2 (ocf::heartbeat:Dummy): Started rhel7-auto3 FAKE3 (ocf::heartbeat:Dummy): Started rhel7-auto2 FAKE4 (ocf::heartbeat:Dummy): Started rhel7-auto3 FAKE5 (ocf::heartbeat:Dummy): Started rhel7-auto3 FAKE6 (ocf::heartbeat:Dummy): Started rhel7-auto2 diff --git a/pengine/test10/remote-recover.summary b/pengine/test10/remote-recover.summary index 8fd74806b2..ad4710da08 100644 --- a/pengine/test10/remote-recover.summary +++ b/pengine/test10/remote-recover.summary @@ -1,36 +1,36 @@ Current cluster status: Node rhel7-alt2 (2): standby RemoteNode rhel7-alt4: UNCLEAN (offline) Online: [ rhel7-alt1 ] OFFLINE: [ rhel7-alt3 ] shooter (stonith:fence_xvm): Stopped rhel7-alt4 (ocf::pacemaker:remote): Stopped - fake (ocf::heartbeat:Dummy): Started rhel7-alt4 + fake (ocf::heartbeat:Dummy): Started rhel7-alt4 (UNCLEAN) Transition Summary: * Start shooter (rhel7-alt1) * Start rhel7-alt4 (rhel7-alt1) * Restart fake (Started rhel7-alt4) Executing cluster transition: * Resource action: shooter start on rhel7-alt1 * Resource action: rhel7-alt4 start on rhel7-alt1 * Resource action: fake stop on rhel7-alt4 * Pseudo action: all_stopped * Resource action: shooter monitor=60000 on rhel7-alt1 * Resource action: rhel7-alt4 monitor=60000 on rhel7-alt1 * Resource action: fake start on rhel7-alt4 * Resource action: fake monitor=10000 on rhel7-alt4 Revised cluster status: Node rhel7-alt2 (2): standby Online: [ rhel7-alt1 ] OFFLINE: [ rhel7-alt3 ] RemoteOnline: [ rhel7-alt4 ] shooter (stonith:fence_xvm): Started rhel7-alt1 rhel7-alt4 (ocf::pacemaker:remote): Started rhel7-alt1 fake (ocf::heartbeat:Dummy): Started rhel7-alt4 diff --git a/pengine/test10/stonith-1.summary b/pengine/test10/stonith-1.summary index ed83ad7774..ef904fe13b 100644 --- a/pengine/test10/stonith-1.summary +++ b/pengine/test10/stonith-1.summary @@ -1,111 +1,112 @@ Current cluster status: Node sles-3 (2298606a-6a8c-499a-9d25-76242f7006ec): UNCLEAN (offline) Online: [ sles-1 sles-2 sles-4 ] Resource Group: group-1 r192.168.100.181 (ocf::heartbeat:IPaddr): Started sles-1 r192.168.100.182 (heartbeat:IPaddr): Started sles-1 r192.168.100.183 (ocf::heartbeat:IPaddr): Stopped lsb_dummy (lsb:/usr/lib64/heartbeat/cts/LSBDummy): Started sles-2 - migrator (ocf::heartbeat:Dummy): Started sles-3 + migrator (ocf::heartbeat:Dummy): Started sles-3 (UNCLEAN) rsc_sles-1 (ocf::heartbeat:IPaddr): Started sles-1 rsc_sles-2 (ocf::heartbeat:IPaddr): Started sles-2 - rsc_sles-3 (ocf::heartbeat:IPaddr): Started sles-3 + rsc_sles-3 (ocf::heartbeat:IPaddr): Started sles-3 (UNCLEAN) rsc_sles-4 (ocf::heartbeat:IPaddr): Started sles-4 Clone Set: DoFencing [child_DoFencing] - Started: [ sles-1 sles-2 sles-3 ] + child_DoFencing (stonith:external/vmware): Started sles-3 (UNCLEAN) + Started: [ sles-1 sles-2 ] Stopped: [ sles-4 ] Master/Slave Set: master_rsc_1 [ocf_msdummy] (unique) ocf_msdummy:0 (ocf::heartbeat:Stateful): Stopped ocf_msdummy:1 (ocf::heartbeat:Stateful): Stopped - ocf_msdummy:2 (ocf::heartbeat:Stateful): Started sles-3 + ocf_msdummy:2 (ocf::heartbeat:Stateful): Started sles-3 (UNCLEAN) ocf_msdummy:3 (ocf::heartbeat:Stateful): Stopped ocf_msdummy:4 (ocf::heartbeat:Stateful): Stopped - ocf_msdummy:5 (ocf::heartbeat:Stateful): Started sles-3 + ocf_msdummy:5 (ocf::heartbeat:Stateful): Started sles-3 (UNCLEAN) ocf_msdummy:6 (ocf::heartbeat:Stateful): Stopped ocf_msdummy:7 (ocf::heartbeat:Stateful): Stopped Transition Summary: * Start r192.168.100.183 (sles-1) * Move migrator (Started sles-3 -> sles-4) * Move rsc_sles-3 (Started sles-3 -> sles-4) * Move child_DoFencing:2 (Started sles-3 -> sles-4) * Start ocf_msdummy:0 (sles-4) * Start ocf_msdummy:1 (sles-1) * Move ocf_msdummy:2 (Slave sles-3 -> sles-2) * Start ocf_msdummy:3 (sles-4) * Start ocf_msdummy:4 (sles-1) * Move ocf_msdummy:5 (Slave sles-3 -> sles-2) Executing cluster transition: * Pseudo action: group-1_start_0 * Resource action: r192.168.100.182 monitor=5000 on sles-1 * Resource action: lsb_dummy monitor=5000 on sles-2 * Resource action: rsc_sles-2 monitor=5000 on sles-2 * Resource action: rsc_sles-4 monitor=5000 on sles-4 * Fencing sles-3 (reboot) * Pseudo action: stonith_complete * Resource action: r192.168.100.183 start on sles-1 * Pseudo action: migrator_stop_0 * Pseudo action: rsc_sles-3_stop_0 * Pseudo action: DoFencing_stop_0 * Pseudo action: master_rsc_1_stop_0 * Pseudo action: group-1_running_0 * Resource action: r192.168.100.183 monitor=5000 on sles-1 * Resource action: migrator start on sles-4 * Resource action: rsc_sles-3 start on sles-4 * Pseudo action: child_DoFencing:2_stop_0 * Pseudo action: DoFencing_stopped_0 * Pseudo action: DoFencing_start_0 * Pseudo action: ocf_msdummy:2_stop_0 * Pseudo action: ocf_msdummy:5_stop_0 * Pseudo action: master_rsc_1_stopped_0 * Pseudo action: master_rsc_1_start_0 * Pseudo action: all_stopped * Resource action: migrator monitor=10000 on sles-4 * Resource action: rsc_sles-3 monitor=5000 on sles-4 * Resource action: child_DoFencing:2 start on sles-4 * Pseudo action: DoFencing_running_0 * Resource action: ocf_msdummy:0 start on sles-4 * Resource action: ocf_msdummy:1 start on sles-1 * Resource action: ocf_msdummy:2 start on sles-2 * Resource action: ocf_msdummy:3 start on sles-4 * Resource action: ocf_msdummy:4 start on sles-1 * Resource action: ocf_msdummy:5 start on sles-2 * Pseudo action: master_rsc_1_running_0 * Resource action: child_DoFencing:2 monitor=60000 on sles-4 * Resource action: ocf_msdummy:0 monitor=5000 on sles-4 * Resource action: ocf_msdummy:1 monitor=5000 on sles-1 * Resource action: ocf_msdummy:2 monitor=5000 on sles-2 * Resource action: ocf_msdummy:3 monitor=5000 on sles-4 * Resource action: ocf_msdummy:4 monitor=5000 on sles-1 * Resource action: ocf_msdummy:5 monitor=5000 on sles-2 Revised cluster status: Online: [ sles-1 sles-2 sles-4 ] OFFLINE: [ sles-3 ] Resource Group: group-1 r192.168.100.181 (ocf::heartbeat:IPaddr): Started sles-1 r192.168.100.182 (heartbeat:IPaddr): Started sles-1 r192.168.100.183 (ocf::heartbeat:IPaddr): Started sles-1 lsb_dummy (lsb:/usr/lib64/heartbeat/cts/LSBDummy): Started sles-2 migrator (ocf::heartbeat:Dummy): Started sles-4 rsc_sles-1 (ocf::heartbeat:IPaddr): Started sles-1 rsc_sles-2 (ocf::heartbeat:IPaddr): Started sles-2 rsc_sles-3 (ocf::heartbeat:IPaddr): Started sles-4 rsc_sles-4 (ocf::heartbeat:IPaddr): Started sles-4 Clone Set: DoFencing [child_DoFencing] Started: [ sles-1 sles-2 sles-4 ] Stopped: [ sles-3 ] Master/Slave Set: master_rsc_1 [ocf_msdummy] (unique) ocf_msdummy:0 (ocf::heartbeat:Stateful): Started sles-4 ocf_msdummy:1 (ocf::heartbeat:Stateful): Started sles-1 ocf_msdummy:2 (ocf::heartbeat:Stateful): Started sles-2 ocf_msdummy:3 (ocf::heartbeat:Stateful): Started sles-4 ocf_msdummy:4 (ocf::heartbeat:Stateful): Started sles-1 ocf_msdummy:5 (ocf::heartbeat:Stateful): Started sles-2 ocf_msdummy:6 (ocf::heartbeat:Stateful): Stopped ocf_msdummy:7 (ocf::heartbeat:Stateful): Stopped diff --git a/pengine/test10/stop-failure-no-fencing.summary b/pengine/test10/stop-failure-no-fencing.summary index 4fc1ed1cff..1148f8291f 100644 --- a/pengine/test10/stop-failure-no-fencing.summary +++ b/pengine/test10/stop-failure-no-fencing.summary @@ -1,29 +1,29 @@ Current cluster status: Node pcmk-3 (103): UNCLEAN (offline) Node pcmk-4 (104): UNCLEAN (offline) Online: [ pcmk-1 pcmk-2 ] Clone Set: dlm-clone [dlm] Stopped: [ pcmk-1 pcmk-2 pcmk-3 pcmk-4 ] Clone Set: clvm-clone [clvm] - clvm (lsb:clvmd): FAILED pcmk-3 (unmanaged) + clvm (lsb:clvmd): FAILED pcmk-3 (UNCLEAN) (unmanaged) Stopped: [ pcmk-1 pcmk-2 pcmk-3 pcmk-4 ] ClusterIP (ocf::heartbeat:IPaddr2): Stopped Transition Summary: Executing cluster transition: Revised cluster status: Node pcmk-3 (103): UNCLEAN (offline) Node pcmk-4 (104): UNCLEAN (offline) Online: [ pcmk-1 pcmk-2 ] Clone Set: dlm-clone [dlm] Stopped: [ pcmk-1 pcmk-2 pcmk-3 pcmk-4 ] Clone Set: clvm-clone [clvm] - clvm (lsb:clvmd): FAILED pcmk-3 (unmanaged) + clvm (lsb:clvmd): FAILED pcmk-3 (UNCLEAN) (unmanaged) Stopped: [ pcmk-1 pcmk-2 pcmk-3 pcmk-4 ] ClusterIP (ocf::heartbeat:IPaddr2): Stopped diff --git a/pengine/test10/stop-failure-no-quorum.summary b/pengine/test10/stop-failure-no-quorum.summary index d57bdc6b32..a5a0949e84 100644 --- a/pengine/test10/stop-failure-no-quorum.summary +++ b/pengine/test10/stop-failure-no-quorum.summary @@ -1,45 +1,45 @@ Current cluster status: Node pcmk-2 (102): UNCLEAN (online) Node pcmk-3 (103): UNCLEAN (offline) Node pcmk-4 (104): UNCLEAN (offline) Online: [ pcmk-1 ] Clone Set: dlm-clone [dlm] Stopped: [ pcmk-1 pcmk-2 pcmk-3 pcmk-4 ] Clone Set: clvm-clone [clvm] clvm (lsb:clvmd): FAILED pcmk-2 - clvm (lsb:clvmd): FAILED pcmk-3 (unmanaged) + clvm (lsb:clvmd): FAILED pcmk-3 (UNCLEAN) (unmanaged) Stopped: [ pcmk-1 pcmk-3 pcmk-4 ] ClusterIP (ocf::heartbeat:IPaddr2): Stopped Fencing (stonith:fence_xvm): Stopped Transition Summary: * Start dlm:0 (pcmk-1 - blocked) * Stop clvm:0 (pcmk-2) * Start clvm:2 (pcmk-1 - blocked) * Start ClusterIP (pcmk-1 - blocked) * Start Fencing (pcmk-1 - blocked) Executing cluster transition: * Fencing pcmk-2 (reboot) * Pseudo action: stonith_complete * Pseudo action: clvm-clone_stop_0 * Pseudo action: clvm_stop_0 * Pseudo action: clvm-clone_stopped_0 * Pseudo action: all_stopped Revised cluster status: Node pcmk-3 (103): UNCLEAN (offline) Node pcmk-4 (104): UNCLEAN (offline) Online: [ pcmk-1 ] OFFLINE: [ pcmk-2 ] Clone Set: dlm-clone [dlm] Stopped: [ pcmk-1 pcmk-2 pcmk-3 pcmk-4 ] Clone Set: clvm-clone [clvm] - clvm (lsb:clvmd): FAILED pcmk-3 (unmanaged) + clvm (lsb:clvmd): FAILED pcmk-3 (UNCLEAN) (unmanaged) Stopped: [ pcmk-1 pcmk-2 pcmk-3 pcmk-4 ] ClusterIP (ocf::heartbeat:IPaddr2): Stopped Fencing (stonith:fence_xvm): Stopped diff --git a/pengine/test10/unrunnable-1.summary b/pengine/test10/unrunnable-1.summary index e9597f3ae1..38a909d08a 100644 --- a/pengine/test10/unrunnable-1.summary +++ b/pengine/test10/unrunnable-1.summary @@ -1,66 +1,66 @@ Current cluster status: Node c001n02 (e9bdfde9-01b0-421f-acd8-8a65a53e775f): UNCLEAN (offline) Online: [ c001n03 ] DcIPaddr (ocf::heartbeat:IPaddr): Stopped Resource Group: group-1 child_192.168.100.181 (ocf::heartbeat:IPaddr): Stopped child_192.168.100.182 (ocf::heartbeat:IPaddr): Stopped child_192.168.100.183 (ocf::heartbeat:IPaddr): Stopped rsc_c001n08 (ocf::heartbeat:IPaddr): Stopped rsc_c001n02 (ocf::heartbeat:IPaddr): Stopped rsc_c001n03 (ocf::heartbeat:IPaddr): Stopped rsc_c001n01 (ocf::heartbeat:IPaddr): Stopped Clone Set: DoFencing [child_DoFencing] (unique) child_DoFencing:0 (stonith:ssh): Started c001n03 - child_DoFencing:1 (stonith:ssh): Started c001n02 + child_DoFencing:1 (stonith:ssh): Started c001n02 (UNCLEAN) child_DoFencing:2 (stonith:ssh): Stopped child_DoFencing:3 (stonith:ssh): Stopped Transition Summary: * Start DcIPaddr (c001n03 - blocked) * Start child_192.168.100.181 (c001n03 - blocked) * Start child_192.168.100.182 (c001n03 - blocked) * Start child_192.168.100.183 (c001n03 - blocked) * Start rsc_c001n08 (c001n03 - blocked) * Start rsc_c001n02 (c001n03 - blocked) * Start rsc_c001n03 (c001n03 - blocked) * Start rsc_c001n01 (c001n03 - blocked) * Stop child_DoFencing:1 (c001n02 - blocked) Executing cluster transition: * Resource action: DcIPaddr monitor on c001n03 * Resource action: child_192.168.100.181 monitor on c001n03 * Resource action: child_192.168.100.182 monitor on c001n03 * Resource action: child_192.168.100.183 monitor on c001n03 * Resource action: rsc_c001n08 monitor on c001n03 * Resource action: rsc_c001n02 monitor on c001n03 * Resource action: rsc_c001n03 monitor on c001n03 * Resource action: rsc_c001n01 monitor on c001n03 * Resource action: child_DoFencing:1 monitor on c001n03 * Resource action: child_DoFencing:2 monitor on c001n03 * Resource action: child_DoFencing:3 monitor on c001n03 * Pseudo action: DoFencing_stop_0 * Pseudo action: probe_complete * Pseudo action: DoFencing_stopped_0 Revised cluster status: Node c001n02 (e9bdfde9-01b0-421f-acd8-8a65a53e775f): UNCLEAN (offline) Online: [ c001n03 ] DcIPaddr (ocf::heartbeat:IPaddr): Stopped Resource Group: group-1 child_192.168.100.181 (ocf::heartbeat:IPaddr): Stopped child_192.168.100.182 (ocf::heartbeat:IPaddr): Stopped child_192.168.100.183 (ocf::heartbeat:IPaddr): Stopped rsc_c001n08 (ocf::heartbeat:IPaddr): Stopped rsc_c001n02 (ocf::heartbeat:IPaddr): Stopped rsc_c001n03 (ocf::heartbeat:IPaddr): Stopped rsc_c001n01 (ocf::heartbeat:IPaddr): Stopped Clone Set: DoFencing [child_DoFencing] (unique) child_DoFencing:0 (stonith:ssh): Started c001n03 - child_DoFencing:1 (stonith:ssh): Started c001n02 + child_DoFencing:1 (stonith:ssh): Started c001n02 (UNCLEAN) child_DoFencing:2 (stonith:ssh): Stopped child_DoFencing:3 (stonith:ssh): Stopped diff --git a/pengine/test10/whitebox-imply-stop-on-fence.summary b/pengine/test10/whitebox-imply-stop-on-fence.summary index 79e77de83a..3bb157273f 100644 --- a/pengine/test10/whitebox-imply-stop-on-fence.summary +++ b/pengine/test10/whitebox-imply-stop-on-fence.summary @@ -1,88 +1,91 @@ Current cluster status: Node kiff-01 (1): UNCLEAN (offline) Online: [ kiff-02 ] Containers: [ lxc-01_kiff-01:R-lxc-01_kiff-01 lxc-01_kiff-02:R-lxc-01_kiff-02 lxc-02_kiff-01:R-lxc-02_kiff-01 lxc-02_kiff-02:R-lxc-02_kiff-02 ] fence-kiff-01 (stonith:fence_ipmilan): Started kiff-02 - fence-kiff-02 (stonith:fence_ipmilan): Started kiff-01 + fence-kiff-02 (stonith:fence_ipmilan): Started kiff-01 (UNCLEAN) Clone Set: dlm-clone [dlm] - Started: [ kiff-01 kiff-02 ] + dlm (ocf::pacemaker:controld): Started kiff-01 (UNCLEAN) + Started: [ kiff-02 ] Stopped: [ lxc-01_kiff-01 lxc-01_kiff-02 lxc-02_kiff-01 lxc-02_kiff-02 ] Clone Set: clvmd-clone [clvmd] - Started: [ kiff-01 kiff-02 ] + clvmd (ocf::heartbeat:clvm): Started kiff-01 (UNCLEAN) + Started: [ kiff-02 ] Stopped: [ lxc-01_kiff-01 lxc-01_kiff-02 lxc-02_kiff-01 lxc-02_kiff-02 ] Clone Set: shared0-clone [shared0] - Started: [ kiff-01 kiff-02 ] + shared0 (ocf::heartbeat:Filesystem): Started kiff-01 (UNCLEAN) + Started: [ kiff-02 ] Stopped: [ lxc-01_kiff-01 lxc-01_kiff-02 lxc-02_kiff-01 lxc-02_kiff-02 ] - R-lxc-01_kiff-01 (ocf::heartbeat:VirtualDomain): Started kiff-01 - R-lxc-02_kiff-01 (ocf::heartbeat:VirtualDomain): Started kiff-01 + R-lxc-01_kiff-01 (ocf::heartbeat:VirtualDomain): Started kiff-01 (UNCLEAN) + R-lxc-02_kiff-01 (ocf::heartbeat:VirtualDomain): Started kiff-01 (UNCLEAN) R-lxc-01_kiff-02 (ocf::heartbeat:VirtualDomain): Started kiff-02 R-lxc-02_kiff-02 (ocf::heartbeat:VirtualDomain): Started kiff-02 vm-fs (ocf::heartbeat:Filesystem): Started lxc-01_kiff-01 Transition Summary: * Move fence-kiff-02 (Started kiff-01 -> kiff-02) * Stop dlm:0 (kiff-01) * Stop clvmd:0 (kiff-01) * Stop shared0:0 (kiff-01) * Move R-lxc-01_kiff-01 (Started kiff-01 -> kiff-02) * Move R-lxc-02_kiff-01 (Started kiff-01 -> kiff-02) * Restart vm-fs (Started lxc-01_kiff-01) * Move lxc-01_kiff-01 (Started kiff-01 -> kiff-02) * Move lxc-02_kiff-01 (Started kiff-01 -> kiff-02) Executing cluster transition: * Fencing kiff-01 (reboot) * Pseudo action: stonith_complete * Pseudo action: fence-kiff-02_stop_0 * Pseudo action: vm-fs_stop_0 * Pseudo action: lxc-01_kiff-01_stop_0 * Pseudo action: lxc-02_kiff-01_stop_0 * Resource action: fence-kiff-02 start on kiff-02 * Pseudo action: R-lxc-01_kiff-01_stop_0 * Pseudo action: R-lxc-02_kiff-01_stop_0 * Resource action: fence-kiff-02 monitor=60000 on kiff-02 * Pseudo action: shared0-clone_stop_0 * Resource action: R-lxc-01_kiff-01 start on kiff-02 * Resource action: R-lxc-02_kiff-01 start on kiff-02 * Resource action: lxc-01_kiff-01 start on kiff-02 * Resource action: lxc-02_kiff-01 start on kiff-02 * Pseudo action: shared0_stop_0 * Pseudo action: shared0-clone_stopped_0 * Resource action: R-lxc-01_kiff-01 monitor=10000 on kiff-02 * Resource action: R-lxc-02_kiff-01 monitor=10000 on kiff-02 * Resource action: vm-fs start on lxc-01_kiff-01 * Resource action: vm-fs monitor=20000 on lxc-01_kiff-01 * Resource action: lxc-01_kiff-01 monitor=30000 on kiff-02 * Resource action: lxc-02_kiff-01 monitor=30000 on kiff-02 * Pseudo action: clvmd-clone_stop_0 * Pseudo action: clvmd_stop_0 * Pseudo action: clvmd-clone_stopped_0 * Pseudo action: dlm-clone_stop_0 * Pseudo action: dlm_stop_0 * Pseudo action: dlm-clone_stopped_0 * Pseudo action: all_stopped Revised cluster status: Online: [ kiff-02 ] OFFLINE: [ kiff-01 ] Containers: [ lxc-01_kiff-01:R-lxc-01_kiff-01 lxc-01_kiff-02:R-lxc-01_kiff-02 lxc-02_kiff-01:R-lxc-02_kiff-01 lxc-02_kiff-02:R-lxc-02_kiff-02 ] fence-kiff-01 (stonith:fence_ipmilan): Started kiff-02 fence-kiff-02 (stonith:fence_ipmilan): Started kiff-02 Clone Set: dlm-clone [dlm] Started: [ kiff-02 ] Stopped: [ kiff-01 lxc-01_kiff-01 lxc-01_kiff-02 lxc-02_kiff-01 lxc-02_kiff-02 ] Clone Set: clvmd-clone [clvmd] Started: [ kiff-02 ] Stopped: [ kiff-01 lxc-01_kiff-01 lxc-01_kiff-02 lxc-02_kiff-01 lxc-02_kiff-02 ] Clone Set: shared0-clone [shared0] Started: [ kiff-02 ] Stopped: [ kiff-01 lxc-01_kiff-01 lxc-01_kiff-02 lxc-02_kiff-01 lxc-02_kiff-02 ] R-lxc-01_kiff-01 (ocf::heartbeat:VirtualDomain): Started kiff-02 R-lxc-02_kiff-01 (ocf::heartbeat:VirtualDomain): Started kiff-02 R-lxc-01_kiff-02 (ocf::heartbeat:VirtualDomain): Started kiff-02 R-lxc-02_kiff-02 (ocf::heartbeat:VirtualDomain): Started kiff-02 vm-fs (ocf::heartbeat:Filesystem): Started lxc-01_kiff-01