diff --git a/cts/cts-exec.in b/cts/cts-exec.in index cc1d0713a3..a89f7477a6 100644 --- a/cts/cts-exec.in +++ b/cts/cts-exec.in @@ -1,1268 +1,1268 @@ #!@PYTHON@ """ Regression tests for Pacemaker's pacemaker-execd """ __copyright__ = "Copyright 2012-2021 the Pacemaker project contributors" __license__ = "GNU General Public License version 2 or later (GPLv2+) WITHOUT ANY WARRANTY" import io import os import stat import sys import subprocess import shlex import shutil import tempfile import time # Where to find test binaries # Prefer the source tree if available BUILD_DIR = "@abs_top_builddir@" TEST_DIR = sys.path[0] SBIN_DIR = "@sbindir@" # Check whether Pacemaker Remote support was built REMOTE_ENABLED = "@PC_NAME_GNUTLS@" != "" # File permissions for executable scripts we create EXECMODE = stat.S_IRUSR | stat.S_IXUSR | stat.S_IRGRP | stat.S_IXGRP | stat.S_IROTH | stat.S_IXOTH # These values must be kept in sync with include/crm/crm.h class CrmExit(object): OK = 0 ERROR = 1 INVALID_PARAM = 2 UNIMPLEMENT_FEATURE = 3 INSUFFICIENT_PRIV = 4 NOT_INSTALLED = 5 NOT_CONFIGURED = 6 NOT_RUNNING = 7 USAGE = 64 DATAERR = 65 NOINPUT = 66 NOUSER = 67 NOHOST = 68 UNAVAILABLE = 69 SOFTWARE = 70 OSERR = 71 OSFILE = 72 CANTCREAT = 73 IOERR = 74 TEMPFAIL = 75 PROTOCOL = 76 NOPERM = 77 CONFIG = 78 FATAL = 100 PANIC = 101 DISCONNECT = 102 SOLO = 103 DIGEST = 104 NOSUCH = 105 QUORUM = 106 UNSAFE = 107 EXISTS = 108 MULTIPLE = 109 OLD = 110 TIMEOUT = 124 MAX = 255 def update_path(): """ Set the PATH environment variable appropriately for the tests """ new_path = os.environ['PATH'] if os.path.exists("%s/cts-exec.in" % TEST_DIR): print("Running tests from the source tree: %s (%s)" % (BUILD_DIR, TEST_DIR)) # For pacemaker-execd, cts-exec-helper, and pacemaker-remoted new_path = "%s/daemons/execd:%s" % (BUILD_DIR, new_path) new_path = "%s/tools:%s" % (BUILD_DIR, new_path) # For crm_resource # For pacemaker-fenced new_path = "%s/daemons/fenced:%s" % (BUILD_DIR, new_path) # For cts-support new_path = "%s/cts/support:%s" % (BUILD_DIR, new_path) else: print("Running tests from the install tree: @CRM_DAEMON_DIR@ (not %s)" % TEST_DIR) # For cts-exec-helper, cts-support, pacemaker-execd, pacemaker-fenced, # and pacemaker-remoted new_path = "@CRM_DAEMON_DIR@:%s" % (new_path) print('Using PATH="{}"'.format(new_path)) os.environ['PATH'] = new_path def output_from_command(command): """ Run a command, and return its standard output. """ test = subprocess.Popen(shlex.split(command), stdout=subprocess.PIPE) test.wait() output = test.communicate()[0].decode(sys.stdout.encoding) return output.split("\n") class TestError(Exception): """ Base class for exceptions in this module """ pass class ExitCodeError(TestError): """ Exception raised when command exit status is unexpected """ def __init__(self, exit_code): self.exit_code = exit_code def __str__(self): return repr(self.exit_code) class OutputNotFoundError(TestError): """ Exception raised when command output does not contain wanted string """ def __init__(self, output): self.output = output def __str__(self): return repr(self.output) class OutputFoundError(TestError): """ Exception raised when command output contains unwanted string """ def __init__(self, output): self.output = output def __str__(self): return repr(self.output) class Test(object): """ Executor for a single pacemaker-execd regression test """ def __init__(self, name, description, verbose=0, tls=0, timeout=2, force_wait=0, logdir="/tmp"): self.name = name self.description = description self.cmds = [] if tls: self.daemon_location = "pacemaker-remoted" else: self.daemon_location = "pacemaker-execd" self.logpath = os.path.join(logdir, self.daemon_location + ".log") self.test_tool_location = "cts-exec-helper" self.verbose = verbose self.tls = tls self.timeout = timeout self.force_wait = force_wait self.result_txt = "" self.cmd_tool_output = "" self.result_exitcode = CrmExit.OK self.execd_process = None self.stonith_process = None self.executed = 0 def __new_cmd(self, cmd, args, exitcode, stdout_match="", no_wait=0, stdout_negative_match="", kill=None): """ Add a command to be executed as part of this test """ if self.verbose and cmd == self.test_tool_location: args = args + " -V " if (cmd == self.test_tool_location) and self.tls: args = args + " -S " self.cmds.append( { "cmd" : cmd, "kill" : kill, "args" : args, "expected_exitcode" : exitcode, "stdout_match" : stdout_match, "stdout_negative_match" : stdout_negative_match, "no_wait" : no_wait, "cmd_output" : "", } ) def start_environment(self): """ Prepare the host for running a test """ ### make sure we are in full control here ### cmd = shlex.split("killall -q -9 pacemaker-fenced lt-pacemaker-fenced pacemaker-execd lt-pacemaker-execd cts-exec-helper lt-cts-exec-helper pacemaker-remoted") test = subprocess.Popen(cmd, stdout=subprocess.PIPE) test.wait() additional_args = "" if self.tls == 0: self.stonith_process = subprocess.Popen(shlex.split("pacemaker-fenced -s")) if self.verbose: additional_args = additional_args + " -V" self.execd_process = subprocess.Popen(shlex.split("%s %s -l %s" % (self.daemon_location, additional_args, self.logpath))) logfile = None init_time = time.time() update_time = init_time while True: time.sleep(0.1) if self.force_wait == 0 and logfile == None \ and os.path.exists(self.logpath): logfile = io.open(self.logpath, 'rt', encoding = "ISO-8859-1") if self.force_wait == 0 and logfile != None: for line in logfile.readlines(): if "successfully started" in line: return now = time.time() if self.timeout > 0 and (now - init_time) >= self.timeout: if self.force_wait == 0: print("\tDaemon %s doesn't seem to have been initialized within %fs." "\n\tConsider specifying a longer '--timeout' value." %(self.daemon_location, self.timeout)) return if self.verbose and (now - update_time) >= 5: print("Waiting for %s to be initialized: %fs ..." %(self.daemon_location, now - init_time)) update_time = now def clean_environment(self): """ Clean up the host after running a test """ if self.execd_process: self.execd_process.terminate() self.execd_process.wait() if self.verbose: print("Daemon output") logfile = io.open(self.logpath, 'rt', errors='replace') for line in logfile: print(line.strip().encode('utf-8', 'replace')) if self.stonith_process: self.stonith_process.terminate() self.stonith_process.wait() self.execd_process = None self.stonith_process = None def add_sys_cmd(self, cmd, args): """ Add a simple command to be executed as part of this test """ self.__new_cmd(cmd, args, CrmExit.OK, "") def add_cmd_check_stdout(self, args, match, no_match=""): """ Add a command with expected output to be executed as part of this test """ self.__new_cmd(self.test_tool_location, args, CrmExit.OK, match, 0, no_match) def add_cmd(self, args): """ Add a cts-exec-helper command to be executed as part of this test """ self.__new_cmd(self.test_tool_location, args, CrmExit.OK, "") def add_cmd_and_kill(self, kill_proc, args): """ Add a cts-exec-helper command and system command to be executed as part of this test """ self.__new_cmd(self.test_tool_location, args, CrmExit.OK, "", kill=kill_proc) def add_expected_fail_cmd(self, args, exitcode=CrmExit.ERROR): """ Add a cts-exec-helper command to be executed as part of this test and expected to fail """ self.__new_cmd(self.test_tool_location, args, exitcode, "") def get_exitcode(self): """ Return the exit status of the last test execution """ return self.result_exitcode def print_result(self, filler): """ Print the result of the last test execution """ print("%s%s" % (filler, self.result_txt)) def run_cmd(self, args): """ Execute a command as part of this test """ cmd = shlex.split(args['args']) cmd.insert(0, args['cmd']) if self.verbose: print("\n\nRunning: "+" ".join(cmd)) test = subprocess.Popen(cmd, stdout=subprocess.PIPE) if args['kill']: if self.verbose: print("Also running: "+args['kill']) ### Typically, the kill argument is used to detect some sort of ### failure. Without yielding for a few seconds here, the process ### launched earlier that is listening for the failure may not have ### time to connect to pacemaker-execd. time.sleep(2) subprocess.Popen(shlex.split(args['kill'])) if args['no_wait'] == 0: test.wait() else: return CrmExit.OK output = test.communicate()[0].decode(sys.stdout.encoding) args['cmd_output'] = output if test.returncode != args['expected_exitcode']: raise ExitCodeError(test.returncode) if args['stdout_match'] != "" and output.count(args['stdout_match']) == 0: raise OutputNotFoundError(output) if args['stdout_negative_match'] != "" and output.count(args['stdout_negative_match']) != 0: raise OutputFoundError(output) def set_error(self, step, cmd): """ Record failure of this test """ msg = "FAILURE - '%s' failed at step %d. Command: %s %s" self.result_txt = msg % (self.name, step, cmd['cmd'], cmd['args']) self.result_exitcode = CrmExit.ERROR def run(self): """ Execute this test. """ res = 0 i = 1 if self.tls and self.name.count("stonith") != 0: self.result_txt = "SKIPPED - '%s' - disabled when testing pacemaker_remote" % (self.name) print(self.result_txt) return res self.start_environment() if self.verbose: print("\n--- START TEST - %s" % self.name) self.result_txt = "SUCCESS - '%s'" % (self.name) self.result_exitcode = CrmExit.OK for cmd in self.cmds: try: self.run_cmd(cmd) except ExitCodeError as e: print(cmd['cmd_output']) print("Step %d FAILED - command returned %s, expected %d" % (i, e, cmd['expected_exitcode'])) self.set_error(i, cmd); break except OutputNotFoundError as e: print("Step %d FAILED - '%s' was not found in command output: %s" % (i, cmd['stdout_match'], e)) self.set_error(i, cmd); break except OutputFoundError as e: print("Step %d FAILED - '%s' was found in command output: %s" % (i, cmd['stdout_negative_match'], e)) self.set_error(i, cmd); break if self.verbose: print(cmd['cmd_output'].strip()) print("Step %d SUCCESS" % (i)) i = i + 1 self.clean_environment() print(self.result_txt) if self.verbose: print("--- END TEST - %s\n" % self.name) self.executed = 1 return res class Tests(object): """ Collection of all pacemaker-execd regression tests """ def __init__(self, verbose=0, tls=0, timeout=2, force_wait=0, logdir="/tmp"): self.tests = [] self.verbose = verbose self.tls = tls self.timeout = timeout self.force_wait = force_wait self.logdir = logdir self.rsc_classes = output_from_command("crm_resource --list-standards") self.rsc_classes = self.rsc_classes[:-1] # Strip trailing empty line self.installed_files = [] self.action_timeout = " -t 9000 " if self.tls: self.rsc_classes.remove("stonith") if "systemd" in self.rsc_classes: try: # This code doesn't need this import, but pacemaker-cts-dummyd # does, so ensure the dependency is available rather than cause # all systemd tests to fail. import systemd.daemon except ImportError: print("Python systemd bindings not found.") print("The tests for systemd class are not going to be run.") self.rsc_classes.remove("systemd") print("Testing resource classes", repr(self.rsc_classes)) self.common_cmds = { "ocf_reg_line" : "-c register_rsc -r ocf_test_rsc "+self.action_timeout+" -C ocf -P pacemaker -T Dummy", "ocf_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:ocf_test_rsc action:none rc:ok op_status:complete\"", "ocf_unreg_line" : "-c unregister_rsc -r \"ocf_test_rsc\" "+self.action_timeout, "ocf_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:ocf_test_rsc action:none rc:ok op_status:complete\"", "ocf_start_line" : "-c exec -r \"ocf_test_rsc\" -a \"start\" "+self.action_timeout, "ocf_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:ocf_test_rsc action:start rc:ok op_status:complete\" ", "ocf_stop_line" : "-c exec -r \"ocf_test_rsc\" -a \"stop\" "+self.action_timeout, "ocf_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:ocf_test_rsc action:stop rc:ok op_status:complete\" ", "ocf_monitor_line" : '-c exec -r ocf_test_rsc -a monitor -i 2s ' + self.action_timeout, "ocf_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:ocf_test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout, "ocf_cancel_line" : '-c cancel -r ocf_test_rsc -a monitor -i 2s ' + self.action_timeout, "ocf_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:ocf_test_rsc action:monitor rc:ok op_status:Cancelled\" ", "systemd_reg_line" : "-c register_rsc -r systemd_test_rsc " + self.action_timeout + " -C systemd -T pacemaker-cts-dummyd@3", "systemd_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:systemd_test_rsc action:none rc:ok op_status:complete\"", "systemd_unreg_line" : "-c unregister_rsc -r \"systemd_test_rsc\" "+self.action_timeout, "systemd_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:systemd_test_rsc action:none rc:ok op_status:complete\"", "systemd_start_line" : "-c exec -r \"systemd_test_rsc\" -a \"start\" "+self.action_timeout, "systemd_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:systemd_test_rsc action:start rc:ok op_status:complete\" ", "systemd_stop_line" : "-c exec -r \"systemd_test_rsc\" -a \"stop\" "+self.action_timeout, "systemd_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:systemd_test_rsc action:stop rc:ok op_status:complete\" ", "systemd_monitor_line" : '-c exec -r systemd_test_rsc -a monitor -i 2s ' + self.action_timeout, "systemd_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:systemd_test_rsc action:monitor rc:ok op_status:complete\" -t 15000 ", "systemd_cancel_line" : '-c cancel -r systemd_test_rsc -a monitor -i 2s ' + self.action_timeout, "systemd_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:systemd_test_rsc action:monitor rc:ok op_status:Cancelled\" ", "upstart_reg_line" : "-c register_rsc -r upstart_test_rsc "+self.action_timeout+" -C upstart -T pacemaker-cts-dummyd", "upstart_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:upstart_test_rsc action:none rc:ok op_status:complete\"", "upstart_unreg_line" : "-c unregister_rsc -r \"upstart_test_rsc\" "+self.action_timeout, "upstart_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:upstart_test_rsc action:none rc:ok op_status:complete\"", "upstart_start_line" : "-c exec -r \"upstart_test_rsc\" -a \"start\" "+self.action_timeout, "upstart_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:upstart_test_rsc action:start rc:ok op_status:complete\" ", "upstart_stop_line" : "-c exec -r \"upstart_test_rsc\" -a \"stop\" "+self.action_timeout, "upstart_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:upstart_test_rsc action:stop rc:ok op_status:complete\" ", "upstart_monitor_line" : '-c exec -r upstart_test_rsc -a monitor -i 2s ' + self.action_timeout, "upstart_monitor_event" : '-l "NEW_EVENT event_type:exec_complete rsc_id:upstart_test_rsc action:monitor rc:ok op_status:complete" -t 15000', "upstart_cancel_line" : '-c cancel -r upstart_test_rsc -a monitor -i 2s ' + self.action_timeout, "upstart_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:upstart_test_rsc action:monitor rc:ok op_status:Cancelled\" ", "service_reg_line" : "-c register_rsc -r service_test_rsc "+self.action_timeout+" -C service -T LSBDummy", "service_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:service_test_rsc action:none rc:ok op_status:complete\"", "service_unreg_line" : "-c unregister_rsc -r \"service_test_rsc\" "+self.action_timeout, "service_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:service_test_rsc action:none rc:ok op_status:complete\"", "service_start_line" : "-c exec -r \"service_test_rsc\" -a \"start\" "+self.action_timeout, "service_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:service_test_rsc action:start rc:ok op_status:complete\" ", "service_stop_line" : "-c exec -r \"service_test_rsc\" -a \"stop\" "+self.action_timeout, "service_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:service_test_rsc action:stop rc:ok op_status:complete\" ", "service_monitor_line" : '-c exec -r service_test_rsc -a monitor -i 2s ' + self.action_timeout, "service_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:service_test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout, "service_cancel_line" : '-c cancel -r service_test_rsc -a monitor -i 2s ' + self.action_timeout, "service_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:service_test_rsc action:monitor rc:ok op_status:Cancelled\" ", "lsb_reg_line" : "-c register_rsc -r lsb_test_rsc "+self.action_timeout+" -C lsb -T LSBDummy", "lsb_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:lsb_test_rsc action:none rc:ok op_status:complete\" ", "lsb_unreg_line" : "-c unregister_rsc -r \"lsb_test_rsc\" "+self.action_timeout, "lsb_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:lsb_test_rsc action:none rc:ok op_status:complete\"", "lsb_start_line" : "-c exec -r \"lsb_test_rsc\" -a \"start\" "+self.action_timeout, "lsb_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:lsb_test_rsc action:start rc:ok op_status:complete\" ", "lsb_stop_line" : "-c exec -r \"lsb_test_rsc\" -a \"stop\" "+self.action_timeout, "lsb_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:lsb_test_rsc action:stop rc:ok op_status:complete\" ", "lsb_monitor_line" : '-c exec -r lsb_test_rsc -a status -i 2s ' + self.action_timeout, "lsb_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:lsb_test_rsc action:status rc:ok op_status:complete\" "+self.action_timeout, "lsb_cancel_line" : '-c cancel -r lsb_test_rsc -a status -i 2s ' + self.action_timeout, "lsb_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:lsb_test_rsc action:status rc:ok op_status:Cancelled\" ", "stonith_reg_line" : "-c register_rsc -r stonith_test_rsc " + self.action_timeout + " -C stonith -P pacemaker -T fence_dummy", "stonith_reg_event" : "-l \"NEW_EVENT event_type:register rsc_id:stonith_test_rsc action:none rc:ok op_status:complete\" ", "stonith_unreg_line" : "-c unregister_rsc -r \"stonith_test_rsc\" "+self.action_timeout, "stonith_unreg_event" : "-l \"NEW_EVENT event_type:unregister rsc_id:stonith_test_rsc action:none rc:ok op_status:complete\"", "stonith_start_line" : '-c exec -r stonith_test_rsc -a start ' + self.action_timeout, "stonith_start_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:start rc:ok op_status:complete\" ", "stonith_stop_line" : "-c exec -r \"stonith_test_rsc\" -a \"stop\" "+self.action_timeout, "stonith_stop_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:stop rc:ok op_status:complete\" ", "stonith_monitor_line" : '-c exec -r stonith_test_rsc -a monitor -i 2s ' + self.action_timeout, "stonith_monitor_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout, "stonith_cancel_line" : '-c cancel -r stonith_test_rsc -a monitor -i 2s ' + self.action_timeout, "stonith_cancel_event" : "-l \"NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:ok op_status:Cancelled\" ", } def new_test(self, name, description): """ Create a named test """ test = Test(name, description, self.verbose, self.tls, self.timeout, self.force_wait, self.logdir) self.tests.append(test) return test def setup_test_environment(self): """ Prepare the host before executing any tests """ if REMOTE_ENABLED: os.system("service pacemaker_remote stop") self.cleanup_test_environment() if self.tls and not os.path.isfile("/etc/pacemaker/authkey"): print("Installing /etc/pacemaker/authkey ...") os.system("mkdir -p /etc/pacemaker") os.system("dd if=/dev/urandom of=/etc/pacemaker/authkey bs=4096 count=1") self.installed_files.append("/etc/pacemaker/authkey") # If we're in build directory, install agents if not already installed if os.path.exists("%s/cts/cts-exec.in" % BUILD_DIR): if not os.path.exists("@OCF_RA_INSTALL_DIR@/pacemaker"): # @TODO remember which components were created and remove them os.makedirs("@OCF_RA_INSTALL_DIR@/pacemaker", 0o755) for agent in ["Dummy", "Stateful", "ping"]: agent_source = "%s/extra/resources/%s" % (BUILD_DIR, agent) agent_dest = "@OCF_RA_INSTALL_DIR@/pacemaker/%s" % (agent) if not os.path.exists(agent_dest): print("Installing %s ..." % (agent_dest)) shutil.copyfile(agent_source, agent_dest) os.chmod(agent_dest, EXECMODE) self.installed_files.append(agent_dest) subprocess.call(["cts-support", "install"]) def cleanup_test_environment(self): """ Clean up the host after executing desired tests """ for installed_file in self.installed_files: print("Removing %s ..." % (installed_file)) os.remove(installed_file) subprocess.call(["cts-support", "uninstall"]) def build_generic_tests(self): """ Register tests that apply to all resource classes """ common_cmds = self.common_cmds ### register/unregister tests ### for rsc in self.rsc_classes: test = self.new_test("generic_registration_%s" % (rsc), "Simple resource registration test for %s standard" % (rsc)) test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)]) test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)]) ### start/stop tests ### for rsc in self.rsc_classes: test = self.new_test("generic_start_stop_%s" % (rsc), "Simple start and stop test for %s standard" % (rsc)) test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)]) test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)]) test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)]) test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)]) ### monitor cancel test ### for rsc in self.rsc_classes: test = self.new_test("generic_monitor_cancel_%s" % (rsc), "Simple monitor cancel test for %s standard" % (rsc)) test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)]) test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)]) test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled #### test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled #### test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) test.add_cmd(common_cmds["%s_cancel_line" % (rsc)] + " " + common_cmds["%s_cancel_event" % (rsc)]) ### If this happens the monitor did not actually cancel correctly. ### test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)], CrmExit.TIMEOUT) ### If this happens the monitor did not actually cancel correctly. ### test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)], CrmExit.TIMEOUT) test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)]) test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)]) ### monitor duplicate test ### for rsc in self.rsc_classes: test = self.new_test("generic_monitor_duplicate_%s" % (rsc), "Test creation and canceling of duplicate monitors for %s standard" % (rsc)) test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)]) test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)]) test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled #### test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled #### test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) # Add the duplicate monitors test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)]) test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)]) test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)]) test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)]) # verify we still get update events ### If this fails, that means the monitor may not be getting rescheduled #### test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) # cancel the monitor, if the duplicate merged with the original, we should no longer see monitor updates test.add_cmd(common_cmds["%s_cancel_line" % (rsc)] + " " + common_cmds["%s_cancel_event" % (rsc)]) ### If this happens the monitor did not actually cancel correctly. ### test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)], CrmExit.TIMEOUT) ### If this happens the monitor did not actually cancel correctly. ### test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)], CrmExit.TIMEOUT) test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)]) test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)]) ### stop implies cancel test ### for rsc in self.rsc_classes: test = self.new_test("generic_stop_implies_cancel_%s" % (rsc), "Verify stopping a resource implies cancel of recurring ops for %s standard" % (rsc)) test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)]) test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)]) test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled #### test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) ### If this fails, that means the monitor may not be getting rescheduled #### test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)]) ### If this happens the monitor did not actually cancel correctly. ### test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)], CrmExit.TIMEOUT) ### If this happens the monitor did not actually cancel correctly. ### test.add_expected_fail_cmd(common_cmds["%s_monitor_event" % (rsc)], CrmExit.TIMEOUT) test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)]) def build_multi_rsc_tests(self): """ Register complex tests that involve managing multiple resouces of different types """ common_cmds = self.common_cmds # do not use service and systemd at the same time, it is the same resource. ### register start monitor stop unregister resources of each type at the same time. ### test = self.new_test("multi_rsc_start_stop_all", "Start, monitor, and stop resources of multiple types and classes") for rsc in self.rsc_classes: test.add_cmd(common_cmds["%s_reg_line" % (rsc)] + " " + common_cmds["%s_reg_event" % (rsc)]) for rsc in self.rsc_classes: test.add_cmd(common_cmds["%s_start_line" % (rsc)] + " " + common_cmds["%s_start_event" % (rsc)]) for rsc in self.rsc_classes: test.add_cmd(common_cmds["%s_monitor_line" % (rsc)] + " " + common_cmds["%s_monitor_event" % (rsc)]) for rsc in self.rsc_classes: ### If this fails, that means the monitor is not being rescheduled #### test.add_cmd(common_cmds["%s_monitor_event" % (rsc)]) for rsc in self.rsc_classes: test.add_cmd(common_cmds["%s_cancel_line" % (rsc)] + " " + common_cmds["%s_cancel_event" % (rsc)]) for rsc in self.rsc_classes: test.add_cmd(common_cmds["%s_stop_line" % (rsc)] + " " + common_cmds["%s_stop_event" % (rsc)]) for rsc in self.rsc_classes: test.add_cmd(common_cmds["%s_unreg_line" % (rsc)] + " " + common_cmds["%s_unreg_event" % (rsc)]) def build_negative_tests(self): """ Register tests related to how pacemaker-execd handles failures """ ### ocf start timeout test ### test = self.new_test("ocf_start_timeout", "Force start timeout to occur, verify start failure.") test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" " + self.action_timeout + "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") # -t must be less than self.action_timeout test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" -k \"op_sleep\" -v \"5\" -t 1000 -w") test.add_cmd('-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:error op_status:Timed Out" ' + self.action_timeout) test.add_cmd("-c exec -r test_rsc -a stop " + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\" ") test.add_cmd("-c unregister_rsc -r test_rsc " + self.action_timeout + "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### stonith start timeout test ### test = self.new_test("stonith_start_timeout", "Force start timeout to occur, verify start failure.") test.add_cmd('-c register_rsc -r test_rsc ' + '-C stonith -P pacemaker -T fence_dummy ' + self.action_timeout + '-l "NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete"') test.add_cmd('-c exec -r test_rsc -a start -k monitor_delay -v 30 ' + '-t 1000 -w') # -t must be less than self.action_timeout - test.add_cmd('-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:OCF_TIMEOUT op_status:Timed Out" ' + test.add_cmd('-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:error op_status:Timed Out" ' + self.action_timeout) test.add_cmd("-c exec -r test_rsc -a stop " + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\" ") test.add_cmd("-c unregister_rsc -r test_rsc " + self.action_timeout + "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### stonith component fail ### common_cmds = self.common_cmds test = self.new_test("stonith_component_fail", "Kill stonith component after pacemaker-execd connects") test.add_cmd(common_cmds["stonith_reg_line"] + " " + common_cmds["stonith_reg_event"]) test.add_cmd(common_cmds["stonith_start_line"] + " " + common_cmds["stonith_start_event"]) test.add_cmd('-c exec -r stonith_test_rsc -a monitor -i 600s ' '-l "NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:ok op_status:complete" ' + self.action_timeout) test.add_cmd_and_kill("killall -9 -q pacemaker-fenced lt-pacemaker-fenced", '-l "NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:error op_status:error" -t 15000') test.add_cmd(common_cmds["stonith_unreg_line"] + " " + common_cmds["stonith_unreg_event"]) ### monitor fail for ocf resources ### test = self.new_test("monitor_fail_ocf", "Force ocf monitor to fail, verify failure is reported.") test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" " + self.action_timeout + "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" " + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" " + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd('-c exec -r test_rsc -a monitor -i 1s ' + self.action_timeout + '-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete"') test.add_cmd('-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete"' + self.action_timeout) test.add_cmd('-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete"' + self.action_timeout) test.add_cmd_and_kill("rm -f @localstatedir@/run/Dummy-test_rsc.state", '-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete" ' + self.action_timeout) test.add_cmd('-c cancel -r test_rsc -a monitor -i 1s ' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ") test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" " + self.action_timeout, CrmExit.TIMEOUT) test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" " + self.action_timeout, CrmExit.TIMEOUT) test.add_cmd("-c unregister_rsc -r \"test_rsc\" " + self.action_timeout + "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### verify notify changes only for monitor operation. ### test = self.new_test("monitor_changes_only", "Verify when flag is set, only monitor changes are notified.") test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+" -o " "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd('-c exec -r test_rsc -a monitor -i 1s ' + self.action_timeout + ' -o -l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ') test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout, CrmExit.TIMEOUT) test.add_cmd_and_kill('rm -f @localstatedir@/run/Dummy-test_rsc.state', '-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete"' + self.action_timeout) test.add_cmd('-c cancel -r test_rsc -a monitor -i 1s' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ") test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" "+self.action_timeout, CrmExit.TIMEOUT) test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout, CrmExit.TIMEOUT) test.add_cmd('-c unregister_rsc -r "test_rsc" ' + self.action_timeout + '-l "NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete"') ### monitor fail for systemd resource ### if "systemd" in self.rsc_classes: test = self.new_test("monitor_fail_systemd", "Force systemd monitor to fail, verify failure is reported..") test.add_cmd("-c register_rsc -r \"test_rsc\" -C systemd -T pacemaker-cts-dummyd@3 " + self.action_timeout + "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd('-c exec -r test_rsc -a monitor -i 1s ' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ") test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout) test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout) test.add_cmd_and_kill("pkill -9 -f pacemaker-cts-dummyd", '-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete"' + self.action_timeout) test.add_cmd('-c cancel -r test_rsc -a monitor -i 1s' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ") test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" "+self.action_timeout, CrmExit.TIMEOUT) test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout, CrmExit.TIMEOUT) test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### monitor fail for upstart resource ### if "upstart" in self.rsc_classes: test = self.new_test("monitor_fail_upstart", "Force upstart monitor to fail, verify failure is reported..") test.add_cmd("-c register_rsc -r \"test_rsc\" -C upstart -T pacemaker-cts-dummyd "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd('-c exec -r test_rsc -a monitor -i 1s ' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ") test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout) test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout) test.add_cmd_and_kill('killall -9 -q dd', '-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete"' + self.action_timeout) test.add_cmd('-c cancel -r test_rsc -a monitor -i 1s' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ") test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete\" "+self.action_timeout, CrmExit.TIMEOUT) test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout, CrmExit.TIMEOUT) test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### Cancel non-existent operation on a resource ### test = self.new_test("cancel_non_existent_op", "Attempt to cancel the wrong monitor operation, verify expected failure") test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd('-c exec -r test_rsc -a monitor -i 1s ' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ") test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout) ### interval is wrong, should fail test.add_expected_fail_cmd('-c cancel -r test_rsc -a monitor -i 2s' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ") ### action name is wrong, should fail test.add_expected_fail_cmd('-c cancel -r test_rsc -a stop -i 1s' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:Cancelled\" ") test.add_cmd("-c unregister_rsc -r \"test_rsc\" " + self.action_timeout + "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### Attempt to invoke non-existent rsc id ### test = self.new_test("invoke_non_existent_rsc", "Attempt to perform operations on a non-existent rsc id.") test.add_expected_fail_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:error op_status:complete\" ") test.add_expected_fail_cmd("-c exec -r test_rsc -a stop "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\" ") test.add_expected_fail_cmd('-c exec -r test_rsc -a monitor -i 6s ' + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" ") test.add_expected_fail_cmd("-c cancel -r test_rsc -a start "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:Cancelled\" ") test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### Register and start a resource that doesn't exist, systemd ### if "systemd" in self.rsc_classes: test = self.new_test("start_uninstalled_systemd", "Register uninstalled systemd agent, try to start, verify expected failure") test.add_cmd("-c register_rsc -r \"test_rsc\" -C systemd -T this_is_fake1234 "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:not installed op_status:Not installed\" ") test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") if "upstart" in self.rsc_classes: test = self.new_test("start_uninstalled_upstart", "Register uninstalled upstart agent, try to start, verify expected failure") test.add_cmd("-c register_rsc -r \"test_rsc\" -C upstart -T this_is_fake1234 "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:not installed op_status:Not installed\" ") test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### Register and start a resource that doesn't exist, ocf ### test = self.new_test("start_uninstalled_ocf", "Register uninstalled ocf agent, try to start, verify expected failure.") test.add_cmd("-c register_rsc -r \"test_rsc\" -C ocf -P pacemaker -T this_is_fake1234 "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:not installed op_status:Not installed\" ") test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### Register ocf with non-existent provider ### test = self.new_test("start_ocf_bad_provider", "Register ocf agent with a non-existent provider, verify expected failure.") test.add_cmd("-c register_rsc -r \"test_rsc\" -C ocf -P pancakes -T Dummy "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:not installed op_status:Not installed\" ") test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### Register ocf with empty provider field ### test = self.new_test("start_ocf_no_provider", "Register ocf agent with a no provider, verify expected failure.") test.add_expected_fail_cmd("-c register_rsc -r \"test_rsc\" -C ocf -T Dummy "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_expected_fail_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:Error\" ") test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") def build_stress_tests(self): """ Register stress tests """ timeout = "-t 20000" iterations = 25 test = self.new_test("ocf_stress", "Verify OCF agent handling works under load") for i in range(iterations): test.add_cmd("-c register_rsc -r rsc_%s %s -C ocf -P heartbeat -T Dummy -l \"NEW_EVENT event_type:register rsc_id:rsc_%s action:none rc:ok op_status:complete\"" % (i, timeout, i)) test.add_cmd("-c exec -r rsc_%s -a start %s -l \"NEW_EVENT event_type:exec_complete rsc_id:rsc_%s action:start rc:ok op_status:complete\"" % (i, timeout, i)) test.add_cmd('-c exec -r rsc_%s -a monitor %s -i 1s ' '-l "NEW_EVENT event_type:exec_complete rsc_id:rsc_%s action:monitor rc:ok op_status:complete"' % (i, timeout, i)) for i in range(iterations): test.add_cmd("-c exec -r rsc_%s -a stop %s -l \"NEW_EVENT event_type:exec_complete rsc_id:rsc_%s action:stop rc:ok op_status:complete\"" % (i, timeout, i)) test.add_cmd("-c unregister_rsc -r rsc_%s %s -l \"NEW_EVENT event_type:unregister rsc_id:rsc_%s action:none rc:ok op_status:complete\"" % (i, timeout, i)) if "systemd" in self.rsc_classes: test = self.new_test("systemd_stress", "Verify systemd dbus connection works under load") for i in range(iterations): test.add_cmd("-c register_rsc -r rsc_%s %s -C systemd -T pacemaker-cts-dummyd@3 -l \"NEW_EVENT event_type:register rsc_id:rsc_%s action:none rc:ok op_status:complete\"" % (i, timeout, i)) test.add_cmd("-c exec -r rsc_%s -a start %s -l \"NEW_EVENT event_type:exec_complete rsc_id:rsc_%s action:start rc:ok op_status:complete\"" % (i, timeout, i)) test.add_cmd('-c exec -r rsc_%s -a monitor %s -i 1s ' '-l "NEW_EVENT event_type:exec_complete rsc_id:rsc_%s action:monitor rc:ok op_status:complete"' % (i, timeout, i)) for i in range(iterations): test.add_cmd("-c exec -r rsc_%s -a stop %s -l \"NEW_EVENT event_type:exec_complete rsc_id:rsc_%s action:stop rc:ok op_status:complete\"" % (i, timeout, i)) test.add_cmd("-c unregister_rsc -r rsc_%s %s -l \"NEW_EVENT event_type:unregister rsc_id:rsc_%s action:none rc:ok op_status:complete\"" % (i, timeout, i)) iterations = 9 timeout = "-t 30000" ### Verify recurring op in-flight collision is handled in series properly test = self.new_test("rsc_inflight_collision", "Verify recurring ops do not collide with other operations for the same rsc.") test.add_cmd("-c register_rsc -r test_rsc -P pacemaker -C ocf -T Dummy " "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" "+self.action_timeout) test.add_cmd("-c exec -r test_rsc -a start %s -k op_sleep -v 1 -l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\"" % (timeout)) for i in range(iterations): test.add_cmd('-c exec -r test_rsc -a monitor %s -i 100%dms ' '-k op_sleep -v 2 ' '-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete"' % (timeout, i)) test.add_cmd("-c exec -r test_rsc -a stop %s -l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\"" % (timeout)) test.add_cmd("-c unregister_rsc -r test_rsc %s -l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\"" % (timeout)) def build_custom_tests(self): """ Register tests that target specific cases """ ### verify resource temporary folder is created and used by OCF agents. ### test = self.new_test("rsc_tmp_dir", "Verify creation and use of rsc temporary state directory") test.add_sys_cmd("ls", "-al @CRM_RSCTMP_DIR@") test.add_cmd("-c register_rsc -r test_rsc -P heartbeat -C ocf -T Dummy " "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" "+self.action_timeout) test.add_cmd("-c exec -r test_rsc -a start -t 4000") test.add_sys_cmd("ls", "-al @CRM_RSCTMP_DIR@") test.add_sys_cmd("ls", "@CRM_RSCTMP_DIR@/Dummy-test_rsc.state") test.add_cmd("-c exec -r test_rsc -a stop -t 4000") test.add_cmd("-c unregister_rsc -r test_rsc "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### start delay then stop test ### test = self.new_test("start_delay", "Verify start delay works as expected.") test.add_cmd("-c register_rsc -r test_rsc -P pacemaker -C ocf -T Dummy " "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" "+self.action_timeout) test.add_cmd("-c exec -r test_rsc -s 6000 -a start -w -t 6000") test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" -t 2000", CrmExit.TIMEOUT) test.add_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" -t 6000") test.add_cmd("-c exec -r test_rsc -a stop " + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:stop rc:ok op_status:complete\" ") test.add_cmd("-c unregister_rsc -r test_rsc " + self.action_timeout + "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### start delay, but cancel before it gets a chance to start. ### test = self.new_test("start_delay_cancel", "Using start_delay, start a rsc, but cancel the start op before execution.") test.add_cmd("-c register_rsc -r test_rsc -P pacemaker -C ocf -T Dummy " "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" "+self.action_timeout) test.add_cmd("-c exec -r test_rsc -s 5000 -a start -w -t 4000") test.add_cmd("-c cancel -r test_rsc -a start " + self.action_timeout + "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:Cancelled\" ") test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" -t 5000", CrmExit.TIMEOUT) test.add_cmd("-c unregister_rsc -r test_rsc " + self.action_timeout + "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### Register a bunch of resources, verify we can get info on them ### test = self.new_test("verify_get_rsc_info", "Register multiple resources, verify retrieval of rsc info.") if "systemd" in self.rsc_classes: test.add_cmd("-c register_rsc -r rsc1 -C systemd -T pacemaker-cts-dummyd@3 "+self.action_timeout) test.add_cmd("-c get_rsc_info -r rsc1 ") test.add_cmd("-c unregister_rsc -r rsc1 "+self.action_timeout) test.add_expected_fail_cmd("-c get_rsc_info -r rsc1 ") if "upstart" in self.rsc_classes: test.add_cmd("-c register_rsc -r rsc1 -C upstart -T pacemaker-cts-dummyd "+self.action_timeout) test.add_cmd("-c get_rsc_info -r rsc1 ") test.add_cmd("-c unregister_rsc -r rsc1 "+self.action_timeout) test.add_expected_fail_cmd("-c get_rsc_info -r rsc1 ") test.add_cmd("-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker "+self.action_timeout) test.add_cmd("-c get_rsc_info -r rsc2 ") test.add_cmd("-c unregister_rsc -r rsc2 "+self.action_timeout) test.add_expected_fail_cmd("-c get_rsc_info -r rsc2 ") ### Register duplicate, verify only one entry exists and can still be removed. test = self.new_test("duplicate_registration", "Register resource multiple times, verify only one entry exists and can be removed.") test.add_cmd("-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker "+self.action_timeout) test.add_cmd_check_stdout("-c get_rsc_info -r rsc2 ", "id:rsc2 class:ocf provider:pacemaker type:Dummy") test.add_cmd("-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker "+self.action_timeout) test.add_cmd_check_stdout("-c get_rsc_info -r rsc2 ", "id:rsc2 class:ocf provider:pacemaker type:Dummy") test.add_cmd("-c register_rsc -r rsc2 -C ocf -T Stateful -P pacemaker "+self.action_timeout) test.add_cmd_check_stdout("-c get_rsc_info -r rsc2 ", "id:rsc2 class:ocf provider:pacemaker type:Stateful") test.add_cmd("-c unregister_rsc -r rsc2 "+self.action_timeout) test.add_expected_fail_cmd("-c get_rsc_info -r rsc2 ") ### verify the option to only send notification to the original client. ### test = self.new_test("notify_orig_client_only", "Verify option to only send notifications to the client originating the action.") test.add_cmd("-c register_rsc -r \"test_rsc\" -C \"ocf\" -P \"pacemaker\" -T \"Dummy\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:register rsc_id:test_rsc action:none rc:ok op_status:complete\" ") test.add_cmd("-c exec -r \"test_rsc\" -a \"start\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete\" ") test.add_cmd('-c exec -r \"test_rsc\" -a \"monitor\" -i 1s ' + self.action_timeout + ' -n ' '-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete"') # this will fail because the monitor notifications should only go to the original caller, which no longer exists. test.add_expected_fail_cmd("-l \"NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete\" "+self.action_timeout, CrmExit.TIMEOUT) test.add_cmd('-c cancel -r test_rsc -a monitor -i 1s -t 6000 ') test.add_cmd("-c unregister_rsc -r \"test_rsc\" "+self.action_timeout+ "-l \"NEW_EVENT event_type:unregister rsc_id:test_rsc action:none rc:ok op_status:complete\" ") ### get metadata ### test = self.new_test("get_ocf_metadata", "Retrieve metadata for a resource") test.add_cmd_check_stdout("-c metadata -C \"ocf\" -P \"pacemaker\" -T \"Dummy\"", "resource-agent name=\"Dummy\"") test.add_cmd("-c metadata -C \"ocf\" -P \"pacemaker\" -T \"Stateful\"") test.add_expected_fail_cmd("-c metadata -P \"pacemaker\" -T \"Stateful\"") test.add_expected_fail_cmd("-c metadata -C \"ocf\" -P \"pacemaker\" -T \"fake_agent\"") ### get metadata ### test = self.new_test("get_lsb_metadata", "Retrieve metadata for a resource") test.add_cmd_check_stdout("-c metadata -C \"lsb\" -T \"LSBDummy\"", "resource-agent name='LSBDummy'") ### get stonith metadata ### test = self.new_test("get_stonith_metadata", "Retrieve stonith metadata for a resource") test.add_cmd_check_stdout("-c metadata -C \"stonith\" -P \"pacemaker\" -T \"fence_dummy\"", "resource-agent name=\"fence_dummy\"") ### get metadata ### if "systemd" in self.rsc_classes: test = self.new_test("get_systemd_metadata", "Retrieve metadata for a resource") test.add_cmd_check_stdout("-c metadata -C \"systemd\" -T \"pacemaker-cts-dummyd@\"", "resource-agent name=\"pacemaker-cts-dummyd@\"") ### get metadata ### if "upstart" in self.rsc_classes: test = self.new_test("get_upstart_metadata", "Retrieve metadata for a resource") test.add_cmd_check_stdout("-c metadata -C \"upstart\" -T \"pacemaker-cts-dummyd\"", "resource-agent name=\"pacemaker-cts-dummyd\"") ### get ocf providers ### test = self.new_test("list_ocf_providers", "Retrieve list of available resource providers, verifies pacemaker is a provider.") test.add_cmd_check_stdout("-c list_ocf_providers ", "pacemaker") test.add_cmd_check_stdout("-c list_ocf_providers -T ping", "pacemaker") ### Verify agents only exist in their lists ### test = self.new_test("verify_agent_lists", "Verify the agent lists contain the right data.") test.add_cmd_check_stdout("-c list_agents ", "Stateful") ### ocf ### test.add_cmd_check_stdout("-c list_agents -C ocf", "Stateful") test.add_cmd_check_stdout("-c list_agents -C lsb", "", "Stateful") ### should not exist test.add_cmd_check_stdout("-c list_agents -C service", "", "Stateful") ### should not exist test.add_cmd_check_stdout("-c list_agents ", "LSBDummy") ### init.d ### test.add_cmd_check_stdout("-c list_agents -C lsb", "LSBDummy") test.add_cmd_check_stdout("-c list_agents -C service", "LSBDummy") test.add_cmd_check_stdout("-c list_agents -C ocf", "", "pacemaker-cts-dummyd@") ### should not exist test.add_cmd_check_stdout("-c list_agents -C ocf", "", "pacemaker-cts-dummyd@") ### should not exist test.add_cmd_check_stdout("-c list_agents -C lsb", "", "fence_dummy") ### should not exist test.add_cmd_check_stdout("-c list_agents -C service", "", "fence_dummy") ### should not exist test.add_cmd_check_stdout("-c list_agents -C ocf", "", "fence_dummy") ### should not exist if "systemd" in self.rsc_classes: test.add_cmd_check_stdout("-c list_agents ", "pacemaker-cts-dummyd@") ### systemd ### test.add_cmd_check_stdout("-c list_agents -C service", "LSBDummy") test.add_cmd_check_stdout("-c list_agents -C systemd", "", "Stateful") ### should not exist test.add_cmd_check_stdout("-c list_agents -C systemd", "pacemaker-cts-dummyd@") test.add_cmd_check_stdout("-c list_agents -C systemd", "", "fence_dummy") ### should not exist if "upstart" in self.rsc_classes: test.add_cmd_check_stdout("-c list_agents ", "pacemaker-cts-dummyd") ### upstart ### test.add_cmd_check_stdout("-c list_agents -C service", "LSBDummy") test.add_cmd_check_stdout("-c list_agents -C upstart", "", "Stateful") ### should not exist test.add_cmd_check_stdout("-c list_agents -C upstart", "pacemaker-cts-dummyd") test.add_cmd_check_stdout("-c list_agents -C upstart", "", "fence_dummy") ### should not exist if "stonith" in self.rsc_classes: test.add_cmd_check_stdout("-c list_agents -C stonith", "fence_dummy") ### stonith ### test.add_cmd_check_stdout("-c list_agents -C stonith", "", "pacemaker-cts-dummyd@") ### should not exist test.add_cmd_check_stdout("-c list_agents -C stonith", "", "Stateful") ### should not exist test.add_cmd_check_stdout("-c list_agents ", "fence_dummy") def print_list(self): """ List all registered tests """ print("\n==== %d TESTS FOUND ====" % (len(self.tests))) print("%35s - %s" % ("TEST NAME", "TEST DESCRIPTION")) print("%35s - %s" % ("--------------------", "--------------------")) for test in self.tests: print("%35s - %s" % (test.name, test.description)) print("==== END OF LIST ====\n") def run_single(self, name): """ Run a single named test """ for test in self.tests: if test.name == name: test.run() break def run_tests_matching(self, pattern): """ Run all tests whose name matches a pattern """ for test in self.tests: if test.name.count(pattern) != 0: test.run() def run_tests(self): """ Run all tests """ for test in self.tests: test.run() def exit(self): """ Exit (with error status code if any test failed) """ for test in self.tests: if test.executed == 0: continue if test.get_exitcode() != CrmExit.OK: sys.exit(CrmExit.ERROR) sys.exit(CrmExit.OK) def print_results(self): """ Print summary of results of executed tests """ failures = 0 success = 0 print("\n\n======= FINAL RESULTS ==========") print("\n--- FAILURE RESULTS:") for test in self.tests: if test.executed == 0: continue if test.get_exitcode() != CrmExit.OK: failures = failures + 1 test.print_result(" ") else: success = success + 1 if failures == 0: print(" None") print("\n--- TOTALS\n Pass:%d\n Fail:%d\n" % (success, failures)) class TestOptions(object): """ Option handler """ def __init__(self): self.options = {} self.options['list-tests'] = 0 self.options['run-all'] = 1 self.options['run-only'] = "" self.options['run-only-pattern'] = "" self.options['verbose'] = 0 self.options['timeout'] = 2 self.options['force-wait'] = 0 self.options['invalid-arg'] = "" self.options['show-usage'] = 0 self.options['pacemaker-remote'] = 0 def build_options(self, argv): """ Set options based on command-line arguments """ args = argv[1:] skip = 0 for i in range(0, len(args)): if skip: skip = 0 continue elif args[i] == "-h" or args[i] == "--help": self.options['show-usage'] = 1 elif args[i] == "-l" or args[i] == "--list-tests": self.options['list-tests'] = 1 elif args[i] == "-V" or args[i] == "--verbose": self.options['verbose'] = 1 elif args[i] == "-t" or args[i] == "--timeout": self.options['timeout'] = float(args[i+1]) elif args[i] == "-w" or args[i] == "--force-wait": self.options['force-wait'] = 1 elif args[i] == "-R" or args[i] == "--pacemaker-remote": if REMOTE_ENABLED: self.options['pacemaker-remote'] = 1 else: print("ERROR: This build does not support Pacemaker Remote") sys.exit(CrmExit.USAGE) elif args[i] == "-r" or args[i] == "--run-only": self.options['run-only'] = args[i+1] skip = 1 elif args[i] == "-p" or args[i] == "--run-only-pattern": self.options['run-only-pattern'] = args[i+1] skip = 1 def show_usage(): """ Show command usage """ print("usage: " + sys.argv[0] + " [options]") print("If no options are provided, all tests will run") print("Options:") print("\t [--help | -h] Show usage") print("\t [--list-tests | -l] Print out all registered tests.") print("\t [--run-only | -r 'testname'] Run a specific test") print("\t [--verbose | -V] Verbose output") print("\t [--timeout | -t 'floating point number']" "\n\t\tUp to how many seconds each test case waits for the daemon to be initialized." "\n\t\tDefaults to 2. The value 0 means no limit.") print("\t [--force-wait | -w]" "\n\t\tEach test case waits the default/specified --timeout for the daemon without tracking the log.") if REMOTE_ENABLED: print("\t [--pacemaker-remote | -R Test pacemaker-remoted binary instead of pacemaker-execd") print("\t [--run-only-pattern | -p 'string'] Run only tests containing the string value") print("\n\tExample: Run only the test 'start_stop'") print("\t\t " + sys.argv[0] + " --run-only start_stop") print("\n\tExample: Run only the tests with the string 'systemd' present in them") print("\t\t " + sys.argv[0] + " --run-only-pattern systemd") def main(argv): """ Run pacemaker-execd regression tests as specified by arguments """ update_path() opts = TestOptions() opts.build_options(argv) if opts.options['show-usage']: show_usage() sys.exit(CrmExit.OK) # Create a temporary directory for log files (the directory will # automatically be erased when done) with tempfile.TemporaryDirectory(prefix="cts-exec-") as logdir: tests = Tests(opts.options['verbose'], opts.options['pacemaker-remote'], opts.options['timeout'], opts.options['force-wait'], logdir) tests.build_generic_tests() tests.build_multi_rsc_tests() tests.build_negative_tests() tests.build_custom_tests() tests.build_stress_tests() if opts.options['list-tests']: tests.print_list() sys.exit(CrmExit.OK) tests.setup_test_environment() print("Starting ...") if opts.options['run-only-pattern'] != "": tests.run_tests_matching(opts.options['run-only-pattern']) tests.print_results() elif opts.options['run-only'] != "": tests.run_single(opts.options['run-only']) tests.print_results() else: tests.run_tests() tests.print_results() tests.cleanup_test_environment() tests.exit() if __name__ == "__main__": main(sys.argv) diff --git a/daemons/controld/controld_te_actions.c b/daemons/controld/controld_te_actions.c index 80a5236c95..b9c34cd61a 100644 --- a/daemons/controld/controld_te_actions.c +++ b/daemons/controld/controld_te_actions.c @@ -1,641 +1,641 @@ /* * Copyright 2004-2021 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU General Public License version 2 * or later (GPLv2+) WITHOUT ANY WARRANTY. */ #include #include #include #include #include // lrmd_event_data_t, lrmd_free_event() #include #include #include #include #include char *te_uuid = NULL; GHashTable *te_targets = NULL; void send_rsc_command(crm_action_t * action); static void te_update_job_count(crm_action_t * action, int offset); static void te_start_action_timer(crm_graph_t * graph, crm_action_t * action) { action->timer = calloc(1, sizeof(crm_action_timer_t)); action->timer->timeout = action->timeout; action->timer->action = action; action->timer->source_id = g_timeout_add(action->timer->timeout + graph->network_delay, action_timer_callback, (void *)action->timer); CRM_ASSERT(action->timer->source_id != 0); } static gboolean te_pseudo_action(crm_graph_t * graph, crm_action_t * pseudo) { const char *task = crm_element_value(pseudo->xml, XML_LRM_ATTR_TASK); /* send to peers as well? */ if (pcmk__str_eq(task, CRM_OP_MAINTENANCE_NODES, pcmk__str_casei)) { GHashTableIter iter; crm_node_t *node = NULL; g_hash_table_iter_init(&iter, crm_peer_cache); while (g_hash_table_iter_next(&iter, NULL, (gpointer *) &node)) { xmlNode *cmd = NULL; if (pcmk__str_eq(fsa_our_uname, node->uname, pcmk__str_casei)) { continue; } cmd = create_request(task, pseudo->xml, node->uname, CRM_SYSTEM_CRMD, CRM_SYSTEM_TENGINE, NULL); send_cluster_message(node, crm_msg_crmd, cmd, FALSE); free_xml(cmd); } remote_ra_process_maintenance_nodes(pseudo->xml); } else { /* Check action for Pacemaker Remote node side effects */ remote_ra_process_pseudo(pseudo->xml); } crm_debug("Pseudo-action %d (%s) fired and confirmed", pseudo->id, crm_element_value(pseudo->xml, XML_LRM_ATTR_TASK_KEY)); te_action_confirmed(pseudo, graph); return TRUE; } static int get_target_rc(crm_action_t * action) { int exit_status; pcmk__scan_min_int(crm_meta_value(action->params, XML_ATTR_TE_TARGET_RC), &exit_status, 0); return exit_status; } static gboolean te_crm_command(crm_graph_t * graph, crm_action_t * action) { char *counter = NULL; xmlNode *cmd = NULL; gboolean is_local = FALSE; const char *id = NULL; const char *task = NULL; const char *value = NULL; const char *on_node = NULL; const char *router_node = NULL; gboolean rc = TRUE; gboolean no_wait = FALSE; id = ID(action->xml); task = crm_element_value(action->xml, XML_LRM_ATTR_TASK); on_node = crm_element_value(action->xml, XML_LRM_ATTR_TARGET); router_node = crm_element_value(action->xml, XML_LRM_ATTR_ROUTER_NODE); if (!router_node) { router_node = on_node; if (pcmk__str_eq(task, CRM_OP_LRM_DELETE, pcmk__str_casei)) { const char *mode = crm_element_value(action->xml, PCMK__XA_MODE); if (pcmk__str_eq(mode, XML_TAG_CIB, pcmk__str_casei)) { router_node = fsa_our_uname; } } } CRM_CHECK(on_node != NULL && strlen(on_node) != 0, crm_err("Corrupted command (id=%s) %s: no node", crm_str(id), crm_str(task)); return FALSE); if (pcmk__str_eq(router_node, fsa_our_uname, pcmk__str_casei)) { is_local = TRUE; } value = crm_meta_value(action->params, XML_ATTR_TE_NOWAIT); if (crm_is_true(value)) { no_wait = TRUE; } crm_info("Executing crm-event (%s)%s%s: %s on %s", crm_str(id), (is_local? " locally" : ""), (no_wait? " without waiting" : ""), crm_str(task), on_node); if (is_local && pcmk__str_eq(task, CRM_OP_SHUTDOWN, pcmk__str_casei)) { /* defer until everything else completes */ crm_info("crm-event (%s) is a local shutdown", crm_str(id)); graph->completion_action = tg_shutdown; graph->abort_reason = "local shutdown"; te_action_confirmed(action, graph); return TRUE; } else if (pcmk__str_eq(task, CRM_OP_SHUTDOWN, pcmk__str_casei)) { crm_node_t *peer = crm_get_peer(0, router_node); pcmk__update_peer_expected(__func__, peer, CRMD_JOINSTATE_DOWN); } cmd = create_request(task, action->xml, router_node, CRM_SYSTEM_CRMD, CRM_SYSTEM_TENGINE, NULL); counter = pcmk__transition_key(transition_graph->id, action->id, get_target_rc(action), te_uuid); crm_xml_add(cmd, XML_ATTR_TRANSITION_KEY, counter); rc = send_cluster_message(crm_get_peer(0, router_node), crm_msg_crmd, cmd, TRUE); free(counter); free_xml(cmd); if (rc == FALSE) { crm_err("Action %d failed: send", action->id); return FALSE; } else if (no_wait) { te_action_confirmed(action, graph); } else { if (action->timeout <= 0) { crm_err("Action %d: %s on %s had an invalid timeout (%dms). Using %ums instead", action->id, task, on_node, action->timeout, graph->network_delay); action->timeout = (int) graph->network_delay; } te_start_action_timer(graph, action); } return TRUE; } void controld_record_action_timeout(crm_action_t *action) { lrmd_event_data_t *op = NULL; xmlNode *state = NULL; xmlNode *rsc = NULL; xmlNode *xml_op = NULL; xmlNode *action_rsc = NULL; int rc = pcmk_ok; const char *rsc_id = NULL; const char *target = crm_element_value(action->xml, XML_LRM_ATTR_TARGET); const char *task_uuid = crm_element_value(action->xml, XML_LRM_ATTR_TASK_KEY); const char *target_uuid = crm_element_value(action->xml, XML_LRM_ATTR_TARGET_UUID); int call_options = cib_quorum_override | cib_scope_local; int target_rc = get_target_rc(action); crm_warn("%s %d: %s on %s timed out", crm_element_name(action->xml), action->id, task_uuid, target); action_rsc = find_xml_node(action->xml, XML_CIB_TAG_RESOURCE, TRUE); if (action_rsc == NULL) { return; } rsc_id = ID(action_rsc); CRM_CHECK(rsc_id != NULL, crm_log_xml_err(action->xml, "Bad:action"); return); /* update the CIB */ state = create_xml_node(NULL, XML_CIB_TAG_STATE); crm_xml_add(state, XML_ATTR_UUID, target_uuid); crm_xml_add(state, XML_ATTR_UNAME, target); rsc = create_xml_node(state, XML_CIB_TAG_LRM); crm_xml_add(rsc, XML_ATTR_ID, target_uuid); rsc = create_xml_node(rsc, XML_LRM_TAG_RESOURCES); rsc = create_xml_node(rsc, XML_LRM_TAG_RESOURCE); crm_xml_add(rsc, XML_ATTR_ID, rsc_id); crm_copy_xml_element(action_rsc, rsc, XML_ATTR_TYPE); crm_copy_xml_element(action_rsc, rsc, XML_AGENT_ATTR_CLASS); crm_copy_xml_element(action_rsc, rsc, XML_AGENT_ATTR_PROVIDER); /* If the executor gets a timeout while waiting for the action to complete, * that will be reported via the usual callback. This timeout means that we * didn't hear from the executor or the controller that relayed the action * to the executor. */ op = pcmk__event_from_graph_action(NULL, action, PCMK_EXEC_TIMEOUT, - PCMK_OCF_TIMEOUT, + PCMK_OCF_UNKNOWN_ERROR, "Cluster communication timeout " "(no response from executor)"); op->call_id = -1; op->user_data = pcmk__transition_key(transition_graph->id, action->id, target_rc, te_uuid); xml_op = pcmk__create_history_xml(rsc, op, CRM_FEATURE_SET, target_rc, target, __func__, LOG_INFO); lrmd_free_event(op); crm_log_xml_trace(xml_op, "Action timeout"); rc = fsa_cib_conn->cmds->update(fsa_cib_conn, XML_CIB_TAG_STATUS, state, call_options); fsa_register_cib_callback(rc, FALSE, NULL, cib_action_updated); free_xml(state); crm_trace("Sent CIB update (call ID %d) for timeout of action %d (%s on %s)", rc, action->id, task_uuid, target); action->sent_update = TRUE; } static gboolean te_rsc_command(crm_graph_t * graph, crm_action_t * action) { /* never overwrite stop actions in the CIB with * anything other than completed results * * Writing pending stops makes it look like the * resource is running again */ xmlNode *cmd = NULL; xmlNode *rsc_op = NULL; gboolean rc = TRUE; gboolean no_wait = FALSE; gboolean is_local = FALSE; char *counter = NULL; const char *task = NULL; const char *value = NULL; const char *on_node = NULL; const char *router_node = NULL; const char *task_uuid = NULL; CRM_ASSERT(action != NULL); CRM_ASSERT(action->xml != NULL); action->executed = FALSE; on_node = crm_element_value(action->xml, XML_LRM_ATTR_TARGET); CRM_CHECK(on_node != NULL && strlen(on_node) != 0, crm_err("Corrupted command(id=%s) %s: no node", ID(action->xml), crm_str(task)); return FALSE); rsc_op = action->xml; task = crm_element_value(rsc_op, XML_LRM_ATTR_TASK); task_uuid = crm_element_value(action->xml, XML_LRM_ATTR_TASK_KEY); router_node = crm_element_value(rsc_op, XML_LRM_ATTR_ROUTER_NODE); if (!router_node) { router_node = on_node; } counter = pcmk__transition_key(transition_graph->id, action->id, get_target_rc(action), te_uuid); crm_xml_add(rsc_op, XML_ATTR_TRANSITION_KEY, counter); if (pcmk__str_eq(router_node, fsa_our_uname, pcmk__str_casei)) { is_local = TRUE; } value = crm_meta_value(action->params, XML_ATTR_TE_NOWAIT); if (crm_is_true(value)) { no_wait = TRUE; } crm_notice("Initiating %s operation %s%s on %s%s "CRM_XS" action %d", task, task_uuid, (is_local? " locally" : ""), on_node, (no_wait? " without waiting" : ""), action->id); cmd = create_request(CRM_OP_INVOKE_LRM, rsc_op, router_node, CRM_SYSTEM_LRMD, CRM_SYSTEM_TENGINE, NULL); if (is_local) { /* shortcut local resource commands */ ha_msg_input_t data = { .msg = cmd, .xml = rsc_op, }; fsa_data_t msg = { .id = 0, .data = &data, .data_type = fsa_dt_ha_msg, .fsa_input = I_NULL, .fsa_cause = C_FSA_INTERNAL, .actions = A_LRM_INVOKE, .origin = __func__, }; do_lrm_invoke(A_LRM_INVOKE, C_FSA_INTERNAL, fsa_state, I_NULL, &msg); } else { rc = send_cluster_message(crm_get_peer(0, router_node), crm_msg_lrmd, cmd, TRUE); } free(counter); free_xml(cmd); action->executed = TRUE; if (rc == FALSE) { crm_err("Action %d failed: send", action->id); return FALSE; } else if (no_wait) { crm_info("Action %d confirmed - no wait", action->id); action->confirmed = TRUE; /* Just mark confirmed. * Don't bump the job count only to immediately decrement it */ pcmk__update_graph(transition_graph, action); trigger_graph(); } else if (action->confirmed == TRUE) { crm_debug("Action %d: %s %s on %s(timeout %dms) was already confirmed.", action->id, task, task_uuid, on_node, action->timeout); } else { if (action->timeout <= 0) { crm_err("Action %d: %s %s on %s had an invalid timeout (%dms). Using %ums instead", action->id, task, task_uuid, on_node, action->timeout, graph->network_delay); action->timeout = (int) graph->network_delay; } te_update_job_count(action, 1); te_start_action_timer(graph, action); } return TRUE; } struct te_peer_s { char *name; int jobs; int migrate_jobs; }; static void te_peer_free(gpointer p) { struct te_peer_s *peer = p; free(peer->name); free(peer); } void te_reset_job_counts(void) { GHashTableIter iter; struct te_peer_s *peer = NULL; if(te_targets == NULL) { te_targets = pcmk__strkey_table(NULL, te_peer_free); } g_hash_table_iter_init(&iter, te_targets); while (g_hash_table_iter_next(&iter, NULL, (gpointer *) & peer)) { peer->jobs = 0; peer->migrate_jobs = 0; } } static void te_update_job_count_on(const char *target, int offset, bool migrate) { struct te_peer_s *r = NULL; if(target == NULL || te_targets == NULL) { return; } r = g_hash_table_lookup(te_targets, target); if(r == NULL) { r = calloc(1, sizeof(struct te_peer_s)); r->name = strdup(target); g_hash_table_insert(te_targets, r->name, r); } r->jobs += offset; if(migrate) { r->migrate_jobs += offset; } crm_trace("jobs[%s] = %d", target, r->jobs); } static void te_update_job_count(crm_action_t * action, int offset) { const char *task = crm_element_value(action->xml, XML_LRM_ATTR_TASK); const char *target = crm_element_value(action->xml, XML_LRM_ATTR_TARGET); if (action->type != action_type_rsc || target == NULL) { /* No limit on these */ return; } /* if we have a router node, this means the action is performing * on a remote node. For now, we count all actions occurring on a * remote node against the job list on the cluster node hosting * the connection resources */ target = crm_element_value(action->xml, XML_LRM_ATTR_ROUTER_NODE); if ((target == NULL) && pcmk__strcase_any_of(task, CRMD_ACTION_MIGRATE, CRMD_ACTION_MIGRATED, NULL)) { const char *t1 = crm_meta_value(action->params, XML_LRM_ATTR_MIGRATE_SOURCE); const char *t2 = crm_meta_value(action->params, XML_LRM_ATTR_MIGRATE_TARGET); te_update_job_count_on(t1, offset, TRUE); te_update_job_count_on(t2, offset, TRUE); return; } else if (target == NULL) { target = crm_element_value(action->xml, XML_LRM_ATTR_TARGET); } te_update_job_count_on(target, offset, FALSE); } static gboolean te_should_perform_action_on(crm_graph_t * graph, crm_action_t * action, const char *target) { int limit = 0; struct te_peer_s *r = NULL; const char *task = crm_element_value(action->xml, XML_LRM_ATTR_TASK); const char *id = crm_element_value(action->xml, XML_LRM_ATTR_TASK_KEY); if(target == NULL) { /* No limit on these */ return TRUE; } else if(te_targets == NULL) { return FALSE; } r = g_hash_table_lookup(te_targets, target); limit = throttle_get_job_limit(target); if(r == NULL) { r = calloc(1, sizeof(struct te_peer_s)); r->name = strdup(target); g_hash_table_insert(te_targets, r->name, r); } if(limit <= r->jobs) { crm_trace("Peer %s is over their job limit of %d (%d): deferring %s", target, limit, r->jobs, id); return FALSE; } else if(graph->migration_limit > 0 && r->migrate_jobs >= graph->migration_limit) { if (pcmk__strcase_any_of(task, CRMD_ACTION_MIGRATE, CRMD_ACTION_MIGRATED, NULL)) { crm_trace("Peer %s is over their migration job limit of %d (%d): deferring %s", target, graph->migration_limit, r->migrate_jobs, id); return FALSE; } } crm_trace("Peer %s has not hit their limit yet. current jobs = %d limit= %d limit", target, r->jobs, limit); return TRUE; } static gboolean te_should_perform_action(crm_graph_t * graph, crm_action_t * action) { const char *target = NULL; const char *task = crm_element_value(action->xml, XML_LRM_ATTR_TASK); if (action->type != action_type_rsc) { /* No limit on these */ return TRUE; } /* if we have a router node, this means the action is performing * on a remote node. For now, we count all actions occurring on a * remote node against the job list on the cluster node hosting * the connection resources */ target = crm_element_value(action->xml, XML_LRM_ATTR_ROUTER_NODE); if ((target == NULL) && pcmk__strcase_any_of(task, CRMD_ACTION_MIGRATE, CRMD_ACTION_MIGRATED, NULL)) { target = crm_meta_value(action->params, XML_LRM_ATTR_MIGRATE_SOURCE); if(te_should_perform_action_on(graph, action, target) == FALSE) { return FALSE; } target = crm_meta_value(action->params, XML_LRM_ATTR_MIGRATE_TARGET); } else if (target == NULL) { target = crm_element_value(action->xml, XML_LRM_ATTR_TARGET); } return te_should_perform_action_on(graph, action, target); } /*! * \brief Confirm a graph action (and optionally update graph) * * \param[in] action Action to confirm * \param[in] graph Update and trigger this graph (if non-NULL) */ void te_action_confirmed(crm_action_t *action, crm_graph_t *graph) { if (action->confirmed == FALSE) { if ((action->type == action_type_rsc) && (crm_element_value(action->xml, XML_LRM_ATTR_TARGET) != NULL)) { te_update_job_count(action, -1); } action->confirmed = TRUE; } if (graph) { pcmk__update_graph(graph, action); trigger_graph(); } } crm_graph_functions_t te_graph_fns = { te_pseudo_action, te_rsc_command, te_crm_command, te_fence_node, te_should_perform_action, }; void notify_crmd(crm_graph_t * graph) { const char *type = "unknown"; enum crmd_fsa_input event = I_NULL; crm_debug("Processing transition completion in state %s", fsa_state2string(fsa_state)); if (graph->complete == FALSE) { CRM_CHECK(graph->complete,); graph->complete = TRUE; } switch (graph->completion_action) { case tg_stop: type = "stop"; if (fsa_state == S_TRANSITION_ENGINE) { event = I_TE_SUCCESS; } break; case tg_done: type = "done"; if (fsa_state == S_TRANSITION_ENGINE) { event = I_TE_SUCCESS; } break; case tg_restart: type = "restart"; if (fsa_state == S_TRANSITION_ENGINE) { if (transition_timer->period_ms > 0) { controld_stop_timer(transition_timer); controld_start_timer(transition_timer); } else { event = I_PE_CALC; } } else if (fsa_state == S_POLICY_ENGINE) { controld_set_fsa_action_flags(A_PE_INVOKE); trigger_fsa(); } break; case tg_shutdown: type = "shutdown"; if (pcmk_is_set(fsa_input_register, R_SHUTDOWN)) { event = I_STOP; } else { crm_err("We didn't ask to be shut down, yet the scheduler is telling us to"); event = I_TERMINATE; } } crm_debug("Transition %d status: %s - %s", graph->id, type, crm_str(graph->abort_reason)); graph->abort_reason = NULL; graph->completion_action = tg_done; controld_clear_fsa_input_flags(R_IN_TRANSITION); if (event != I_NULL) { register_fsa_input(C_FSA_INTERNAL, event, NULL); } else if (fsa_source) { mainloop_set_trigger(fsa_source); } } diff --git a/daemons/execd/execd_commands.c b/daemons/execd/execd_commands.c index c4ce5182f1..5727528c5c 100644 --- a/daemons/execd/execd_commands.c +++ b/daemons/execd/execd_commands.c @@ -1,2013 +1,2008 @@ /* * Copyright 2012-2021 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #include // Check whether we have a high-resolution monotonic clock #undef PCMK__TIME_USE_CGT #if HAVE_DECL_CLOCK_MONOTONIC && defined(CLOCK_MONOTONIC) # define PCMK__TIME_USE_CGT # include /* clock_gettime */ #endif #include #include #include #include #include #include #include #include #include "pacemaker-execd.h" #define EXIT_REASON_MAX_LEN 128 GHashTable *rsc_list = NULL; typedef struct lrmd_cmd_s { int timeout; guint interval_ms; int start_delay; int timeout_orig; int call_id; int call_opts; /* Timer ids, must be removed on cmd destruction. */ int delay_id; int stonith_recurring_id; int rsc_deleted; int service_flags; char *client_id; char *origin; char *rsc_id; char *action; char *real_action; char *userdata_str; pcmk__action_result_t result; /* We can track operation queue time and run time, to be saved with the CIB * resource history (and displayed in cluster status). We need * high-resolution monotonic time for this purpose, so we use * clock_gettime(CLOCK_MONOTONIC, ...) (if available, otherwise this feature * is disabled). * * However, we also need epoch timestamps for recording the time the command * last ran and the time its return value last changed, for use in time * displays (as opposed to interval calculations). We keep time_t values for * this purpose. * * The last run time is used for both purposes, so we keep redundant * monotonic and epoch values for this. Technically the two could represent * different times, but since time_t has only second resolution and the * values are used for distinct purposes, that is not significant. */ #ifdef PCMK__TIME_USE_CGT /* Recurring and systemd operations may involve more than one executor * command per operation, so they need info about the original and the most * recent. */ struct timespec t_first_run; // When op first ran struct timespec t_run; // When op most recently ran struct timespec t_first_queue; // When op was first queued struct timespec t_queue; // When op was most recently queued #endif time_t epoch_last_run; // Epoch timestamp of when op last ran time_t epoch_rcchange; // Epoch timestamp of when rc last changed bool first_notify_sent; int last_notify_rc; int last_notify_op_status; int last_pid; GHashTable *params; } lrmd_cmd_t; static void cmd_finalize(lrmd_cmd_t * cmd, lrmd_rsc_t * rsc); static gboolean lrmd_rsc_dispatch(gpointer user_data); static void cancel_all_recurring(lrmd_rsc_t * rsc, const char *client_id); #ifdef PCMK__TIME_USE_CGT /*! * \internal * \brief Check whether a struct timespec has been set * * \param[in] timespec Time to check * * \return true if timespec has been set (i.e. is nonzero), false otherwise */ static inline bool time_is_set(struct timespec *timespec) { return (timespec != NULL) && ((timespec->tv_sec != 0) || (timespec->tv_nsec != 0)); } /* * \internal * \brief Set a timespec (and its original if unset) to the current time * * \param[out] t_current Where to store current time * \param[out] t_orig Where to copy t_current if unset */ static void get_current_time(struct timespec *t_current, struct timespec *t_orig) { clock_gettime(CLOCK_MONOTONIC, t_current); if ((t_orig != NULL) && !time_is_set(t_orig)) { *t_orig = *t_current; } } /*! * \internal * \brief Return difference between two times in milliseconds * * \param[in] now More recent time (or NULL to use current time) * \param[in] old Earlier time * * \return milliseconds difference (or 0 if old is NULL or unset) * * \note Can overflow on 32bit machines when the differences is around * 24 days or more. */ static int time_diff_ms(struct timespec *now, struct timespec *old) { int diff_ms = 0; if (time_is_set(old)) { struct timespec local_now = { 0, }; if (now == NULL) { clock_gettime(CLOCK_MONOTONIC, &local_now); now = &local_now; } diff_ms = (now->tv_sec - old->tv_sec) * 1000 + (now->tv_nsec - old->tv_nsec) / 1000000; } return diff_ms; } /*! * \internal * \brief Reset a command's operation times to their original values. * * Reset a command's run and queued timestamps to the timestamps of the original * command, so we report the entire time since then and not just the time since * the most recent command (for recurring and systemd operations). * * \param[in] cmd Executor command object to reset * * \note It's not obvious what the queued time should be for a systemd * start/stop operation, which might go like this: * initial command queued 5ms, runs 3s * monitor command queued 10ms, runs 10s * monitor command queued 10ms, runs 10s * Is the queued time for that operation 5ms, 10ms or 25ms? The current * implementation will report 5ms. If it's 25ms, then we need to * subtract 20ms from the total exec time so as not to count it twice. * We can implement that later if it matters to anyone ... */ static void cmd_original_times(lrmd_cmd_t * cmd) { cmd->t_run = cmd->t_first_run; cmd->t_queue = cmd->t_first_queue; } #endif static inline bool action_matches(lrmd_cmd_t *cmd, const char *action, guint interval_ms) { return (cmd->interval_ms == interval_ms) && pcmk__str_eq(cmd->action, action, pcmk__str_casei); } static void log_finished(lrmd_cmd_t * cmd, int exec_time, int queue_time) { char pid_str[32] = { 0, }; int log_level = LOG_INFO; if (cmd->last_pid) { snprintf(pid_str, 32, "%d", cmd->last_pid); } if (pcmk__str_eq(cmd->action, "monitor", pcmk__str_casei)) { log_level = LOG_DEBUG; } #ifdef PCMK__TIME_USE_CGT do_crm_log(log_level, "%s %s (call %d%s%s) exited with status %d" " (execution time %dms, queue time %dms)", cmd->rsc_id, cmd->action, cmd->call_id, (cmd->last_pid? ", PID " : ""), pid_str, cmd->result.exit_status, exec_time, queue_time); #else do_crm_log(log_level, "%s %s (call %d%s%s) exited with status %d", cmd->rsc_id, cmd->action, cmd->call_id, (cmd->last_pid? ", PID " : ""), pid_str, cmd->result.exit_status); #endif } static void log_execute(lrmd_cmd_t * cmd) { int log_level = LOG_INFO; if (pcmk__str_eq(cmd->action, "monitor", pcmk__str_casei)) { log_level = LOG_DEBUG; } do_crm_log(log_level, "executing - rsc:%s action:%s call_id:%d", cmd->rsc_id, cmd->action, cmd->call_id); } static const char * normalize_action_name(lrmd_rsc_t * rsc, const char *action) { if (pcmk__str_eq(action, "monitor", pcmk__str_casei) && pcmk_is_set(pcmk_get_ra_caps(rsc->class), pcmk_ra_cap_status)) { return "status"; } return action; } static lrmd_rsc_t * build_rsc_from_xml(xmlNode * msg) { xmlNode *rsc_xml = get_xpath_object("//" F_LRMD_RSC, msg, LOG_ERR); lrmd_rsc_t *rsc = NULL; rsc = calloc(1, sizeof(lrmd_rsc_t)); crm_element_value_int(msg, F_LRMD_CALLOPTS, &rsc->call_opts); rsc->rsc_id = crm_element_value_copy(rsc_xml, F_LRMD_RSC_ID); rsc->class = crm_element_value_copy(rsc_xml, F_LRMD_CLASS); rsc->provider = crm_element_value_copy(rsc_xml, F_LRMD_PROVIDER); rsc->type = crm_element_value_copy(rsc_xml, F_LRMD_TYPE); rsc->work = mainloop_add_trigger(G_PRIORITY_HIGH, lrmd_rsc_dispatch, rsc); rsc->st_probe_rc = -ENODEV; // if stonith, initialize to "not running" return rsc; } static lrmd_cmd_t * create_lrmd_cmd(xmlNode *msg, pcmk__client_t *client) { int call_options = 0; xmlNode *rsc_xml = get_xpath_object("//" F_LRMD_RSC, msg, LOG_ERR); lrmd_cmd_t *cmd = NULL; cmd = calloc(1, sizeof(lrmd_cmd_t)); crm_element_value_int(msg, F_LRMD_CALLOPTS, &call_options); cmd->call_opts = call_options; cmd->client_id = strdup(client->id); crm_element_value_int(msg, F_LRMD_CALLID, &cmd->call_id); crm_element_value_ms(rsc_xml, F_LRMD_RSC_INTERVAL, &cmd->interval_ms); crm_element_value_int(rsc_xml, F_LRMD_TIMEOUT, &cmd->timeout); crm_element_value_int(rsc_xml, F_LRMD_RSC_START_DELAY, &cmd->start_delay); cmd->timeout_orig = cmd->timeout; cmd->origin = crm_element_value_copy(rsc_xml, F_LRMD_ORIGIN); cmd->action = crm_element_value_copy(rsc_xml, F_LRMD_RSC_ACTION); cmd->userdata_str = crm_element_value_copy(rsc_xml, F_LRMD_RSC_USERDATA_STR); cmd->rsc_id = crm_element_value_copy(rsc_xml, F_LRMD_RSC_ID); cmd->params = xml2list(rsc_xml); if (pcmk__str_eq(g_hash_table_lookup(cmd->params, "CRM_meta_on_fail"), "block", pcmk__str_casei)) { crm_debug("Setting flag to leave pid group on timeout and " "only kill action pid for " PCMK__OP_FMT, cmd->rsc_id, cmd->action, cmd->interval_ms); cmd->service_flags = pcmk__set_flags_as(__func__, __LINE__, LOG_TRACE, "Action", cmd->action, 0, SVC_ACTION_LEAVE_GROUP, "SVC_ACTION_LEAVE_GROUP"); } return cmd; } static void stop_recurring_timer(lrmd_cmd_t *cmd) { if (cmd) { if (cmd->stonith_recurring_id) { g_source_remove(cmd->stonith_recurring_id); } cmd->stonith_recurring_id = 0; } } static void free_lrmd_cmd(lrmd_cmd_t * cmd) { stop_recurring_timer(cmd); if (cmd->delay_id) { g_source_remove(cmd->delay_id); } if (cmd->params) { g_hash_table_destroy(cmd->params); } pcmk__reset_result(&(cmd->result)); free(cmd->origin); free(cmd->action); free(cmd->real_action); free(cmd->userdata_str); free(cmd->rsc_id); free(cmd->client_id); free(cmd); } static gboolean stonith_recurring_op_helper(gpointer data) { lrmd_cmd_t *cmd = data; lrmd_rsc_t *rsc; cmd->stonith_recurring_id = 0; if (!cmd->rsc_id) { return FALSE; } rsc = g_hash_table_lookup(rsc_list, cmd->rsc_id); CRM_ASSERT(rsc != NULL); /* take it out of recurring_ops list, and put it in the pending ops * to be executed */ rsc->recurring_ops = g_list_remove(rsc->recurring_ops, cmd); rsc->pending_ops = g_list_append(rsc->pending_ops, cmd); #ifdef PCMK__TIME_USE_CGT get_current_time(&(cmd->t_queue), &(cmd->t_first_queue)); #endif mainloop_set_trigger(rsc->work); return FALSE; } static inline void start_recurring_timer(lrmd_cmd_t *cmd) { if (cmd && (cmd->interval_ms > 0)) { cmd->stonith_recurring_id = g_timeout_add(cmd->interval_ms, stonith_recurring_op_helper, cmd); } } static gboolean start_delay_helper(gpointer data) { lrmd_cmd_t *cmd = data; lrmd_rsc_t *rsc = NULL; cmd->delay_id = 0; rsc = cmd->rsc_id ? g_hash_table_lookup(rsc_list, cmd->rsc_id) : NULL; if (rsc) { mainloop_set_trigger(rsc->work); } return FALSE; } /*! * \internal * \brief Check whether a list already contains the equivalent of a given action */ static lrmd_cmd_t * find_duplicate_action(GList *action_list, lrmd_cmd_t *cmd) { for (GList *item = action_list; item != NULL; item = item->next) { lrmd_cmd_t *dup = item->data; if (action_matches(cmd, dup->action, dup->interval_ms)) { return dup; } } return NULL; } static bool merge_recurring_duplicate(lrmd_rsc_t * rsc, lrmd_cmd_t * cmd) { lrmd_cmd_t * dup = NULL; bool dup_pending = true; if (cmd->interval_ms == 0) { return false; } // Search for a duplicate of this action (in-flight or not) dup = find_duplicate_action(rsc->pending_ops, cmd); if (dup == NULL) { dup_pending = false; dup = find_duplicate_action(rsc->recurring_ops, cmd); if (dup == NULL) { return false; } } /* Do not merge fencing monitors marked for cancellation, so we can reply to * the cancellation separately. */ if (pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei) && (dup->result.execution_status == PCMK_EXEC_CANCELLED)) { return false; } /* This should not occur. If it does, we need to investigate how something * like this is possible in the controller. */ crm_warn("Duplicate recurring op entry detected (" PCMK__OP_FMT "), merging with previous op entry", rsc->rsc_id, normalize_action_name(rsc, dup->action), dup->interval_ms); // Merge new action's call ID and user data into existing action dup->first_notify_sent = false; free(dup->userdata_str); dup->userdata_str = cmd->userdata_str; cmd->userdata_str = NULL; dup->call_id = cmd->call_id; free_lrmd_cmd(cmd); cmd = NULL; /* If dup is not pending, that means it has already executed at least once * and is waiting in the interval. In that case, stop waiting and initiate * a new instance now. */ if (!dup_pending) { if (pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei)) { stop_recurring_timer(dup); stonith_recurring_op_helper(dup); } else { services_action_kick(rsc->rsc_id, normalize_action_name(rsc, dup->action), dup->interval_ms); } } return true; } static void schedule_lrmd_cmd(lrmd_rsc_t * rsc, lrmd_cmd_t * cmd) { CRM_CHECK(cmd != NULL, return); CRM_CHECK(rsc != NULL, return); crm_trace("Scheduling %s on %s", cmd->action, rsc->rsc_id); if (merge_recurring_duplicate(rsc, cmd)) { // Equivalent of cmd has already been scheduled return; } /* The controller expects the executor to automatically cancel * recurring operations before a resource stops. */ if (pcmk__str_eq(cmd->action, "stop", pcmk__str_casei)) { cancel_all_recurring(rsc, NULL); } rsc->pending_ops = g_list_append(rsc->pending_ops, cmd); #ifdef PCMK__TIME_USE_CGT get_current_time(&(cmd->t_queue), &(cmd->t_first_queue)); #endif mainloop_set_trigger(rsc->work); if (cmd->start_delay) { cmd->delay_id = g_timeout_add(cmd->start_delay, start_delay_helper, cmd); } } static xmlNode * create_lrmd_reply(const char *origin, int rc, int call_id) { xmlNode *reply = create_xml_node(NULL, T_LRMD_REPLY); crm_xml_add(reply, F_LRMD_ORIGIN, origin); crm_xml_add_int(reply, F_LRMD_RC, rc); crm_xml_add_int(reply, F_LRMD_CALLID, call_id); return reply; } static void send_client_notify(gpointer key, gpointer value, gpointer user_data) { xmlNode *update_msg = user_data; pcmk__client_t *client = value; int rc; int log_level = LOG_WARNING; const char *msg = NULL; CRM_CHECK(client != NULL, return); if (client->name == NULL) { crm_trace("Skipping notification to client without name"); return; } if (pcmk_is_set(client->flags, pcmk__client_to_proxy)) { /* We only want to notify clients of the executor IPC API. If we are * running as Pacemaker Remote, we may have clients proxied to other * IPC services in the cluster, so skip those. */ crm_trace("Skipping executor API notification to client %s", pcmk__client_name(client)); return; } rc = lrmd_server_send_notify(client, update_msg); if (rc == pcmk_rc_ok) { return; } switch (rc) { case ENOTCONN: case EPIPE: // Client exited without waiting for notification log_level = LOG_INFO; msg = "Disconnected"; break; default: msg = pcmk_rc_str(rc); break; } do_crm_log(log_level, "Could not notify client %s: %s " CRM_XS " rc=%d", pcmk__client_name(client), msg, rc); } static void send_cmd_complete_notify(lrmd_cmd_t * cmd) { xmlNode *notify = NULL; #ifdef PCMK__TIME_USE_CGT int exec_time = time_diff_ms(NULL, &(cmd->t_run)); int queue_time = time_diff_ms(&cmd->t_run, &(cmd->t_queue)); log_finished(cmd, exec_time, queue_time); #else log_finished(cmd, 0, 0); #endif /* if the first notify result for a cmd has already been sent earlier, and the * the option to only send notifies on result changes is set. Check to see * if the last result is the same as the new one. If so, suppress this update */ if (cmd->first_notify_sent && (cmd->call_opts & lrmd_opt_notify_changes_only)) { if ((cmd->last_notify_rc == cmd->result.exit_status) && (cmd->last_notify_op_status == cmd->result.execution_status)) { /* only send changes */ return; } } cmd->first_notify_sent = true; cmd->last_notify_rc = cmd->result.exit_status; cmd->last_notify_op_status = cmd->result.execution_status; notify = create_xml_node(NULL, T_LRMD_NOTIFY); crm_xml_add(notify, F_LRMD_ORIGIN, __func__); crm_xml_add_int(notify, F_LRMD_TIMEOUT, cmd->timeout); crm_xml_add_ms(notify, F_LRMD_RSC_INTERVAL, cmd->interval_ms); crm_xml_add_int(notify, F_LRMD_RSC_START_DELAY, cmd->start_delay); crm_xml_add_int(notify, F_LRMD_EXEC_RC, cmd->result.exit_status); crm_xml_add_int(notify, F_LRMD_OP_STATUS, cmd->result.execution_status); crm_xml_add_int(notify, F_LRMD_CALLID, cmd->call_id); crm_xml_add_int(notify, F_LRMD_RSC_DELETED, cmd->rsc_deleted); crm_xml_add_ll(notify, F_LRMD_RSC_RUN_TIME, (long long) cmd->epoch_last_run); crm_xml_add_ll(notify, F_LRMD_RSC_RCCHANGE_TIME, (long long) cmd->epoch_rcchange); #ifdef PCMK__TIME_USE_CGT crm_xml_add_int(notify, F_LRMD_RSC_EXEC_TIME, exec_time); crm_xml_add_int(notify, F_LRMD_RSC_QUEUE_TIME, queue_time); #endif crm_xml_add(notify, F_LRMD_OPERATION, LRMD_OP_RSC_EXEC); crm_xml_add(notify, F_LRMD_RSC_ID, cmd->rsc_id); if(cmd->real_action) { crm_xml_add(notify, F_LRMD_RSC_ACTION, cmd->real_action); } else { crm_xml_add(notify, F_LRMD_RSC_ACTION, cmd->action); } crm_xml_add(notify, F_LRMD_RSC_USERDATA_STR, cmd->userdata_str); crm_xml_add(notify, F_LRMD_RSC_EXIT_REASON, cmd->result.exit_reason); if (cmd->result.action_stderr != NULL) { crm_xml_add(notify, F_LRMD_RSC_OUTPUT, cmd->result.action_stderr); } else if (cmd->result.action_stdout != NULL) { crm_xml_add(notify, F_LRMD_RSC_OUTPUT, cmd->result.action_stdout); } if (cmd->params) { char *key = NULL; char *value = NULL; GHashTableIter iter; xmlNode *args = create_xml_node(notify, XML_TAG_ATTRS); g_hash_table_iter_init(&iter, cmd->params); while (g_hash_table_iter_next(&iter, (gpointer *) & key, (gpointer *) & value)) { hash2smartfield((gpointer) key, (gpointer) value, args); } } if (cmd->client_id && (cmd->call_opts & lrmd_opt_notify_orig_only)) { pcmk__client_t *client = pcmk__find_client_by_id(cmd->client_id); if (client) { send_client_notify(client->id, client, notify); } } else { pcmk__foreach_ipc_client(send_client_notify, notify); } free_xml(notify); } static void send_generic_notify(int rc, xmlNode * request) { if (pcmk__ipc_client_count() != 0) { int call_id = 0; xmlNode *notify = NULL; xmlNode *rsc_xml = get_xpath_object("//" F_LRMD_RSC, request, LOG_ERR); const char *rsc_id = crm_element_value(rsc_xml, F_LRMD_RSC_ID); const char *op = crm_element_value(request, F_LRMD_OPERATION); crm_element_value_int(request, F_LRMD_CALLID, &call_id); notify = create_xml_node(NULL, T_LRMD_NOTIFY); crm_xml_add(notify, F_LRMD_ORIGIN, __func__); crm_xml_add_int(notify, F_LRMD_RC, rc); crm_xml_add_int(notify, F_LRMD_CALLID, call_id); crm_xml_add(notify, F_LRMD_OPERATION, op); crm_xml_add(notify, F_LRMD_RSC_ID, rsc_id); pcmk__foreach_ipc_client(send_client_notify, notify); free_xml(notify); } } static void cmd_reset(lrmd_cmd_t * cmd) { cmd->last_pid = 0; #ifdef PCMK__TIME_USE_CGT memset(&cmd->t_run, 0, sizeof(cmd->t_run)); memset(&cmd->t_queue, 0, sizeof(cmd->t_queue)); #endif cmd->epoch_last_run = 0; pcmk__reset_result(&(cmd->result)); cmd->result.execution_status = PCMK_EXEC_DONE; } static void cmd_finalize(lrmd_cmd_t * cmd, lrmd_rsc_t * rsc) { crm_trace("Resource operation rsc:%s action:%s completed (%p %p)", cmd->rsc_id, cmd->action, rsc ? rsc->active : NULL, cmd); if (rsc && (rsc->active == cmd)) { rsc->active = NULL; mainloop_set_trigger(rsc->work); } if (!rsc) { cmd->rsc_deleted = 1; } /* reset original timeout so client notification has correct information */ cmd->timeout = cmd->timeout_orig; send_cmd_complete_notify(cmd); if ((cmd->interval_ms != 0) && (cmd->result.execution_status == PCMK_EXEC_CANCELLED)) { if (rsc) { rsc->recurring_ops = g_list_remove(rsc->recurring_ops, cmd); rsc->pending_ops = g_list_remove(rsc->pending_ops, cmd); } free_lrmd_cmd(cmd); } else if (cmd->interval_ms == 0) { if (rsc) { rsc->pending_ops = g_list_remove(rsc->pending_ops, cmd); } free_lrmd_cmd(cmd); } else { /* Clear all the values pertaining just to the last iteration of a recurring op. */ cmd_reset(cmd); } } static int ocf2uniform_rc(int rc) { switch (rc) { case PCMK_OCF_DEGRADED: case PCMK_OCF_DEGRADED_PROMOTED: break; default: if (rc < 0 || rc > PCMK_OCF_FAILED_PROMOTED) { return PCMK_OCF_UNKNOWN_ERROR; } } return rc; } static int stonith2uniform_rc(const char *action, int rc) { switch (rc) { case pcmk_ok: rc = PCMK_OCF_OK; break; case -ENODEV: /* This should be possible only for probes in practice, but * interpret for all actions to be safe. */ if (pcmk__str_eq(action, "monitor", pcmk__str_casei)) { rc = PCMK_OCF_NOT_RUNNING; } else if (pcmk__str_eq(action, "stop", pcmk__str_casei)) { rc = PCMK_OCF_OK; } else { rc = PCMK_OCF_NOT_INSTALLED; } break; case -EOPNOTSUPP: rc = PCMK_OCF_UNIMPLEMENT_FEATURE; break; - case -ETIME: - case -ETIMEDOUT: - rc = PCMK_OCF_TIMEOUT; - break; - default: rc = PCMK_OCF_UNKNOWN_ERROR; break; } return rc; } #if SUPPORT_NAGIOS static int nagios2uniform_rc(const char *action, int rc) { if (rc < 0) { return PCMK_OCF_UNKNOWN_ERROR; } switch (rc) { case NAGIOS_STATE_OK: return PCMK_OCF_OK; case NAGIOS_INSUFFICIENT_PRIV: return PCMK_OCF_INSUFFICIENT_PRIV; case NAGIOS_NOT_INSTALLED: return PCMK_OCF_NOT_INSTALLED; case NAGIOS_STATE_WARNING: case NAGIOS_STATE_CRITICAL: case NAGIOS_STATE_UNKNOWN: case NAGIOS_STATE_DEPENDENT: default: return PCMK_OCF_UNKNOWN_ERROR; } return PCMK_OCF_UNKNOWN_ERROR; } #endif static int get_uniform_rc(const char *standard, const char *action, int rc) { if (pcmk__str_eq(standard, PCMK_RESOURCE_CLASS_OCF, pcmk__str_casei)) { return ocf2uniform_rc(rc); } else if (pcmk__str_eq(standard, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei)) { return stonith2uniform_rc(action, rc); } else if (pcmk__str_eq(standard, PCMK_RESOURCE_CLASS_SYSTEMD, pcmk__str_casei)) { return rc; } else if (pcmk__str_eq(standard, PCMK_RESOURCE_CLASS_UPSTART, pcmk__str_casei)) { return rc; #if SUPPORT_NAGIOS } else if (pcmk__str_eq(standard, PCMK_RESOURCE_CLASS_NAGIOS, pcmk__str_casei)) { return nagios2uniform_rc(action, rc); #endif } else { return services_get_ocf_exitcode(action, rc); } } static int action_get_uniform_rc(svc_action_t * action) { lrmd_cmd_t *cmd = action->cb_data; return get_uniform_rc(action->standard, cmd->action, action->rc); } struct notify_new_client_data { xmlNode *notify; pcmk__client_t *new_client; }; static void notify_one_client(gpointer key, gpointer value, gpointer user_data) { pcmk__client_t *client = value; struct notify_new_client_data *data = user_data; if (!pcmk__str_eq(client->id, data->new_client->id, pcmk__str_casei)) { send_client_notify(key, (gpointer) client, (gpointer) data->notify); } } void notify_of_new_client(pcmk__client_t *new_client) { struct notify_new_client_data data; data.new_client = new_client; data.notify = create_xml_node(NULL, T_LRMD_NOTIFY); crm_xml_add(data.notify, F_LRMD_ORIGIN, __func__); crm_xml_add(data.notify, F_LRMD_OPERATION, LRMD_OP_NEW_CLIENT); pcmk__foreach_ipc_client(notify_one_client, &data); free_xml(data.notify); } static char * parse_exit_reason(const char *output) { const char *cur = NULL; const char *last = NULL; static int cookie_len = 0; char *eol = NULL; size_t reason_len = EXIT_REASON_MAX_LEN; if (output == NULL) { return NULL; } if (!cookie_len) { cookie_len = strlen(PCMK_OCF_REASON_PREFIX); } cur = strstr(output, PCMK_OCF_REASON_PREFIX); for (; cur != NULL; cur = strstr(cur, PCMK_OCF_REASON_PREFIX)) { /* skip over the cookie delimiter string */ cur += cookie_len; last = cur; } if (last == NULL) { return NULL; } // Truncate everything after a new line, and limit reason string size eol = strchr(last, '\n'); if (eol) { reason_len = QB_MIN(reason_len, eol - last); } return strndup(last, reason_len); } void client_disconnect_cleanup(const char *client_id) { GHashTableIter iter; lrmd_rsc_t *rsc = NULL; char *key = NULL; g_hash_table_iter_init(&iter, rsc_list); while (g_hash_table_iter_next(&iter, (gpointer *) & key, (gpointer *) & rsc)) { if (rsc->call_opts & lrmd_opt_drop_recurring) { /* This client is disconnecting, drop any recurring operations * it may have initiated on the resource */ cancel_all_recurring(rsc, client_id); } } } static void action_complete(svc_action_t * action) { lrmd_rsc_t *rsc; lrmd_cmd_t *cmd = action->cb_data; #ifdef PCMK__TIME_USE_CGT const char *rclass = NULL; bool goagain = false; #endif if (!cmd) { crm_err("Completed executor action (%s) does not match any known operations", action->id); return; } #ifdef PCMK__TIME_USE_CGT if (cmd->result.exit_status != action->rc) { cmd->epoch_rcchange = time(NULL); } #endif cmd->last_pid = action->pid; pcmk__set_result(&(cmd->result), action_get_uniform_rc(action), action->status, NULL); rsc = cmd->rsc_id ? g_hash_table_lookup(rsc_list, cmd->rsc_id) : NULL; #ifdef PCMK__TIME_USE_CGT if (rsc && pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_SERVICE, pcmk__str_casei)) { rclass = resources_find_service_class(rsc->type); } else if(rsc) { rclass = rsc->class; } if (pcmk__str_eq(rclass, PCMK_RESOURCE_CLASS_SYSTEMD, pcmk__str_casei)) { if ((cmd->result.exit_status == PCMK_OCF_OK) && pcmk__strcase_any_of(cmd->action, "start", "stop", NULL)) { /* systemd returns from start and stop actions after the action * begins, not after it completes. We have to jump through a few * hoops so that we don't report 'complete' to the rest of pacemaker * until it's actually done. */ goagain = true; cmd->real_action = cmd->action; cmd->action = strdup("monitor"); } else if (cmd->real_action != NULL) { // This is follow-up monitor to check whether start/stop completed if (cmd->result.execution_status == PCMK_EXEC_PENDING) { goagain = true; } else if ((cmd->result.exit_status == PCMK_OCF_OK) && pcmk__str_eq(cmd->real_action, "stop", pcmk__str_casei)) { goagain = true; } else { int time_sum = time_diff_ms(NULL, &(cmd->t_first_run)); int timeout_left = cmd->timeout_orig - time_sum; crm_debug("%s systemd %s is now complete (elapsed=%dms, " "remaining=%dms): %s (%d)", cmd->rsc_id, cmd->real_action, time_sum, timeout_left, services_ocf_exitcode_str(cmd->result.exit_status), cmd->result.exit_status); cmd_original_times(cmd); // Monitors may return "not running", but start/stop shouldn't if ((cmd->result.execution_status == PCMK_EXEC_DONE) && (cmd->result.exit_status == PCMK_OCF_NOT_RUNNING)) { if (pcmk__str_eq(cmd->real_action, "start", pcmk__str_casei)) { cmd->result.exit_status = PCMK_OCF_UNKNOWN_ERROR; } else if (pcmk__str_eq(cmd->real_action, "stop", pcmk__str_casei)) { cmd->result.exit_status = PCMK_OCF_OK; } } } } } #endif #if SUPPORT_NAGIOS if (rsc && pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_NAGIOS, pcmk__str_casei)) { if (action_matches(cmd, "monitor", 0) && (cmd->result.exit_status == PCMK_OCF_OK)) { /* Successfully executed --version for the nagios plugin */ cmd->result.exit_status = PCMK_OCF_NOT_RUNNING; } else if (pcmk__str_eq(cmd->action, "start", pcmk__str_casei) && (cmd->result.exit_status != PCMK_OCF_OK)) { #ifdef PCMK__TIME_USE_CGT goagain = true; #endif } } #endif #ifdef PCMK__TIME_USE_CGT if (goagain) { int time_sum = time_diff_ms(NULL, &(cmd->t_first_run)); int timeout_left = cmd->timeout_orig - time_sum; int delay = cmd->timeout_orig / 10; if(delay >= timeout_left && timeout_left > 20) { delay = timeout_left/2; } delay = QB_MIN(2000, delay); if (delay < timeout_left) { cmd->start_delay = delay; cmd->timeout = timeout_left; if (cmd->result.exit_status == PCMK_OCF_OK) { crm_debug("%s %s may still be in progress: re-scheduling (elapsed=%dms, remaining=%dms, start_delay=%dms)", cmd->rsc_id, cmd->real_action, time_sum, timeout_left, delay); } else if (cmd->result.execution_status == PCMK_EXEC_PENDING) { crm_info("%s %s is still in progress: re-scheduling (elapsed=%dms, remaining=%dms, start_delay=%dms)", cmd->rsc_id, cmd->action, time_sum, timeout_left, delay); } else { crm_notice("%s %s failed '%s' (%d): re-scheduling (elapsed=%dms, remaining=%dms, start_delay=%dms)", cmd->rsc_id, cmd->action, services_ocf_exitcode_str(cmd->result.exit_status), cmd->result.exit_status, time_sum, timeout_left, delay); } cmd_reset(cmd); if(rsc) { rsc->active = NULL; } schedule_lrmd_cmd(rsc, cmd); /* Don't finalize cmd, we're not done with it yet */ return; } else { crm_notice("Giving up on %s %s (rc=%d): timeout (elapsed=%dms, remaining=%dms)", cmd->rsc_id, (cmd->real_action? cmd->real_action : cmd->action), cmd->result.exit_status, time_sum, timeout_left); - pcmk__set_result(&(cmd->result), PCMK_OCF_TIMEOUT, + pcmk__set_result(&(cmd->result), PCMK_OCF_UNKNOWN_ERROR, PCMK_EXEC_TIMEOUT, "Investigate reason for timeout, and adjust " "configured operation timeout if necessary"); cmd_original_times(cmd); } } #endif pcmk__set_result_output(&(cmd->result), action->stdout_data, action->stderr_data); if (action->stderr_data) { cmd->result.exit_reason = parse_exit_reason(action->stderr_data); } cmd_finalize(cmd, rsc); } /*! * \internal * \brief Determine operation status of a stonith operation * * Non-stonith resource operations get their operation status directly from the * service library, but the fencer does not have an equivalent, so we must infer * an operation status from the fencer API's return code. * * \param[in] action Name of action performed on stonith resource * \param[in] interval_ms Action interval * \param[in] rc Action result from fencer * * \return Operation status corresponding to fencer API return code */ static int stonith_rc2status(const char *action, guint interval_ms, int rc) { int status = PCMK_EXEC_DONE; switch (rc) { case pcmk_ok: break; case -EOPNOTSUPP: case -EPROTONOSUPPORT: status = PCMK_EXEC_NOT_SUPPORTED; break; case -ETIME: case -ETIMEDOUT: status = PCMK_EXEC_TIMEOUT; break; case -ENOTCONN: case -ECOMM: // Couldn't talk to fencer status = PCMK_EXEC_ERROR; break; case -ENODEV: // The device is not registered with the fencer status = PCMK_EXEC_ERROR; break; default: break; } return status; } static void stonith_action_complete(lrmd_cmd_t * cmd, int rc) { // This can be NULL if resource was removed before command completed lrmd_rsc_t *rsc = g_hash_table_lookup(rsc_list, cmd->rsc_id); cmd->result.exit_status = stonith2uniform_rc(cmd->action, rc); /* This function may be called with status already set to cancelled, if a * pending action was aborted. Otherwise, we need to determine status from * the fencer return code. */ if (cmd->result.execution_status != PCMK_EXEC_CANCELLED) { cmd->result.execution_status = stonith_rc2status(cmd->action, cmd->interval_ms, rc); // Certain successful actions change the known state of the resource if ((rsc != NULL) && (cmd->result.exit_status == PCMK_OCF_OK)) { if (pcmk__str_eq(cmd->action, "start", pcmk__str_casei)) { rsc->st_probe_rc = pcmk_ok; // maps to PCMK_OCF_OK } else if (pcmk__str_eq(cmd->action, "stop", pcmk__str_casei)) { rsc->st_probe_rc = -ENODEV; // maps to PCMK_OCF_NOT_RUNNING } } } // Give the user more detail than an OCF code if (rc != -pcmk_err_generic) { cmd->result.exit_reason = strdup(pcmk_strerror(rc)); } /* The recurring timer should not be running at this point in any case, but * as a failsafe, stop it if it is. */ stop_recurring_timer(cmd); /* Reschedule this command if appropriate. If a recurring command is *not* * rescheduled, its status must be PCMK_EXEC_CANCELLED, otherwise it will * not be removed from recurring_ops by cmd_finalize(). */ if (rsc && (cmd->interval_ms > 0) && (cmd->result.execution_status != PCMK_EXEC_CANCELLED)) { start_recurring_timer(cmd); } cmd_finalize(cmd, rsc); } static void lrmd_stonith_callback(stonith_t * stonith, stonith_callback_data_t * data) { stonith_action_complete(data->userdata, data->rc); } void stonith_connection_failed(void) { GHashTableIter iter; GList *cmd_list = NULL; GList *cmd_iter = NULL; lrmd_rsc_t *rsc = NULL; char *key = NULL; g_hash_table_iter_init(&iter, rsc_list); while (g_hash_table_iter_next(&iter, (gpointer *) & key, (gpointer *) & rsc)) { if (pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei)) { /* If we registered this fence device, we don't know whether the * fencer still has the registration or not. Cause future probes to * return PCMK_OCF_UNKNOWN_ERROR until the resource is stopped or * started successfully. This is especially important if the * controller also went away (possibly due to a cluster layer * restart) and won't receive our client notification of any * monitors finalized below. */ if (rsc->st_probe_rc == pcmk_ok) { rsc->st_probe_rc = pcmk_err_generic; } if (rsc->active) { cmd_list = g_list_append(cmd_list, rsc->active); } if (rsc->recurring_ops) { cmd_list = g_list_concat(cmd_list, rsc->recurring_ops); } if (rsc->pending_ops) { cmd_list = g_list_concat(cmd_list, rsc->pending_ops); } rsc->pending_ops = rsc->recurring_ops = NULL; } } if (!cmd_list) { return; } crm_err("Connection to fencer failed, finalizing %d pending operations", g_list_length(cmd_list)); for (cmd_iter = cmd_list; cmd_iter; cmd_iter = cmd_iter->next) { stonith_action_complete(cmd_iter->data, -ENOTCONN); } g_list_free(cmd_list); } /*! * \internal * \brief Execute a stonith resource "start" action * * Start a stonith resource by registering it with the fencer. * (Stonith agents don't have a start command.) * * \param[in] stonith_api Connection to fencer * \param[in] rsc Stonith resource to start * \param[in] cmd Start command to execute * * \return pcmk_ok on success, -errno otherwise */ static int execd_stonith_start(stonith_t *stonith_api, lrmd_rsc_t *rsc, lrmd_cmd_t *cmd) { char *key = NULL; char *value = NULL; stonith_key_value_t *device_params = NULL; int rc = pcmk_ok; // Convert command parameters to stonith API key/values if (cmd->params) { GHashTableIter iter; g_hash_table_iter_init(&iter, cmd->params); while (g_hash_table_iter_next(&iter, (gpointer *) & key, (gpointer *) & value)) { device_params = stonith_key_value_add(device_params, key, value); } } /* The fencer will automatically register devices via CIB notifications * when the CIB changes, but to avoid a possible race condition between * the fencer receiving the notification and the executor requesting that * resource, the executor registers the device as well. The fencer knows how * to handle duplicate registrations. */ rc = stonith_api->cmds->register_device(stonith_api, st_opt_sync_call, cmd->rsc_id, rsc->provider, rsc->type, device_params); stonith_key_value_freeall(device_params, 1, 1); return rc; } /*! * \internal * \brief Execute a stonith resource "stop" action * * Stop a stonith resource by unregistering it with the fencer. * (Stonith agents don't have a stop command.) * * \param[in] stonith_api Connection to fencer * \param[in] rsc Stonith resource to stop * * \return pcmk_ok on success, -errno otherwise */ static inline int execd_stonith_stop(stonith_t *stonith_api, const lrmd_rsc_t *rsc) { /* @TODO Failure would indicate a problem communicating with fencer; * perhaps we should try reconnecting and retrying a few times? */ return stonith_api->cmds->remove_device(stonith_api, st_opt_sync_call, rsc->rsc_id); } /*! * \internal * \brief Initiate a stonith resource agent recurring "monitor" action * * \param[in] stonith_api Connection to fencer * \param[in] rsc Stonith resource to monitor * \param[in] cmd Monitor command being executed * * \return pcmk_ok if monitor was successfully initiated, -errno otherwise */ static inline int execd_stonith_monitor(stonith_t *stonith_api, lrmd_rsc_t *rsc, lrmd_cmd_t *cmd) { int rc = stonith_api->cmds->monitor(stonith_api, 0, cmd->rsc_id, cmd->timeout / 1000); rc = stonith_api->cmds->register_callback(stonith_api, rc, 0, 0, cmd, "lrmd_stonith_callback", lrmd_stonith_callback); if (rc == TRUE) { rsc->active = cmd; rc = pcmk_ok; } else { rc = -pcmk_err_generic; } return rc; } static void lrmd_rsc_execute_stonith(lrmd_rsc_t * rsc, lrmd_cmd_t * cmd) { int rc = 0; bool do_monitor = FALSE; stonith_t *stonith_api = get_stonith_connection(); if (!stonith_api) { rc = -ENOTCONN; } else if (pcmk__str_eq(cmd->action, "start", pcmk__str_casei)) { rc = execd_stonith_start(stonith_api, rsc, cmd); if (rc == 0) { do_monitor = TRUE; } } else if (pcmk__str_eq(cmd->action, "stop", pcmk__str_casei)) { rc = execd_stonith_stop(stonith_api, rsc); } else if (pcmk__str_eq(cmd->action, "monitor", pcmk__str_casei)) { if (cmd->interval_ms > 0) { do_monitor = TRUE; } else { rc = rsc->st_probe_rc; } } if (do_monitor) { rc = execd_stonith_monitor(stonith_api, rsc, cmd); if (rc == pcmk_ok) { // Don't clean up yet, we will find out result of the monitor later return; } } stonith_action_complete(cmd, rc); } static int lrmd_rsc_execute_service_lib(lrmd_rsc_t * rsc, lrmd_cmd_t * cmd) { svc_action_t *action = NULL; GHashTable *params_copy = NULL; CRM_ASSERT(rsc); CRM_ASSERT(cmd); crm_trace("Creating action, resource:%s action:%s class:%s provider:%s agent:%s", rsc->rsc_id, cmd->action, rsc->class, rsc->provider, rsc->type); #if SUPPORT_NAGIOS /* Recurring operations are cancelled anyway for a stop operation */ if (pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_NAGIOS, pcmk__str_casei) && pcmk__str_eq(cmd->action, "stop", pcmk__str_casei)) { cmd->result.exit_status = PCMK_OCF_OK; goto exec_done; } #endif params_copy = pcmk__str_table_dup(cmd->params); action = services__create_resource_action(rsc->rsc_id, rsc->class, rsc->provider, rsc->type, normalize_action_name(rsc, cmd->action), cmd->interval_ms, cmd->timeout, params_copy, cmd->service_flags); if (!action) { // Invalid arguments (which would be a bug) or out-of-memory crm_err("Failed to create action, action:%s on resource %s", cmd->action, rsc->rsc_id); pcmk__set_result(&(cmd->result), PCMK_OCF_UNKNOWN_ERROR, PCMK_EXEC_ERROR, "Internal Pacemaker error"); goto exec_done; } if (action->rc != PCMK_OCF_OK) { pcmk__set_result(&(cmd->result), action->rc, action->status, NULL); services_action_free(action); goto exec_done; } action->cb_data = cmd; /* 'cmd' may not be valid after this point if * services_action_async() returned TRUE * * Upstart and systemd both synchronously determine monitor/status * results and call action_complete (which may free 'cmd') if necessary. */ if (services_action_async(action, action_complete)) { return TRUE; } /* Asynchronous execution could not be initiated. services_action_async() * should have already set an appropriate execution status, but as a * fail-safe for cases where we've neglected to do so, make sure this is * considered an execution error. */ if (action->status == PCMK_EXEC_DONE) { action->status = PCMK_EXEC_ERROR; } pcmk__set_result(&(cmd->result), action->rc, action->status, NULL); services_action_free(action); action = NULL; exec_done: cmd_finalize(cmd, rsc); return TRUE; } static gboolean lrmd_rsc_execute(lrmd_rsc_t * rsc) { lrmd_cmd_t *cmd = NULL; CRM_CHECK(rsc != NULL, return FALSE); if (rsc->active) { crm_trace("%s is still active", rsc->rsc_id); return TRUE; } if (rsc->pending_ops) { GList *first = rsc->pending_ops; cmd = first->data; if (cmd->delay_id) { crm_trace ("Command %s %s was asked to run too early, waiting for start_delay timeout of %dms", cmd->rsc_id, cmd->action, cmd->start_delay); return TRUE; } rsc->pending_ops = g_list_remove_link(rsc->pending_ops, first); g_list_free_1(first); #ifdef PCMK__TIME_USE_CGT get_current_time(&(cmd->t_run), &(cmd->t_first_run)); #endif cmd->epoch_last_run = time(NULL); } if (!cmd) { crm_trace("Nothing further to do for %s", rsc->rsc_id); return TRUE; } rsc->active = cmd; /* only one op at a time for a rsc */ if (cmd->interval_ms) { rsc->recurring_ops = g_list_append(rsc->recurring_ops, cmd); } log_execute(cmd); if (pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei)) { lrmd_rsc_execute_stonith(rsc, cmd); } else { lrmd_rsc_execute_service_lib(rsc, cmd); } return TRUE; } static gboolean lrmd_rsc_dispatch(gpointer user_data) { return lrmd_rsc_execute(user_data); } void free_rsc(gpointer data) { GList *gIter = NULL; lrmd_rsc_t *rsc = data; int is_stonith = pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei); gIter = rsc->pending_ops; while (gIter != NULL) { GList *next = gIter->next; lrmd_cmd_t *cmd = gIter->data; /* command was never executed */ cmd->result.execution_status = PCMK_EXEC_CANCELLED; cmd_finalize(cmd, NULL); gIter = next; } /* frees list, but not list elements. */ g_list_free(rsc->pending_ops); gIter = rsc->recurring_ops; while (gIter != NULL) { GList *next = gIter->next; lrmd_cmd_t *cmd = gIter->data; if (is_stonith) { cmd->result.execution_status = PCMK_EXEC_CANCELLED; /* If a stonith command is in-flight, just mark it as cancelled; * it is not safe to finalize/free the cmd until the stonith api * says it has either completed or timed out. */ if (rsc->active != cmd) { cmd_finalize(cmd, NULL); } } else { /* This command is already handed off to service library, * let service library cancel it and tell us via the callback * when it is cancelled. The rsc can be safely destroyed * even if we are waiting for the cancel result */ services_action_cancel(rsc->rsc_id, normalize_action_name(rsc, cmd->action), cmd->interval_ms); } gIter = next; } /* frees list, but not list elements. */ g_list_free(rsc->recurring_ops); free(rsc->rsc_id); free(rsc->class); free(rsc->provider); free(rsc->type); mainloop_destroy_trigger(rsc->work); free(rsc); } static int process_lrmd_signon(pcmk__client_t *client, xmlNode *request, int call_id, xmlNode **reply) { int rc = pcmk_ok; const char *is_ipc_provider = crm_element_value(request, F_LRMD_IS_IPC_PROVIDER); const char *protocol_version = crm_element_value(request, F_LRMD_PROTOCOL_VERSION); if (compare_version(protocol_version, LRMD_MIN_PROTOCOL_VERSION) < 0) { crm_err("Cluster API version must be greater than or equal to %s, not %s", LRMD_MIN_PROTOCOL_VERSION, protocol_version); rc = -EPROTO; } if (crm_is_true(is_ipc_provider)) { #ifdef PCMK__COMPILE_REMOTE if ((client->remote != NULL) && client->remote->tls_handshake_complete) { // This is a remote connection from a cluster node's controller ipc_proxy_add_provider(client); } else { rc = -EACCES; } #else rc = -EPROTONOSUPPORT; #endif } *reply = create_lrmd_reply(__func__, rc, call_id); crm_xml_add(*reply, F_LRMD_OPERATION, CRM_OP_REGISTER); crm_xml_add(*reply, F_LRMD_CLIENTID, client->id); crm_xml_add(*reply, F_LRMD_PROTOCOL_VERSION, LRMD_PROTOCOL_VERSION); return rc; } static int process_lrmd_rsc_register(pcmk__client_t *client, uint32_t id, xmlNode *request) { int rc = pcmk_ok; lrmd_rsc_t *rsc = build_rsc_from_xml(request); lrmd_rsc_t *dup = g_hash_table_lookup(rsc_list, rsc->rsc_id); if (dup && pcmk__str_eq(rsc->class, dup->class, pcmk__str_casei) && pcmk__str_eq(rsc->provider, dup->provider, pcmk__str_casei) && pcmk__str_eq(rsc->type, dup->type, pcmk__str_casei)) { crm_notice("Ignoring duplicate registration of '%s'", rsc->rsc_id); free_rsc(rsc); return rc; } g_hash_table_replace(rsc_list, rsc->rsc_id, rsc); crm_info("Cached agent information for '%s'", rsc->rsc_id); return rc; } static xmlNode * process_lrmd_get_rsc_info(xmlNode *request, int call_id) { int rc = pcmk_ok; xmlNode *rsc_xml = get_xpath_object("//" F_LRMD_RSC, request, LOG_ERR); const char *rsc_id = crm_element_value(rsc_xml, F_LRMD_RSC_ID); xmlNode *reply = NULL; lrmd_rsc_t *rsc = NULL; if (rsc_id == NULL) { rc = -ENODEV; } else { rsc = g_hash_table_lookup(rsc_list, rsc_id); if (rsc == NULL) { crm_info("Agent information for '%s' not in cache", rsc_id); rc = -ENODEV; } } reply = create_lrmd_reply(__func__, rc, call_id); if (rsc) { crm_xml_add(reply, F_LRMD_RSC_ID, rsc->rsc_id); crm_xml_add(reply, F_LRMD_CLASS, rsc->class); crm_xml_add(reply, F_LRMD_PROVIDER, rsc->provider); crm_xml_add(reply, F_LRMD_TYPE, rsc->type); } return reply; } static int process_lrmd_rsc_unregister(pcmk__client_t *client, uint32_t id, xmlNode *request) { int rc = pcmk_ok; lrmd_rsc_t *rsc = NULL; xmlNode *rsc_xml = get_xpath_object("//" F_LRMD_RSC, request, LOG_ERR); const char *rsc_id = crm_element_value(rsc_xml, F_LRMD_RSC_ID); if (!rsc_id) { return -ENODEV; } rsc = g_hash_table_lookup(rsc_list, rsc_id); if (rsc == NULL) { crm_info("Ignoring unregistration of resource '%s', which is not registered", rsc_id); return pcmk_ok; } if (rsc->active) { /* let the caller know there are still active ops on this rsc to watch for */ crm_trace("Operation (0x%p) still in progress for unregistered resource %s", rsc->active, rsc_id); rc = -EINPROGRESS; } g_hash_table_remove(rsc_list, rsc_id); return rc; } static int process_lrmd_rsc_exec(pcmk__client_t *client, uint32_t id, xmlNode *request) { lrmd_rsc_t *rsc = NULL; lrmd_cmd_t *cmd = NULL; xmlNode *rsc_xml = get_xpath_object("//" F_LRMD_RSC, request, LOG_ERR); const char *rsc_id = crm_element_value(rsc_xml, F_LRMD_RSC_ID); int call_id; if (!rsc_id) { return -EINVAL; } if (!(rsc = g_hash_table_lookup(rsc_list, rsc_id))) { crm_info("Resource '%s' not found (%d active resources)", rsc_id, g_hash_table_size(rsc_list)); return -ENODEV; } cmd = create_lrmd_cmd(request, client); call_id = cmd->call_id; /* Don't reference cmd after handing it off to be scheduled. * The cmd could get merged and freed. */ schedule_lrmd_cmd(rsc, cmd); return call_id; } static int cancel_op(const char *rsc_id, const char *action, guint interval_ms) { GList *gIter = NULL; lrmd_rsc_t *rsc = g_hash_table_lookup(rsc_list, rsc_id); /* How to cancel an action. * 1. Check pending ops list, if it hasn't been handed off * to the service library or stonith recurring list remove * it there and that will stop it. * 2. If it isn't in the pending ops list, then it's either a * recurring op in the stonith recurring list, or the service * library's recurring list. Stop it there * 3. If not found in any lists, then this operation has either * been executed already and is not a recurring operation, or * never existed. */ if (!rsc) { return -ENODEV; } for (gIter = rsc->pending_ops; gIter != NULL; gIter = gIter->next) { lrmd_cmd_t *cmd = gIter->data; if (action_matches(cmd, action, interval_ms)) { cmd->result.execution_status = PCMK_EXEC_CANCELLED; cmd_finalize(cmd, rsc); return pcmk_ok; } } if (pcmk__str_eq(rsc->class, PCMK_RESOURCE_CLASS_STONITH, pcmk__str_casei)) { /* The service library does not handle stonith operations. * We have to handle recurring stonith operations ourselves. */ for (gIter = rsc->recurring_ops; gIter != NULL; gIter = gIter->next) { lrmd_cmd_t *cmd = gIter->data; if (action_matches(cmd, action, interval_ms)) { cmd->result.execution_status = PCMK_EXEC_CANCELLED; if (rsc->active != cmd) { cmd_finalize(cmd, rsc); } return pcmk_ok; } } } else if (services_action_cancel(rsc_id, normalize_action_name(rsc, action), interval_ms) == TRUE) { /* The service library will tell the action_complete callback function * this action was cancelled, which will destroy the cmd and remove * it from the recurring_op list. Do not do that in this function * if the service library says it cancelled it. */ return pcmk_ok; } return -EOPNOTSUPP; } static void cancel_all_recurring(lrmd_rsc_t * rsc, const char *client_id) { GList *cmd_list = NULL; GList *cmd_iter = NULL; /* Notice a copy of each list is created when concat is called. * This prevents odd behavior from occurring when the cmd_list * is iterated through later on. It is possible the cancel_op * function may end up modifying the recurring_ops and pending_ops * lists. If we did not copy those lists, our cmd_list iteration * could get messed up.*/ if (rsc->recurring_ops) { cmd_list = g_list_concat(cmd_list, g_list_copy(rsc->recurring_ops)); } if (rsc->pending_ops) { cmd_list = g_list_concat(cmd_list, g_list_copy(rsc->pending_ops)); } if (!cmd_list) { return; } for (cmd_iter = cmd_list; cmd_iter; cmd_iter = cmd_iter->next) { lrmd_cmd_t *cmd = cmd_iter->data; if (cmd->interval_ms == 0) { continue; } if (client_id && !pcmk__str_eq(cmd->client_id, client_id, pcmk__str_casei)) { continue; } cancel_op(rsc->rsc_id, cmd->action, cmd->interval_ms); } /* frees only the copied list data, not the cmds */ g_list_free(cmd_list); } static int process_lrmd_rsc_cancel(pcmk__client_t *client, uint32_t id, xmlNode *request) { xmlNode *rsc_xml = get_xpath_object("//" F_LRMD_RSC, request, LOG_ERR); const char *rsc_id = crm_element_value(rsc_xml, F_LRMD_RSC_ID); const char *action = crm_element_value(rsc_xml, F_LRMD_RSC_ACTION); guint interval_ms = 0; crm_element_value_ms(rsc_xml, F_LRMD_RSC_INTERVAL, &interval_ms); if (!rsc_id || !action) { return -EINVAL; } return cancel_op(rsc_id, action, interval_ms); } static void add_recurring_op_xml(xmlNode *reply, lrmd_rsc_t *rsc) { xmlNode *rsc_xml = create_xml_node(reply, F_LRMD_RSC); crm_xml_add(rsc_xml, F_LRMD_RSC_ID, rsc->rsc_id); for (GList *item = rsc->recurring_ops; item != NULL; item = item->next) { lrmd_cmd_t *cmd = item->data; xmlNode *op_xml = create_xml_node(rsc_xml, T_LRMD_RSC_OP); crm_xml_add(op_xml, F_LRMD_RSC_ACTION, (cmd->real_action? cmd->real_action : cmd->action)); crm_xml_add_ms(op_xml, F_LRMD_RSC_INTERVAL, cmd->interval_ms); crm_xml_add_int(op_xml, F_LRMD_TIMEOUT, cmd->timeout_orig); } } static xmlNode * process_lrmd_get_recurring(xmlNode *request, int call_id) { int rc = pcmk_ok; const char *rsc_id = NULL; lrmd_rsc_t *rsc = NULL; xmlNode *reply = NULL; xmlNode *rsc_xml = NULL; // Resource ID is optional rsc_xml = first_named_child(request, F_LRMD_CALLDATA); if (rsc_xml) { rsc_xml = first_named_child(rsc_xml, F_LRMD_RSC); } if (rsc_xml) { rsc_id = crm_element_value(rsc_xml, F_LRMD_RSC_ID); } // If resource ID is specified, resource must exist if (rsc_id != NULL) { rsc = g_hash_table_lookup(rsc_list, rsc_id); if (rsc == NULL) { crm_info("Resource '%s' not found (%d active resources)", rsc_id, g_hash_table_size(rsc_list)); rc = -ENODEV; } } reply = create_lrmd_reply(__func__, rc, call_id); // If resource ID is not specified, check all resources if (rsc_id == NULL) { GHashTableIter iter; char *key = NULL; g_hash_table_iter_init(&iter, rsc_list); while (g_hash_table_iter_next(&iter, (gpointer *) &key, (gpointer *) &rsc)) { add_recurring_op_xml(reply, rsc); } } else if (rsc) { add_recurring_op_xml(reply, rsc); } return reply; } void process_lrmd_message(pcmk__client_t *client, uint32_t id, xmlNode *request) { int rc = pcmk_ok; int call_id = 0; const char *op = crm_element_value(request, F_LRMD_OPERATION); int do_reply = 0; int do_notify = 0; xmlNode *reply = NULL; /* Certain IPC commands may be done only by privileged users (i.e. root or * hacluster), because they would otherwise provide a means of bypassing * ACLs. */ bool allowed = pcmk_is_set(client->flags, pcmk__client_privileged); crm_trace("Processing %s operation from %s", op, client->id); crm_element_value_int(request, F_LRMD_CALLID, &call_id); if (pcmk__str_eq(op, CRM_OP_IPC_FWD, pcmk__str_none)) { #ifdef PCMK__COMPILE_REMOTE if (allowed) { ipc_proxy_forward_client(client, request); } else { rc = -EACCES; } #else rc = -EPROTONOSUPPORT; #endif do_reply = 1; } else if (pcmk__str_eq(op, CRM_OP_REGISTER, pcmk__str_none)) { rc = process_lrmd_signon(client, request, call_id, &reply); do_reply = 1; } else if (pcmk__str_eq(op, LRMD_OP_RSC_REG, pcmk__str_none)) { if (allowed) { rc = process_lrmd_rsc_register(client, id, request); do_notify = 1; } else { rc = -EACCES; } do_reply = 1; } else if (pcmk__str_eq(op, LRMD_OP_RSC_INFO, pcmk__str_none)) { if (allowed) { reply = process_lrmd_get_rsc_info(request, call_id); } else { rc = -EACCES; } do_reply = 1; } else if (pcmk__str_eq(op, LRMD_OP_RSC_UNREG, pcmk__str_none)) { if (allowed) { rc = process_lrmd_rsc_unregister(client, id, request); /* don't notify anyone about failed un-registers */ if (rc == pcmk_ok || rc == -EINPROGRESS) { do_notify = 1; } } else { rc = -EACCES; } do_reply = 1; } else if (pcmk__str_eq(op, LRMD_OP_RSC_EXEC, pcmk__str_none)) { if (allowed) { rc = process_lrmd_rsc_exec(client, id, request); } else { rc = -EACCES; } do_reply = 1; } else if (pcmk__str_eq(op, LRMD_OP_RSC_CANCEL, pcmk__str_none)) { if (allowed) { rc = process_lrmd_rsc_cancel(client, id, request); } else { rc = -EACCES; } do_reply = 1; } else if (pcmk__str_eq(op, LRMD_OP_POKE, pcmk__str_none)) { do_notify = 1; do_reply = 1; } else if (pcmk__str_eq(op, LRMD_OP_CHECK, pcmk__str_none)) { if (allowed) { xmlNode *data = get_message_xml(request, F_LRMD_CALLDATA); CRM_LOG_ASSERT(data != NULL); pcmk__valid_sbd_timeout(crm_element_value(data, F_LRMD_WATCHDOG)); } else { rc = -EACCES; } } else if (pcmk__str_eq(op, LRMD_OP_ALERT_EXEC, pcmk__str_none)) { if (allowed) { rc = process_lrmd_alert_exec(client, id, request); } else { rc = -EACCES; } do_reply = 1; } else if (pcmk__str_eq(op, LRMD_OP_GET_RECURRING, pcmk__str_none)) { if (allowed) { reply = process_lrmd_get_recurring(request, call_id); } else { rc = -EACCES; } do_reply = 1; } else { rc = -EOPNOTSUPP; do_reply = 1; crm_err("Unknown IPC request '%s' from client %s", op, pcmk__client_name(client)); } if (rc == -EACCES) { crm_warn("Rejecting IPC request '%s' from unprivileged client %s", op, pcmk__client_name(client)); } crm_debug("Processed %s operation from %s: rc=%d, reply=%d, notify=%d", op, client->id, rc, do_reply, do_notify); if (do_reply) { int send_rc = pcmk_rc_ok; if (reply == NULL) { reply = create_lrmd_reply(__func__, rc, call_id); } send_rc = lrmd_server_send_reply(client, id, reply); free_xml(reply); if (send_rc != pcmk_rc_ok) { crm_warn("Reply to client %s failed: %s " CRM_XS " rc=%d", pcmk__client_name(client), pcmk_rc_str(send_rc), send_rc); } } if (do_notify) { send_generic_notify(rc, request); } } diff --git a/lib/services/services_linux.c b/lib/services/services_linux.c index 2d73301e1b..84d26a766b 100644 --- a/lib/services/services_linux.c +++ b/lib/services/services_linux.c @@ -1,1381 +1,1381 @@ /* * Copyright 2010-2021 the Pacemaker project contributors * * The version control history for this file may have further details. * * This source code is licensed under the GNU Lesser General Public License * version 2.1 or later (LGPLv2.1+) WITHOUT ANY WARRANTY. */ #include #ifndef _GNU_SOURCE # define _GNU_SOURCE #endif #include #include #include #include #include #include #include #include #include #include #include "crm/crm.h" #include "crm/common/mainloop.h" #include "crm/services.h" #include "services_private.h" static void close_pipe(int fildes[]); /* We have two alternative ways of handling SIGCHLD when synchronously waiting * for spawned processes to complete. Both rely on polling a file descriptor to * discover SIGCHLD events. * * If sys/signalfd.h is available (e.g. on Linux), we call signalfd() to * generate the file descriptor. Otherwise, we use the "self-pipe trick" * (opening a pipe and writing a byte to it when SIGCHLD is received). */ #ifdef HAVE_SYS_SIGNALFD_H // signalfd() implementation #include // Everything needed to manage SIGCHLD handling struct sigchld_data_s { sigset_t mask; // Signals to block now (including SIGCHLD) sigset_t old_mask; // Previous set of blocked signals }; // Initialize SIGCHLD data and prepare for use static bool sigchld_setup(struct sigchld_data_s *data) { sigemptyset(&(data->mask)); sigaddset(&(data->mask), SIGCHLD); sigemptyset(&(data->old_mask)); // Block SIGCHLD (saving previous set of blocked signals to restore later) if (sigprocmask(SIG_BLOCK, &(data->mask), &(data->old_mask)) < 0) { crm_err("Wait for child process completion failed: %s " CRM_XS " source=sigprocmask", pcmk_strerror(errno)); return false; } return true; } // Get a file descriptor suitable for polling for SIGCHLD events static int sigchld_open(struct sigchld_data_s *data) { int fd; CRM_CHECK(data != NULL, return -1); fd = signalfd(-1, &(data->mask), SFD_NONBLOCK); if (fd < 0) { crm_err("Wait for child process completion failed: %s " CRM_XS " source=signalfd", pcmk_strerror(errno)); } return fd; } // Close a file descriptor returned by sigchld_open() static void sigchld_close(int fd) { if (fd > 0) { close(fd); } } // Return true if SIGCHLD was received from polled fd static bool sigchld_received(int fd) { struct signalfd_siginfo fdsi; ssize_t s; if (fd < 0) { return false; } s = read(fd, &fdsi, sizeof(struct signalfd_siginfo)); if (s != sizeof(struct signalfd_siginfo)) { crm_err("Wait for child process completion failed: %s " CRM_XS " source=read", pcmk_strerror(errno)); } else if (fdsi.ssi_signo == SIGCHLD) { return true; } return false; } // Do anything needed after done waiting for SIGCHLD static void sigchld_cleanup(struct sigchld_data_s *data) { // Restore the original set of blocked signals if ((sigismember(&(data->old_mask), SIGCHLD) == 0) && (sigprocmask(SIG_UNBLOCK, &(data->mask), NULL) < 0)) { crm_warn("Could not clean up after child process completion: %s", pcmk_strerror(errno)); } } #else // HAVE_SYS_SIGNALFD_H not defined // Self-pipe implementation (see above for function descriptions) struct sigchld_data_s { int pipe_fd[2]; // Pipe file descriptors struct sigaction sa; // Signal handling info (with SIGCHLD) struct sigaction old_sa; // Previous signal handling info }; // We need a global to use in the signal handler volatile struct sigchld_data_s *last_sigchld_data = NULL; static void sigchld_handler() { // We received a SIGCHLD, so trigger pipe polling if ((last_sigchld_data != NULL) && (last_sigchld_data->pipe_fd[1] >= 0) && (write(last_sigchld_data->pipe_fd[1], "", 1) == -1)) { crm_err("Wait for child process completion failed: %s " CRM_XS " source=write", pcmk_strerror(errno)); } } static bool sigchld_setup(struct sigchld_data_s *data) { int rc; data->pipe_fd[0] = data->pipe_fd[1] = -1; if (pipe(data->pipe_fd) == -1) { crm_err("Wait for child process completion failed: %s " CRM_XS " source=pipe", pcmk_strerror(errno)); return false; } rc = pcmk__set_nonblocking(data->pipe_fd[0]); if (rc != pcmk_rc_ok) { crm_warn("Could not set pipe input non-blocking: %s " CRM_XS " rc=%d", pcmk_rc_str(rc), rc); } rc = pcmk__set_nonblocking(data->pipe_fd[1]); if (rc != pcmk_rc_ok) { crm_warn("Could not set pipe output non-blocking: %s " CRM_XS " rc=%d", pcmk_rc_str(rc), rc); } // Set SIGCHLD handler data->sa.sa_handler = sigchld_handler; data->sa.sa_flags = 0; sigemptyset(&(data->sa.sa_mask)); if (sigaction(SIGCHLD, &(data->sa), &(data->old_sa)) < 0) { crm_err("Wait for child process completion failed: %s " CRM_XS " source=sigaction", pcmk_strerror(errno)); } // Remember data for use in signal handler last_sigchld_data = data; return true; } static int sigchld_open(struct sigchld_data_s *data) { CRM_CHECK(data != NULL, return -1); return data->pipe_fd[0]; } static void sigchld_close(int fd) { // Pipe will be closed in sigchld_cleanup() return; } static bool sigchld_received(int fd) { char ch; if (fd < 0) { return false; } // Clear out the self-pipe while (read(fd, &ch, 1) == 1) /*omit*/; return true; } static void sigchld_cleanup(struct sigchld_data_s *data) { // Restore the previous SIGCHLD handler if (sigaction(SIGCHLD, &(data->old_sa), NULL) < 0) { crm_warn("Could not clean up after child process completion: %s", pcmk_strerror(errno)); } close_pipe(data->pipe_fd); } #endif /*! * \internal * \brief Close the two file descriptors of a pipe * * \param[in] fildes Array of file descriptors opened by pipe() */ static void close_pipe(int fildes[]) { if (fildes[0] >= 0) { close(fildes[0]); fildes[0] = -1; } if (fildes[1] >= 0) { close(fildes[1]); fildes[1] = -1; } } static gboolean svc_read_output(int fd, svc_action_t * op, bool is_stderr) { char *data = NULL; int rc = 0, len = 0; char buf[500]; static const size_t buf_read_len = sizeof(buf) - 1; if (fd < 0) { crm_trace("No fd for %s", op->id); return FALSE; } if (is_stderr && op->stderr_data) { len = strlen(op->stderr_data); data = op->stderr_data; crm_trace("Reading %s stderr into offset %d", op->id, len); } else if (is_stderr == FALSE && op->stdout_data) { len = strlen(op->stdout_data); data = op->stdout_data; crm_trace("Reading %s stdout into offset %d", op->id, len); } else { crm_trace("Reading %s %s into offset %d", op->id, is_stderr?"stderr":"stdout", len); } do { rc = read(fd, buf, buf_read_len); if (rc > 0) { buf[rc] = 0; crm_trace("Got %d chars: %.80s", rc, buf); data = pcmk__realloc(data, len + rc + 1); len += sprintf(data + len, "%s", buf); } else if (errno != EINTR) { /* error or EOF * Cleanup happens in pipe_done() */ rc = FALSE; break; } } while (rc == buf_read_len || rc < 0); if (is_stderr) { op->stderr_data = data; } else { op->stdout_data = data; } return rc; } static int dispatch_stdout(gpointer userdata) { svc_action_t *op = (svc_action_t *) userdata; return svc_read_output(op->opaque->stdout_fd, op, FALSE); } static int dispatch_stderr(gpointer userdata) { svc_action_t *op = (svc_action_t *) userdata; return svc_read_output(op->opaque->stderr_fd, op, TRUE); } static void pipe_out_done(gpointer user_data) { svc_action_t *op = (svc_action_t *) user_data; crm_trace("%p", op); op->opaque->stdout_gsource = NULL; if (op->opaque->stdout_fd > STDOUT_FILENO) { close(op->opaque->stdout_fd); } op->opaque->stdout_fd = -1; } static void pipe_err_done(gpointer user_data) { svc_action_t *op = (svc_action_t *) user_data; op->opaque->stderr_gsource = NULL; if (op->opaque->stderr_fd > STDERR_FILENO) { close(op->opaque->stderr_fd); } op->opaque->stderr_fd = -1; } static struct mainloop_fd_callbacks stdout_callbacks = { .dispatch = dispatch_stdout, .destroy = pipe_out_done, }; static struct mainloop_fd_callbacks stderr_callbacks = { .dispatch = dispatch_stderr, .destroy = pipe_err_done, }; static void set_ocf_env(const char *key, const char *value, gpointer user_data) { if (setenv(key, value, 1) != 0) { crm_perror(LOG_ERR, "setenv failed for key:%s and value:%s", key, value); } } static void set_ocf_env_with_prefix(gpointer key, gpointer value, gpointer user_data) { char buffer[500]; snprintf(buffer, sizeof(buffer), strcmp(key, "OCF_CHECK_LEVEL") != 0 ? "OCF_RESKEY_%s" : "%s", (char *)key); set_ocf_env(buffer, value, user_data); } static void set_alert_env(gpointer key, gpointer value, gpointer user_data) { int rc; if (value != NULL) { rc = setenv(key, value, 1); } else { rc = unsetenv(key); } if (rc < 0) { crm_perror(LOG_ERR, "setenv %s=%s", (char*)key, (value? (char*)value : "")); } else { crm_trace("setenv %s=%s", (char*)key, (value? (char*)value : "")); } } /*! * \internal * \brief Add environment variables suitable for an action * * \param[in] op Action to use */ static void add_action_env_vars(const svc_action_t *op) { void (*env_setter)(gpointer, gpointer, gpointer) = NULL; if (op->agent == NULL) { env_setter = set_alert_env; /* we deal with alert handler */ } else if (pcmk__str_eq(op->standard, PCMK_RESOURCE_CLASS_OCF, pcmk__str_casei)) { env_setter = set_ocf_env_with_prefix; } if (env_setter != NULL && op->params != NULL) { g_hash_table_foreach(op->params, env_setter, NULL); } if (env_setter == NULL || env_setter == set_alert_env) { return; } set_ocf_env("OCF_RA_VERSION_MAJOR", PCMK_OCF_MAJOR_VERSION, NULL); set_ocf_env("OCF_RA_VERSION_MINOR", PCMK_OCF_MINOR_VERSION, NULL); set_ocf_env("OCF_ROOT", OCF_ROOT_DIR, NULL); set_ocf_env("OCF_EXIT_REASON_PREFIX", PCMK_OCF_REASON_PREFIX, NULL); if (op->rsc) { set_ocf_env("OCF_RESOURCE_INSTANCE", op->rsc, NULL); } if (op->agent != NULL) { set_ocf_env("OCF_RESOURCE_TYPE", op->agent, NULL); } /* Notes: this is not added to specification yet. Sept 10,2004 */ if (op->provider != NULL) { set_ocf_env("OCF_RESOURCE_PROVIDER", op->provider, NULL); } } static void pipe_in_single_parameter(gpointer key, gpointer value, gpointer user_data) { svc_action_t *op = user_data; char *buffer = crm_strdup_printf("%s=%s\n", (char *)key, (char *) value); int ret, total = 0, len = strlen(buffer); do { errno = 0; ret = write(op->opaque->stdin_fd, buffer + total, len - total); if (ret > 0) { total += ret; } } while ((errno == EINTR) && (total < len)); free(buffer); } /*! * \internal * \brief Pipe parameters in via stdin for action * * \param[in] op Action to use */ static void pipe_in_action_stdin_parameters(const svc_action_t *op) { crm_debug("sending args"); if (op->params) { g_hash_table_foreach(op->params, pipe_in_single_parameter, (gpointer) op); } } gboolean recurring_action_timer(gpointer data) { svc_action_t *op = data; crm_debug("Scheduling another invocation of %s", op->id); /* Clean out the old result */ free(op->stdout_data); op->stdout_data = NULL; free(op->stderr_data); op->stderr_data = NULL; op->opaque->repeat_timer = 0; services_action_async(op, NULL); return FALSE; } /*! * \internal * \brief Finalize handling of an asynchronous operation * * Given a completed asynchronous operation, cancel or reschedule it as * appropriate if recurring, call its callback if registered, stop tracking it, * and clean it up. * * \param[in,out] op Operation to finalize * * \return Standard Pacemaker return code * \retval EINVAL Caller supplied NULL or invalid \p op * \retval EBUSY Uncanceled recurring action has only been cleaned up * \retval pcmk_rc_ok Action has been freed * * \note If the return value is not pcmk_rc_ok, the caller is responsible for * freeing the action. */ int services__finalize_async_op(svc_action_t *op) { CRM_CHECK((op != NULL) && !(op->synchronous), return EINVAL); if (op->interval_ms != 0) { // Recurring operations must be either cancelled or rescheduled if (op->cancel) { op->status = PCMK_EXEC_CANCELLED; cancel_recurring_action(op); } else { op->opaque->repeat_timer = g_timeout_add(op->interval_ms, recurring_action_timer, (void *) op); } } if (op->opaque->callback != NULL) { op->opaque->callback(op); } // Stop tracking the operation (as in-flight or blocked) op->pid = 0; services_untrack_op(op); if ((op->interval_ms != 0) && !(op->cancel)) { // Do not free recurring actions (they will get freed when cancelled) services_action_cleanup(op); return EBUSY; } services_action_free(op); return pcmk_rc_ok; } static void close_op_input(svc_action_t *op) { if (op->opaque->stdin_fd >= 0) { close(op->opaque->stdin_fd); } } static void finish_op_output(svc_action_t *op, bool is_stderr) { mainloop_io_t **source; int fd; if (is_stderr) { source = &(op->opaque->stderr_gsource); fd = op->opaque->stderr_fd; } else { source = &(op->opaque->stdout_gsource); fd = op->opaque->stdout_fd; } if (op->synchronous || *source) { crm_trace("Finish reading %s[%d] %s", op->id, op->pid, (is_stderr? "stdout" : "stderr")); svc_read_output(fd, op, is_stderr); if (op->synchronous) { close(fd); } else { mainloop_del_fd(*source); *source = NULL; } } } // Log an operation's stdout and stderr static void log_op_output(svc_action_t *op) { char *prefix = crm_strdup_printf("%s[%d] error output", op->id, op->pid); crm_log_output(LOG_NOTICE, prefix, op->stderr_data); strcpy(prefix + strlen(prefix) - strlen("error output"), "output"); crm_log_output(LOG_DEBUG, prefix, op->stdout_data); free(prefix); } /*! * \internal * \brief Process the completion of an asynchronous child process * * \param[in] p Child process that completed * \param[in] pid Process ID of child * \param[in] core (unused) * \param[in] signo Signal that interrupted child, if any * \param[in] exitcode Exit status of child process */ static void async_action_complete(mainloop_child_t *p, pid_t pid, int core, int signo, int exitcode) { svc_action_t *op = mainloop_child_userdata(p); mainloop_clear_child_userdata(p); CRM_CHECK(op->pid == pid, op->rc = services__generic_error(op); op->status = PCMK_EXEC_ERROR; return); /* Depending on the priority the mainloop gives the stdout and stderr * file descriptors, this function could be called before everything has * been read from them, so force a final read now. */ finish_op_output(op, true); finish_op_output(op, false); close_op_input(op); if (signo == 0) { crm_debug("%s[%d] exited with status %d", op->id, op->pid, exitcode); op->status = PCMK_EXEC_DONE; op->rc = exitcode; } else if (mainloop_child_timeout(p)) { crm_warn("%s[%d] timed out after %dms", op->id, op->pid, op->timeout); op->status = PCMK_EXEC_TIMEOUT; - op->rc = PCMK_OCF_TIMEOUT; + op->rc = services__generic_error(op); } else if (op->cancel) { /* If an in-flight recurring operation was killed because it was * cancelled, don't treat that as a failure. */ crm_info("%s[%d] terminated with signal: %s " CRM_XS " (%d)", op->id, op->pid, strsignal(signo), signo); op->status = PCMK_EXEC_CANCELLED; op->rc = PCMK_OCF_OK; } else { crm_warn("%s[%d] terminated with signal: %s " CRM_XS " (%d)", op->id, op->pid, strsignal(signo), signo); op->status = PCMK_EXEC_ERROR; op->rc = PCMK_OCF_UNKNOWN_ERROR; } log_op_output(op); services__finalize_async_op(op); } /*! * \internal * \brief Return agent standard's exit status for "generic error" * * When returning an internal error for an action, a value that is appropriate * to the action's agent standard must be used. This function returns a value * appropriate for errors in general. * * \param[in] op Action that error is for * * \return Exit status appropriate to agent standard * \note Actions without a standard will get PCMK_OCF_UNKNOWN_ERROR. */ int services__generic_error(svc_action_t *op) { if ((op == NULL) || (op->standard == NULL)) { return PCMK_OCF_UNKNOWN_ERROR; } if (pcmk__str_eq(op->standard, PCMK_RESOURCE_CLASS_LSB, pcmk__str_casei) && pcmk__str_eq(op->action, "status", pcmk__str_casei)) { return PCMK_LSB_STATUS_UNKNOWN; } #if SUPPORT_NAGIOS if (pcmk__str_eq(op->standard, PCMK_RESOURCE_CLASS_NAGIOS, pcmk__str_casei)) { return NAGIOS_STATE_UNKNOWN; } #endif return PCMK_OCF_UNKNOWN_ERROR; } /*! * \internal * \brief Return agent standard's exit status for "not installed" * * When returning an internal error for an action, a value that is appropriate * to the action's agent standard must be used. This function returns a value * appropriate for "not installed" errors. * * \param[in] op Action that error is for * * \return Exit status appropriate to agent standard * \note Actions without a standard will get PCMK_OCF_UNKNOWN_ERROR. */ int services__not_installed_error(svc_action_t *op) { if ((op == NULL) || (op->standard == NULL)) { return PCMK_OCF_UNKNOWN_ERROR; } if (pcmk__str_eq(op->standard, PCMK_RESOURCE_CLASS_LSB, pcmk__str_casei) && pcmk__str_eq(op->action, "status", pcmk__str_casei)) { return PCMK_LSB_STATUS_NOT_INSTALLED; } #if SUPPORT_NAGIOS if (pcmk__str_eq(op->standard, PCMK_RESOURCE_CLASS_NAGIOS, pcmk__str_casei)) { return NAGIOS_NOT_INSTALLED; } #endif return PCMK_OCF_NOT_INSTALLED; } /*! * \internal * \brief Return agent standard's exit status for "insufficient privileges" * * When returning an internal error for an action, a value that is appropriate * to the action's agent standard must be used. This function returns a value * appropriate for "insufficient privileges" errors. * * \param[in] op Action that error is for * * \return Exit status appropriate to agent standard * \note Actions without a standard will get PCMK_OCF_UNKNOWN_ERROR. */ int services__authorization_error(svc_action_t *op) { if ((op == NULL) || (op->standard == NULL)) { return PCMK_OCF_UNKNOWN_ERROR; } if (pcmk__str_eq(op->standard, PCMK_RESOURCE_CLASS_LSB, pcmk__str_casei) && pcmk__str_eq(op->action, "status", pcmk__str_casei)) { return PCMK_LSB_STATUS_INSUFFICIENT_PRIV; } #if SUPPORT_NAGIOS if (pcmk__str_eq(op->standard, PCMK_RESOURCE_CLASS_NAGIOS, pcmk__str_casei)) { return NAGIOS_INSUFFICIENT_PRIV; } #endif return PCMK_OCF_INSUFFICIENT_PRIV; } /*! * \internal * \brief Set operation rc and status per errno from stat(), fork() or execvp() * * \param[in,out] op Operation to set rc and status for * \param[in] error Value of errno after system call * * \return void */ void services__handle_exec_error(svc_action_t * op, int error) { switch (error) { /* see execve(2), stat(2) and fork(2) */ case ENOENT: /* No such file or directory */ case EISDIR: /* Is a directory */ case ENOTDIR: /* Path component is not a directory */ case EINVAL: /* Invalid executable format */ case ENOEXEC: /* Invalid executable format */ op->rc = services__not_installed_error(op); op->status = PCMK_EXEC_NOT_INSTALLED; break; case EACCES: /* permission denied (various errors) */ case EPERM: /* permission denied (various errors) */ op->rc = services__authorization_error(op); op->status = PCMK_EXEC_ERROR; break; default: op->rc = services__generic_error(op); op->status = PCMK_EXEC_ERROR; } } static void action_launch_child(svc_action_t *op) { /* SIGPIPE is ignored (which is different from signal blocking) by the gnutls library. * Depending on the libqb version in use, libqb may set SIGPIPE to be ignored as well. * We do not want this to be inherited by the child process. By resetting this the signal * to the default behavior, we avoid some potential odd problems that occur during OCF * scripts when SIGPIPE is ignored by the environment. */ signal(SIGPIPE, SIG_DFL); #if defined(HAVE_SCHED_SETSCHEDULER) if (sched_getscheduler(0) != SCHED_OTHER) { struct sched_param sp; memset(&sp, 0, sizeof(sp)); sp.sched_priority = 0; if (sched_setscheduler(0, SCHED_OTHER, &sp) == -1) { crm_perror(LOG_ERR, "Could not reset scheduling policy to SCHED_OTHER for %s", op->id); } } #endif if (setpriority(PRIO_PROCESS, 0, 0) == -1) { crm_perror(LOG_ERR, "Could not reset process priority to 0 for %s", op->id); } /* Man: The call setpgrp() is equivalent to setpgid(0,0) * _and_ compiles on BSD variants too * need to investigate if it works the same too. */ setpgid(0, 0); pcmk__close_fds_in_child(false); #if SUPPORT_CIBSECRETS if (pcmk__substitute_secrets(op->rsc, op->params) != pcmk_rc_ok) { /* replacing secrets failed! */ if (pcmk__str_eq(op->action, "stop", pcmk__str_casei)) { /* don't fail on stop! */ crm_info("proceeding with the stop operation for %s", op->rsc); } else { crm_err("failed to get secrets for %s, " "considering resource not configured", op->rsc); _exit(PCMK_OCF_NOT_CONFIGURED); } } #endif add_action_env_vars(op); /* Become the desired user */ if (op->opaque->uid && (geteuid() == 0)) { // If requested, set effective group if (op->opaque->gid && (setgid(op->opaque->gid) < 0)) { crm_perror(LOG_ERR, "Could not set child group to %d", op->opaque->gid); _exit(PCMK_OCF_NOT_CONFIGURED); } // Erase supplementary group list // (We could do initgroups() if we kept a copy of the username) if (setgroups(0, NULL) < 0) { crm_perror(LOG_ERR, "Could not set child groups"); _exit(PCMK_OCF_NOT_CONFIGURED); } // Set effective user if (setuid(op->opaque->uid) < 0) { crm_perror(LOG_ERR, "setting user to %d", op->opaque->uid); _exit(PCMK_OCF_NOT_CONFIGURED); } } /* execute the RA */ execvp(op->opaque->exec, op->opaque->args); /* Most cases should have been already handled by stat() */ services__handle_exec_error(op, errno); _exit(op->rc); } /*! * \internal * \brief Wait for synchronous action to complete, and set its result * * \param[in] op Action to wait for * \param[in] data Child signal data */ static void wait_for_sync_result(svc_action_t *op, struct sigchld_data_s *data) { int status = 0; int timeout = op->timeout; time_t start = time(NULL); struct pollfd fds[3]; int wait_rc = 0; fds[0].fd = op->opaque->stdout_fd; fds[0].events = POLLIN; fds[0].revents = 0; fds[1].fd = op->opaque->stderr_fd; fds[1].events = POLLIN; fds[1].revents = 0; fds[2].fd = sigchld_open(data); fds[2].events = POLLIN; fds[2].revents = 0; crm_trace("Waiting for %s[%d]", op->id, op->pid); do { int poll_rc = poll(fds, 3, timeout); if (poll_rc > 0) { if (fds[0].revents & POLLIN) { svc_read_output(op->opaque->stdout_fd, op, FALSE); } if (fds[1].revents & POLLIN) { svc_read_output(op->opaque->stderr_fd, op, TRUE); } if ((fds[2].revents & POLLIN) && sigchld_received(fds[2].fd)) { wait_rc = waitpid(op->pid, &status, WNOHANG); if ((wait_rc > 0) || ((wait_rc < 0) && (errno == ECHILD))) { // Child process exited or doesn't exist break; } else if (wait_rc < 0) { crm_warn("Wait for completion of %s[%d] failed: %s " CRM_XS " source=waitpid", op->id, op->pid, pcmk_strerror(errno)); wait_rc = 0; // Act as if process is still running } } } else if (poll_rc == 0) { // Poll timed out with no descriptors ready timeout = 0; break; } else if ((poll_rc < 0) && (errno != EINTR)) { crm_err("Wait for completion of %s[%d] failed: %s " CRM_XS " source=poll", op->id, op->pid, pcmk_strerror(errno)); break; } timeout = op->timeout - (time(NULL) - start) * 1000; } while ((op->timeout < 0 || timeout > 0)); crm_trace("Stopped waiting for %s[%d]", op->id, op->pid); if (wait_rc <= 0) { if ((op->timeout > 0) && (timeout <= 0)) { op->rc = services__generic_error(op); op->status = PCMK_EXEC_TIMEOUT; crm_warn("%s[%d] timed out after %dms", op->id, op->pid, op->timeout); } else { op->rc = services__generic_error(op); op->status = PCMK_EXEC_ERROR; } /* If only child hasn't been successfully waited for, yet. This is to limit killing wrong target a bit more. */ if ((wait_rc == 0) && (waitpid(op->pid, &status, WNOHANG) == 0)) { if (kill(op->pid, SIGKILL)) { crm_warn("Could not kill rogue child %s[%d]: %s", op->id, op->pid, pcmk_strerror(errno)); } /* Safe to skip WNOHANG here as we sent non-ignorable signal. */ while ((waitpid(op->pid, &status, 0) == (pid_t) -1) && (errno == EINTR)) { /* keep waiting */; } } } else if (WIFEXITED(status)) { op->status = PCMK_EXEC_DONE; op->rc = WEXITSTATUS(status); crm_info("%s[%d] exited with status %d", op->id, op->pid, op->rc); } else if (WIFSIGNALED(status)) { int signo = WTERMSIG(status); op->rc = services__generic_error(op); op->status = PCMK_EXEC_ERROR; crm_err("%s[%d] terminated with signal: %s " CRM_XS " (%d)", op->id, op->pid, strsignal(signo), signo); #ifdef WCOREDUMP if (WCOREDUMP(status)) { crm_warn("%s[%d] dumped core", op->id, op->pid); } #endif } else { // Shouldn't be possible to get here op->rc = services__generic_error(op); op->status = PCMK_EXEC_ERROR; } finish_op_output(op, true); finish_op_output(op, false); close_op_input(op); sigchld_close(fds[2].fd); } /*! * \internal * \brief Execute an action whose standard uses executable files * * \param[in] op Action to execute * * \return Standard Pacemaker return value * \retval EBUSY Recurring operation could not be initiated * \retval pcmk_rc_error Synchronous action failed * \retval pcmk_rc_ok Synchronous action succeeded, or asynchronous action * should not be freed (because it already was or is * pending) * * \note If the return value for an asynchronous action is not pcmk_rc_ok, the * caller is responsible for freeing the action. */ int services__execute_file(svc_action_t *op) { int stdout_fd[2]; int stderr_fd[2]; int stdin_fd[2] = {-1, -1}; int rc; struct stat st; struct sigchld_data_s data; // Catch common failure conditions early if (stat(op->opaque->exec, &st) != 0) { rc = errno; crm_warn("Cannot execute '%s': %s " CRM_XS " stat rc=%d", op->opaque->exec, pcmk_strerror(rc), rc); services__handle_exec_error(op, rc); goto done; } if (pipe(stdout_fd) < 0) { rc = errno; crm_err("Cannot execute '%s': %s " CRM_XS " pipe(stdout) rc=%d", op->opaque->exec, pcmk_strerror(rc), rc); services__handle_exec_error(op, rc); goto done; } if (pipe(stderr_fd) < 0) { rc = errno; close_pipe(stdout_fd); crm_err("Cannot execute '%s': %s " CRM_XS " pipe(stderr) rc=%d", op->opaque->exec, pcmk_strerror(rc), rc); services__handle_exec_error(op, rc); goto done; } if (pcmk_is_set(pcmk_get_ra_caps(op->standard), pcmk_ra_cap_stdin)) { if (pipe(stdin_fd) < 0) { rc = errno; close_pipe(stdout_fd); close_pipe(stderr_fd); crm_err("Cannot execute '%s': %s " CRM_XS " pipe(stdin) rc=%d", op->opaque->exec, pcmk_strerror(rc), rc); services__handle_exec_error(op, rc); goto done; } } if (op->synchronous && !sigchld_setup(&data)) { close_pipe(stdin_fd); close_pipe(stdout_fd); close_pipe(stderr_fd); sigchld_cleanup(&data); op->rc = services__generic_error(op); op->status = PCMK_EXEC_ERROR; goto done; } op->pid = fork(); switch (op->pid) { case -1: rc = errno; close_pipe(stdin_fd); close_pipe(stdout_fd); close_pipe(stderr_fd); crm_err("Cannot execute '%s': %s " CRM_XS " fork rc=%d", op->opaque->exec, pcmk_strerror(rc), rc); services__handle_exec_error(op, rc); if (op->synchronous) { sigchld_cleanup(&data); } goto done; break; case 0: /* Child */ close(stdout_fd[0]); close(stderr_fd[0]); if (stdin_fd[1] >= 0) { close(stdin_fd[1]); } if (STDOUT_FILENO != stdout_fd[1]) { if (dup2(stdout_fd[1], STDOUT_FILENO) != STDOUT_FILENO) { crm_warn("Can't redirect output from '%s': %s " CRM_XS " errno=%d", op->opaque->exec, pcmk_strerror(errno), errno); } close(stdout_fd[1]); } if (STDERR_FILENO != stderr_fd[1]) { if (dup2(stderr_fd[1], STDERR_FILENO) != STDERR_FILENO) { crm_warn("Can't redirect error output from '%s': %s " CRM_XS " errno=%d", op->opaque->exec, pcmk_strerror(errno), errno); } close(stderr_fd[1]); } if ((stdin_fd[0] >= 0) && (STDIN_FILENO != stdin_fd[0])) { if (dup2(stdin_fd[0], STDIN_FILENO) != STDIN_FILENO) { crm_warn("Can't redirect input to '%s': %s " CRM_XS " errno=%d", op->opaque->exec, pcmk_strerror(errno), errno); } close(stdin_fd[0]); } if (op->synchronous) { sigchld_cleanup(&data); } action_launch_child(op); CRM_ASSERT(0); /* action_launch_child is effectively noreturn */ } /* Only the parent reaches here */ close(stdout_fd[1]); close(stderr_fd[1]); if (stdin_fd[0] >= 0) { close(stdin_fd[0]); } op->opaque->stdout_fd = stdout_fd[0]; rc = pcmk__set_nonblocking(op->opaque->stdout_fd); if (rc != pcmk_rc_ok) { crm_warn("Could not set '%s' output non-blocking: %s " CRM_XS " rc=%d", op->opaque->exec, pcmk_rc_str(rc), rc); } op->opaque->stderr_fd = stderr_fd[0]; rc = pcmk__set_nonblocking(op->opaque->stderr_fd); if (rc != pcmk_rc_ok) { crm_warn("Could not set '%s' error output non-blocking: %s " CRM_XS " rc=%d", op->opaque->exec, pcmk_rc_str(rc), rc); } op->opaque->stdin_fd = stdin_fd[1]; if (op->opaque->stdin_fd >= 0) { // using buffer behind non-blocking-fd here - that could be improved // as long as no other standard uses stdin_fd assume stonith rc = pcmk__set_nonblocking(op->opaque->stdin_fd); if (rc != pcmk_rc_ok) { crm_warn("Could not set '%s' input non-blocking: %s " CRM_XS " fd=%d,rc=%d", op->opaque->exec, pcmk_rc_str(rc), op->opaque->stdin_fd, rc); } pipe_in_action_stdin_parameters(op); // as long as we are handling parameters directly in here just close close(op->opaque->stdin_fd); op->opaque->stdin_fd = -1; } // after fds are setup properly and before we plug anything into mainloop if (op->opaque->fork_callback) { op->opaque->fork_callback(op); } if (op->synchronous) { wait_for_sync_result(op, &data); sigchld_cleanup(&data); goto done; } crm_trace("Waiting async for '%s'[%d]", op->opaque->exec, op->pid); mainloop_child_add_with_flags(op->pid, op->timeout, op->id, op, pcmk_is_set(op->flags, SVC_ACTION_LEAVE_GROUP)? mainloop_leave_pid_group : 0, async_action_complete); op->opaque->stdout_gsource = mainloop_add_fd(op->id, G_PRIORITY_LOW, op->opaque->stdout_fd, op, &stdout_callbacks); op->opaque->stderr_gsource = mainloop_add_fd(op->id, G_PRIORITY_LOW, op->opaque->stderr_fd, op, &stderr_callbacks); services_add_inflight_op(op); return pcmk_rc_ok; done: if (op->synchronous) { return (op->rc == PCMK_OCF_OK)? pcmk_rc_ok : pcmk_rc_error; } else { return services__finalize_async_op(op); } } static GList * services_os_get_single_directory_list(const char *root, gboolean files, gboolean executable) { GList *list = NULL; struct dirent **namelist; int entries = 0, lpc = 0; char buffer[PATH_MAX]; entries = scandir(root, &namelist, NULL, alphasort); if (entries <= 0) { return list; } for (lpc = 0; lpc < entries; lpc++) { struct stat sb; if ('.' == namelist[lpc]->d_name[0]) { free(namelist[lpc]); continue; } snprintf(buffer, sizeof(buffer), "%s/%s", root, namelist[lpc]->d_name); if (stat(buffer, &sb)) { continue; } if (S_ISDIR(sb.st_mode)) { if (files) { free(namelist[lpc]); continue; } } else if (S_ISREG(sb.st_mode)) { if (files == FALSE) { free(namelist[lpc]); continue; } else if (executable && (sb.st_mode & S_IXUSR) == 0 && (sb.st_mode & S_IXGRP) == 0 && (sb.st_mode & S_IXOTH) == 0) { free(namelist[lpc]); continue; } } list = g_list_append(list, strdup(namelist[lpc]->d_name)); free(namelist[lpc]); } free(namelist); return list; } GList * services_os_get_directory_list(const char *root, gboolean files, gboolean executable) { GList *result = NULL; char *dirs = strdup(root); char *dir = NULL; if (pcmk__str_empty(dirs)) { free(dirs); return result; } for (dir = strtok(dirs, ":"); dir != NULL; dir = strtok(NULL, ":")) { GList *tmp = services_os_get_single_directory_list(dir, files, executable); if (tmp) { result = g_list_concat(result, tmp); } } free(dirs); return result; } static GList * services_os_get_directory_list_provider(const char *root, const char *provider, gboolean files, gboolean executable) { GList *result = NULL; char *dirs = strdup(root); char *dir = NULL; char buffer[PATH_MAX]; if (pcmk__str_empty(dirs)) { free(dirs); return result; } for (dir = strtok(dirs, ":"); dir != NULL; dir = strtok(NULL, ":")) { GList *tmp = NULL; sprintf(buffer, "%s/%s", dir, provider); tmp = services_os_get_single_directory_list(buffer, files, executable); if (tmp) { result = g_list_concat(result, tmp); } } free(dirs); return result; } GList * resources_os_list_ocf_providers(void) { return get_directory_list(OCF_RA_PATH, FALSE, TRUE); } GList * resources_os_list_ocf_agents(const char *provider) { GList *gIter = NULL; GList *result = NULL; GList *providers = NULL; if (provider) { return services_os_get_directory_list_provider(OCF_RA_PATH, provider, TRUE, TRUE); } providers = resources_os_list_ocf_providers(); for (gIter = providers; gIter != NULL; gIter = gIter->next) { GList *tmp1 = result; GList *tmp2 = resources_os_list_ocf_agents(gIter->data); if (tmp2) { result = g_list_concat(tmp1, tmp2); } } g_list_free_full(providers, free); return result; } gboolean services__ocf_agent_exists(const char *provider, const char *agent) { gboolean rc = FALSE; struct stat st; char *dirs = strdup(OCF_RA_PATH); char *dir = NULL; char *buf = NULL; if (provider == NULL || agent == NULL || pcmk__str_empty(dirs)) { free(dirs); return rc; } for (dir = strtok(dirs, ":"); dir != NULL; dir = strtok(NULL, ":")) { buf = crm_strdup_printf("%s/%s/%s", dir, provider, agent); if (stat(buf, &st) == 0) { free(buf); rc = TRUE; break; } free(buf); } free(dirs); return rc; }