diff --git a/cts/cts-exec.in b/cts/cts-exec.in index b944e84956..c451841df5 100644 --- a/cts/cts-exec.in +++ b/cts/cts-exec.in @@ -1,1230 +1,1230 @@ #!@PYTHON@ """ Regression tests for Pacemaker's pacemaker-execd """ # Pacemaker targets compatibility with Python 2.7 and 3.2+ from __future__ import print_function, unicode_literals, absolute_import, division __copyright__ = "Copyright 2012-2018 Andrew Beekhof " __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 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@" # 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: +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:%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 pipe_output(pipes, stdout=True, stderr=False): """ Wrapper to get text output from pipes regardless of Python version """ output = "" pipe_outputs = pipes.communicate() if sys.version_info < (3,): if stdout: output = output + pipe_outputs[0] if stderr: output = output + pipe_outputs[1] else: if stdout: output = output + pipe_outputs[0].decode(sys.stdout.encoding) if stderr: output = output + pipe_outputs[1].decode(sys.stderr.encoding) return output def output_from_command(command): """ Run a command, and return its standard output. """ test = subprocess.Popen(shlex.split(command), stdout=subprocess.PIPE) test.wait() return pipe_output(test).split("\n") def write_file(filename, contents, executable=False): """ Create a file. """ print("Installing %s ..." % (filename)) f = io.open(filename, "w+") f.write(contents) f.close() if executable: os.chmod(filename, EXECMODE) 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): self.name = name self.description = description self.cmds = [] if tls: self.daemon_location = "pacemaker-remoted" else: self.daemon_location = "pacemaker-execd" self.test_tool_location = "cts-exec-helper" self.verbose = verbose self.tls = tls 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 /tmp/pacemaker-execd-regression.log" % (self.daemon_location, additional_args))) time.sleep(1) 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('/tmp/pacemaker-execd-regression.log', 'rt', errors='replace') for line in logfile: print(line.strip().encode('utf-8', 'replace')) os.remove('/tmp/pacemaker-execd-regression.log') 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 = pipe_output(test) 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): self.tests = [] self.verbose = verbose self.tls = tls 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("Fatal error: python systemd bindings not found. Is package installed?", file=sys.stderr) sys.exit(CrmExit.ERROR) 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.tests.append(test) return test def setup_test_environment(self): """ Prepare the host before executing any tests """ 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_DIR@/pacemaker"): # @TODO remember which components were created and remove them os.makedirs("@OCF_RA_DIR@/pacemaker", 0o755) for agent in ["Dummy", "Stateful", "ping"]: agent_source = "%s/extra/resources/%s" % (BUILD_DIR, agent) agent_dest = "@OCF_RA_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:unknown 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:unknown 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:unknown 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("killall -9 -q 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:unknown 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['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] == "-R" or args[i] == "--pacemaker-remote": self.options['pacemaker-remote'] = 1 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(self): """ 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 [--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) tests = Tests(opts.options['verbose'], opts.options['pacemaker-remote']) tests.build_generic_tests() tests.build_multi_rsc_tests() tests.build_negative_tests() tests.build_custom_tests() tests.build_stress_tests() tests.setup_test_environment() print("Starting ...") if opts.options['list-tests']: tests.print_list() elif opts.options['show-usage']: opts.show_usage() elif 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/cts/cts-fencing.in b/cts/cts-fencing.in index f7319a4a5e..ce5331f34d 100644 --- a/cts/cts-fencing.in +++ b/cts/cts-fencing.in @@ -1,1442 +1,1442 @@ #!@PYTHON@ """ Regression tests for Pacemaker's fencer """ # Pacemaker targets compatibility with Python 2.7 and 3.2+ from __future__ import print_function, unicode_literals, absolute_import, division __copyright__ = "Copyright 2012-2018 Andrew Beekhof " __license__ = "GNU General Public License version 2 or later (GPLv2+) WITHOUT ANY WARRANTY" import io import os import sys import subprocess import shlex import time # Where to find test binaries # Prefer the source tree if available BUILD_DIR = "@abs_top_builddir@" TEST_DIR = sys.path[0] # These values must be kept in sync with include/crm/crm.h -class CrmExit: +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-fencing.in" % TEST_DIR): print("Running tests from the source tree: %s (%s)" % (BUILD_DIR, TEST_DIR)) # For pacemaker-fenced and cts-fence-helper new_path = "%s/daemons/fenced:%s" % (BUILD_DIR, new_path) new_path = "%s/tools:%s" % (BUILD_DIR, new_path) # For stonith_admin new_path = "%s/cts:%s" % (BUILD_DIR, new_path) # For cts-support else: print("Running tests from the install tree: @CRM_DAEMON_DIR@ (not %s)" % TEST_DIR) # For pacemaker-fenced, cts-fence-helper, and cts-support new_path = "@CRM_DAEMON_DIR@:%s" % (new_path) print('Using PATH="{}"'.format(new_path)) os.environ['PATH'] = new_path def pipe_output(pipes, stdout=True, stderr=False): """ Wrapper to get text output from pipes regardless of Python version """ output = "" pipe_outputs = pipes.communicate() if sys.version_info < (3,): if stdout: output = output + pipe_outputs[0] if stderr: output = output + pipe_outputs[1] else: if stdout: output = output + pipe_outputs[0].decode(sys.stdout.encoding) if stderr: output = output + pipe_outputs[1].decode(sys.stderr.encoding) return output def output_from_command(command): """ Execute command and return its standard output """ test = subprocess.Popen(shlex.split(command), stdout=subprocess.PIPE) test.wait() return pipe_output(test).split("\n") def localname(): """ Return the uname of the local host """ our_uname = output_from_command("uname -n") if our_uname: our_uname = our_uname[0] else: our_uname = "localhost" return our_uname def killall(process): """ Kill all instances of a process """ cmd = shlex.split("killall -9 -q %s" % process) test = subprocess.Popen(cmd, stdout=subprocess.PIPE) test.wait() 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, exit_code): self.output = output def __str__(self): return repr(self.output) class OutputFoundError(TestError): """ Exception raised when command output contains unwanted string """ def __init__(self, exit_code): self.output = output def __str__(self): return repr(self.output) class Test(object): """ Executor for a single test """ def __init__(self, name, description, verbose=0, with_cpg=0): self.name = name self.description = description self.cmds = [] self.verbose = verbose self.result_txt = "" self.cmd_tool_output = "" self.result_exitcode = CrmExit.OK if with_cpg: self.stonith_options = "-c" self.enable_corosync = 1 else: self.stonith_options = "-s" self.enable_corosync = 0 self.stonith_process = None self.stonith_output = "" self.stonith_patterns = [] self.negative_stonith_patterns = [] 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 """ 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, } ) def start_environment(self): """ Prepare the host for executing a test """ # Make sure we are in full control killall("pacemakerd") killall("pacemaker-fenced") if self.verbose: self.stonith_options = self.stonith_options + " -V" print("Starting pacemaker-fenced with %s" % self.stonith_options) if os.path.exists("/tmp/stonith-regression.log"): os.remove('/tmp/stonith-regression.log') cmd = "pacemaker-fenced %s -l /tmp/stonith-regression.log" % self.stonith_options self.stonith_process = subprocess.Popen(shlex.split(cmd)) time.sleep(1) def clean_environment(self): """ Clean up the host after executing a test """ if self.stonith_process: self.stonith_process.terminate() self.stonith_process.wait() self.stonith_output = "" self.stonith_process = None logfile = io.open('/tmp/stonith-regression.log', 'rt') for line in logfile.readlines(): self.stonith_output = self.stonith_output + line if self.verbose: print("Daemon Output Start") print(self.stonith_output) print("Daemon Output End") os.remove('/tmp/stonith-regression.log') def add_stonith_log_pattern(self, pattern): """ Add a log pattern to expect from this test """ self.stonith_patterns.append(pattern) def add_stonith_neg_log_pattern(self, pattern): """ Add a log pattern that should not occur with this test """ self.negative_stonith_patterns.append(pattern) def add_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_no_wait(self, cmd, args): """ Add a simple command to be executed (without waiting) as part of this test """ self.__new_cmd(cmd, args, CrmExit.OK, "", 1) def add_cmd_check_stdout(self, cmd, args, match, no_match=""): """ Add a simple command with expected output to be executed as part of this test """ self.__new_cmd(cmd, args, CrmExit.OK, match, 0, no_match) def add_expected_fail_cmd(self, cmd, args, exitcode=CrmExit.ERROR): """ Add a command to be executed as part of this test and expected to fail """ self.__new_cmd(cmd, 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, stderr=subprocess.PIPE) if args['kill']: if self.verbose: print("Also running: "+args['kill']) subprocess.Popen(shlex.split(args['kill'])) if args['no_wait'] == 0: test.wait() else: return CrmExit.OK output = pipe_output(test, stderr=True) if self.verbose: print(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 count_negative_matches(self, outline): """ Return 1 if a line matches patterns that shouldn't have occurred """ count = 0 for line in self.negative_stonith_patterns: if outline.count(line): count = 1 if self.verbose: print("This pattern should not have matched = '%s" % (line)) return count def match_stonith_patterns(self): """ Check test output for expected patterns """ negative_matches = 0 cur = 0 pats = self.stonith_patterns total_patterns = len(self.stonith_patterns) if len(self.stonith_patterns) == 0 and len(self.negative_stonith_patterns) == 0: return for line in self.stonith_output.split("\n"): negative_matches = negative_matches + self.count_negative_matches(line) if len(pats) == 0: continue cur = -1 for pat in pats: cur = cur + 1 if line.count(pats[cur]): del pats[cur] break if len(pats) > 0 or negative_matches: if self.verbose: for pat in pats: print("Pattern Not Matched = '%s'" % pat) msg = "FAILURE - '%s' failed. %d patterns out of %d not matched. %d negative matches." self.result_txt = msg % (self.name, len(pats), total_patterns, negative_matches) self.result_exitcode = CrmExit.ERROR 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 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("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("Step %d SUCCESS" % (i)) i = i + 1 self.clean_environment() if self.result_exitcode == CrmExit.OK: self.match_stonith_patterns() 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 fencing regression tests """ def __init__(self, verbose=0): self.tests = [] self.verbose = verbose self.autogen_corosync_cfg = 0 if not os.path.exists("/etc/corosync/corosync.conf"): self.autogen_corosync_cfg = 1 def new_test(self, name, description, with_cpg=0): """ Create a named test """ test = Test(name, description, self.verbose, with_cpg) self.tests.append(test) return test 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 start_corosync(self): """ Start the corosync process """ if self.verbose: print("Starting corosync") test = subprocess.Popen("corosync", stdout=subprocess.PIPE) test.wait() time.sleep(10) 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_cpg_only(self): """ Run all corosync-enabled tests """ for test in self.tests: if test.enable_corosync: test.run() def run_no_cpg(self): """ Run all standalone tests """ for test in self.tests: if not test.enable_corosync: 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)) def build_api_sanity_tests(self): """ Register tests to verify basic API usage """ verbose_arg = "" if self.verbose: verbose_arg = "-V" test = self.new_test("standalone_low_level_api_test", "Sanity test client api in standalone mode.") test.add_cmd("cts-fence-helper", "-t %s" % (verbose_arg)) test = self.new_test("cpg_low_level_api_test", "Sanity test client api using mainloop and cpg.", 1) test.add_cmd("cts-fence-helper", "-m %s" % (verbose_arg)) def build_custom_timeout_tests(self): """ Register tests to verify custom timeout usage """ # custom timeout without topology test = self.new_test("cpg_custom_timeout_1", "Verify per device timeouts work as expected without using topology.", 1) test.add_cmd('stonith_admin', '-R false1 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=node1 node2 node3"') test.add_cmd('stonith_admin', '-R true1 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node3" -o "pcmk_off_timeout=1"') test.add_cmd('stonith_admin', '-R false2 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=node3" -o "pcmk_off_timeout=4"') test.add_cmd("stonith_admin", "-F node3 -t 2") # timeout is 2+1+4 = 7 test.add_stonith_log_pattern("Total timeout set to 7") # custom timeout _WITH_ topology test = self.new_test("cpg_custom_timeout_2", "Verify per device timeouts work as expected _WITH_ topology.", 1) test.add_cmd('stonith_admin', '-R false1 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=node1 node2 node3"') test.add_cmd('stonith_admin', '-R true1 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node3" -o "pcmk_off_timeout=1"') test.add_cmd('stonith_admin', '-R false2 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=node3" -o "pcmk_off_timeout=4000"') test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1") test.add_cmd("stonith_admin", "-r node3 -i 2 -v true1") test.add_cmd("stonith_admin", "-r node3 -i 3 -v false2") test.add_cmd("stonith_admin", "-F node3 -t 2") # timeout is 2+1+4000 = 4003 test.add_stonith_log_pattern("Total timeout set to 4003") def build_fence_merge_tests(self): """ Register tests to verify when fence operations should be merged """ ### Simple test that overlapping fencing operations get merged test = self.new_test("cpg_custom_merge_single", "Verify overlapping identical fencing operations are merged, no fencing levels used.", 1) test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node3\" ") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3\"") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd("stonith_admin", "-F node3 -t 10") ### one merger will happen test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") ### the pattern below signifies that both the original and duplicate operation completed test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") ### Test that multiple mergers occur test = self.new_test("cpg_custom_merge_multiple", "Verify multiple overlapping identical fencing operations are merged", 1) test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"delay=2\" -o \"pcmk_host_list=node3\" ") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3\"") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd("stonith_admin", "-F node3 -t 10") ### 4 mergers should occur test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") ### the pattern below signifies that both the original and duplicate operation completed test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") ### Test that multiple mergers occur with topologies used test = self.new_test("cpg_custom_merge_with_topology", "Verify multiple overlapping identical fencing operations are merged with fencing levels.", 1) test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node3\" ") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1") test.add_cmd("stonith_admin", "-r node3 -i 1 -v false2") test.add_cmd("stonith_admin", "-r node3 -i 2 -v true1") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd_no_wait("stonith_admin", "-F node3 -t 10") test.add_cmd("stonith_admin", "-F node3 -t 10") ### 4 mergers should occur test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") test.add_stonith_log_pattern("Merging stonith action off for node node3 originating from client") ### the pattern below signifies that both the original and duplicate operation completed test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") test.add_stonith_log_pattern("Operation off of node3 by") def build_fence_no_merge_tests(self): """ Register tests to verify when fence operations should not be merged """ test = self.new_test("cpg_custom_no_merge", "Verify differing fencing operations are not merged", 1) test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3 node2\"") test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node3 node2\" ") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3 node2\"") test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1") test.add_cmd("stonith_admin", "-r node3 -i 1 -v false2") test.add_cmd("stonith_admin", "-r node3 -i 2 -v true1") test.add_cmd_no_wait("stonith_admin", "-F node2 -t 10") test.add_cmd("stonith_admin", "-F node3 -t 10") test.add_stonith_neg_log_pattern("Merging stonith action off for node node3 originating from client") def build_standalone_tests(self): """ Register a grab bag of tests that can be executed in standalone or corosync mode """ test_types = [ { "prefix" : "standalone", "use_cpg" : 0, }, { "prefix" : "cpg", "use_cpg" : 1, }, ] # test what happens when all devices timeout for test_type in test_types: test = self.new_test("%s_fence_multi_device_failure" % test_type["prefix"], "Verify that all devices timeout, a fencing failure is returned.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R false3 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") if test_type["use_cpg"] == 1: test.add_expected_fail_cmd("stonith_admin", "-F node3 -t 2", CrmExit.TIMEOUT) test.add_stonith_log_pattern("Total timeout set to 6") else: test.add_expected_fail_cmd("stonith_admin", "-F node3 -t 2", CrmExit.ERROR) test.add_stonith_log_pattern("for host 'node3' with device 'false1' returned: ") test.add_stonith_log_pattern("for host 'node3' with device 'false2' returned: ") test.add_stonith_log_pattern("for host 'node3' with device 'false3' returned: ") # test what happens when multiple devices can fence a node, but the first device fails. for test_type in test_types: test = self.new_test("%s_fence_device_failure_rollover" % test_type["prefix"], "Verify that when one fence device fails for a node, the others are tried.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-F node3 -t 2") if test_type["use_cpg"] == 1: test.add_stonith_log_pattern("Total timeout set to 6") # simple topology test for one device for test_type in test_types: if test_type["use_cpg"] == 0: continue test = self.new_test("%s_topology_simple" % test_type["prefix"], "Verify all fencing devices at a level are used.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-r node3 -i 1 -v true") test.add_cmd("stonith_admin", "-F node3 -t 2") test.add_stonith_log_pattern("Total timeout set to 2") test.add_stonith_log_pattern("for host 'node3' with device 'true' returned: 0") # add topology, delete topology, verify fencing still works for test_type in test_types: if test_type["use_cpg"] == 0: continue test = self.new_test("%s_topology_add_remove" % test_type["prefix"], "Verify fencing occurrs after all topology levels are removed", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-r node3 -i 1 -v true") test.add_cmd("stonith_admin", "-d node3 -i 1") test.add_cmd("stonith_admin", "-F node3 -t 2") test.add_stonith_log_pattern("Total timeout set to 2") test.add_stonith_log_pattern("for host 'node3' with device 'true' returned: 0") # test what happens when the first fencing level has multiple devices. for test_type in test_types: if test_type["use_cpg"] == 0: continue test = self.new_test("%s_topology_device_fails" % test_type["prefix"], "Verify if one device in a level fails, the other is tried.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R false -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-r node3 -i 1 -v false") test.add_cmd("stonith_admin", "-r node3 -i 2 -v true") test.add_cmd("stonith_admin", "-F node3 -t 20") test.add_stonith_log_pattern("Total timeout set to 40") test.add_stonith_log_pattern("for host 'node3' with device 'false' returned: -201") test.add_stonith_log_pattern("for host 'node3' with device 'true' returned: 0") # test what happens when the first fencing level fails. for test_type in test_types: if test_type["use_cpg"] == 0: continue test = self.new_test("%s_topology_multi_level_fails" % test_type["prefix"], "Verify if one level fails, the next leve is tried.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true3 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true4 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1") test.add_cmd("stonith_admin", "-r node3 -i 1 -v true1") test.add_cmd("stonith_admin", "-r node3 -i 2 -v true2") test.add_cmd("stonith_admin", "-r node3 -i 2 -v false2") test.add_cmd("stonith_admin", "-r node3 -i 3 -v true3") test.add_cmd("stonith_admin", "-r node3 -i 3 -v true4") test.add_cmd("stonith_admin", "-F node3 -t 3") test.add_stonith_log_pattern("Total timeout set to 18") test.add_stonith_log_pattern("for host 'node3' with device 'false1' returned: -201") test.add_stonith_log_pattern("for host 'node3' with device 'false2' returned: -201") test.add_stonith_log_pattern("for host 'node3' with device 'true3' returned: 0") test.add_stonith_log_pattern("for host 'node3' with device 'true4' returned: 0") # test what happens when the first fencing level had devices that no one has registered for test_type in test_types: if test_type["use_cpg"] == 0: continue test = self.new_test("%s_topology_missing_devices" % test_type["prefix"], "Verify topology can continue with missing devices.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true3 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true4 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1") test.add_cmd("stonith_admin", "-r node3 -i 1 -v true1") test.add_cmd("stonith_admin", "-r node3 -i 2 -v true2") test.add_cmd("stonith_admin", "-r node3 -i 2 -v false2") test.add_cmd("stonith_admin", "-r node3 -i 3 -v true3") test.add_cmd("stonith_admin", "-r node3 -i 3 -v true4") test.add_cmd("stonith_admin", "-F node3 -t 2") # Test what happens if multiple fencing levels are defined, and then the first one is removed. for test_type in test_types: if test_type["use_cpg"] == 0: continue test = self.new_test("%s_topology_level_removal" % test_type["prefix"], "Verify level removal works.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true3 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true4 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R false2 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-r node3 -i 1 -v false1") test.add_cmd("stonith_admin", "-r node3 -i 1 -v true1") test.add_cmd("stonith_admin", "-r node3 -i 2 -v true2") test.add_cmd("stonith_admin", "-r node3 -i 2 -v false2") test.add_cmd("stonith_admin", "-r node3 -i 3 -v true3") test.add_cmd("stonith_admin", "-r node3 -i 3 -v true4") # Now remove level 2, verify none of the devices in level two are hit. test.add_cmd("stonith_admin", "-d node3 -i 2") test.add_cmd("stonith_admin", "-F node3 -t 20") test.add_stonith_log_pattern("Total timeout set to 8") test.add_stonith_log_pattern("for host 'node3' with device 'false1' returned: -201") test.add_stonith_neg_log_pattern("for host 'node3' with device 'false2' returned: ") test.add_stonith_log_pattern("for host 'node3' with device 'true3' returned: 0") test.add_stonith_log_pattern("for host 'node3' with device 'true4' returned: 0") # Test targeting a topology level by node name pattern. for test_type in test_types: if test_type["use_cpg"] == 0: continue test = self.new_test("%s_topology_level_pattern" % test_type["prefix"], "Verify targeting topology by node name pattern works.", test_type["use_cpg"]) test.add_cmd("stonith_admin", """-R true -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node1 node2 node3" """) test.add_cmd("stonith_admin", """-r '@node.*' -i 1 -v true""") test.add_cmd("stonith_admin", "-F node3 -t 2") test.add_stonith_log_pattern("for host 'node3' with device 'true' returned: 0") # test allowing commas and semicolons as delimiters in pcmk_host_list for test_type in test_types: test = self.new_test("%s_host_list_delimiters" % test_type["prefix"], "Verify commas and semicolons can be used as pcmk_host_list delimiters", test_type["use_cpg"]) test.add_cmd("stonith_admin", """-R true1 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node1,node2,node3" """) test.add_cmd("stonith_admin", """-R true2 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=pcmk1;pcmk2;pcmk3" """) test.add_cmd("stonith_admin", "stonith_admin -F node2 -t 2") test.add_cmd("stonith_admin", "stonith_admin -F pcmk3 -t 2") test.add_stonith_log_pattern("for host 'node2' with device 'true1' returned: 0") test.add_stonith_log_pattern("for host 'pcmk3' with device 'true2' returned: 0") # test the stonith builds the correct list of devices that can fence a node. for test_type in test_types: test = self.new_test("%s_list_devices" % test_type["prefix"], "Verify list of devices that can fence a node is correct", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-R true3 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd_check_stdout("stonith_admin", "-l node1 -V", "true2", "true1") test.add_cmd_check_stdout("stonith_admin", "-l node1 -V", "true3", "true1") # simple test of device monitor for test_type in test_types: test = self.new_test("%s_monitor" % test_type["prefix"], "Verify device is reachable", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-R false1 -a fence_dummy -o \"mode=fail\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-Q true1") test.add_cmd("stonith_admin", "-Q false1") test.add_expected_fail_cmd("stonith_admin", "-Q true2", CrmExit.ERROR) # Verify monitor occurs for duration of timeout period on failure for test_type in test_types: test = self.new_test("%s_monitor_timeout" % test_type["prefix"], "Verify monitor uses duration of timeout period given.", test_type["use_cpg"]) test.add_cmd("stonith_admin", '-R true1 -a fence_dummy -o "mode=fail" -o "monitor_mode=fail" -o "pcmk_host_list=node3"') test.add_expected_fail_cmd("stonith_admin", "-Q true1 -t 5", CrmExit.ERROR) test.add_stonith_log_pattern("Attempt 2 to execute") # Verify monitor occurs for duration of timeout period on failure, but stops at max retries for test_type in test_types: test = self.new_test("%s_monitor_timeout_max_retries" % test_type["prefix"], "Verify monitor retries until max retry value or timeout is hit.", test_type["use_cpg"]) test.add_cmd("stonith_admin", '-R true1 -a fence_dummy -o "mode=fail" -o "monitor_mode=fail" -o "pcmk_host_list=node3"') test.add_expected_fail_cmd("stonith_admin", "-Q true1 -t 15", CrmExit.ERROR) test.add_stonith_log_pattern("Attempted to execute agent fence_dummy (list) the maximum number of times") # simple register test for test_type in test_types: test = self.new_test("%s_register" % test_type["prefix"], "Verify devices can be registered and un-registered", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-Q true1") test.add_cmd("stonith_admin", "-D true1") test.add_expected_fail_cmd("stonith_admin", "-Q true1", CrmExit.ERROR) # simple reboot test for test_type in test_types: test = self.new_test("%s_reboot" % test_type["prefix"], "Verify devices can be rebooted", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-B node3 -t 2") test.add_cmd("stonith_admin", "-D true1") test.add_expected_fail_cmd("stonith_admin", "-Q true1", CrmExit.ERROR) # test fencing history. for test_type in test_types: if test_type["use_cpg"] == 0: continue test = self.new_test("%s_fence_history" % test_type["prefix"], "Verify last fencing operation is returned.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node3\"") test.add_cmd("stonith_admin", "-F node3 -t 2 -V") test.add_cmd_check_stdout("stonith_admin", "-H node3", "was able to turn off node node3", "") # simple test of dynamic list query for test_type in test_types: test = self.new_test("%s_dynamic_list_query" % test_type["prefix"], "Verify dynamic list of fencing devices can be retrieved.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o mode=pass -o mock_dynamic_hosts=fake_port_1") test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o mode=pass -o mock_dynamic_hosts=fake_port_1") test.add_cmd("stonith_admin", "-R true3 -a fence_dummy -o mode=pass -o mock_dynamic_hosts=fake_port_1") test.add_cmd_check_stdout("stonith_admin", "-l fake_port_1", "3 devices found") # fence using dynamic list query for test_type in test_types: test = self.new_test("%s_fence_dynamic_list_query" % test_type["prefix"], "Verify dynamic list of fencing devices can be retrieved.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o mode=pass -o mock_dynamic_hosts=fake_port_1") test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o mode=pass -o mock_dynamic_hosts=fake_port_1") test.add_cmd("stonith_admin", "-R true3 -a fence_dummy -o mode=pass -o mock_dynamic_hosts=fake_port_1") test.add_cmd("stonith_admin", "-F fake_port_1 -t 5 -V") # simple test of query using status action for test_type in test_types: test = self.new_test("%s_status_query" % test_type["prefix"], "Verify dynamic list of fencing devices can be retrieved.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_check=status\"") test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_check=status\"") test.add_cmd("stonith_admin", "-R true3 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_check=status\"") test.add_cmd_check_stdout("stonith_admin", "-l fake_port_1", "3 devices found") # test what happens when no reboot action is advertised for test_type in test_types: test = self.new_test("%s_no_reboot_support" % test_type["prefix"], "Verify reboot action defaults to off when no reboot action is advertised by agent.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy_no_reboot -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-B node1 -t 5 -V") test.add_stonith_log_pattern("does not advertise support for 'reboot', performing 'off'") test.add_stonith_log_pattern("with device 'true1' returned: 0 (OK)") # make sure reboot is used when reboot action is advertised for test_type in test_types: test = self.new_test("%s_with_reboot_support" % test_type["prefix"], "Verify reboot action can be used when metadata advertises it.", test_type["use_cpg"]) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=node1 node2 node3\"") test.add_cmd("stonith_admin", "-B node1 -t 5 -V") test.add_stonith_neg_log_pattern("does not advertise support for 'reboot', performing 'off'") test.add_stonith_log_pattern("with device 'true1' returned: 0 (OK)") def build_nodeid_tests(self): """ Register tests that use a corosync node id """ our_uname = localname() ### verify nodeid is supplied when nodeid is in the metadata parameters test = self.new_test("cpg_supply_nodeid", "Verify nodeid is given when fence agent has nodeid as parameter", 1) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=%s\"" % (our_uname)) test.add_cmd("stonith_admin", "-F %s -t 3" % (our_uname)) test.add_stonith_log_pattern("For stonith action (off) for victim %s, adding nodeid" % (our_uname)) ### verify nodeid is _NOT_ supplied when nodeid is not in the metadata parameters test = self.new_test("cpg_do_not_supply_nodeid", "Verify nodeid is _NOT_ given when fence agent does not have nodeid as parameter", 1) # use a host name that won't be in corosync.conf test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=regr-test\"") test.add_cmd("stonith_admin", "-F regr-test -t 3") test.add_stonith_neg_log_pattern("For stonith action (off) for victim regr-test, adding nodeid") ### verify nodeid use doesn't explode standalone mode test = self.new_test("standalone_do_not_supply_nodeid", "Verify nodeid in metadata parameter list doesn't kill standalone mode", 0) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=%s\"" % (our_uname)) test.add_cmd("stonith_admin", "-F %s -t 3" % (our_uname)) test.add_stonith_neg_log_pattern("For stonith action (off) for victim %s, adding nodeid" % (our_uname)) def build_unfence_tests(self): """ Register tests that verify unfencing """ our_uname = localname() ### verify unfencing using automatic unfencing test = self.new_test("cpg_unfence_required_1", "Verify require unfencing on all devices when automatic=true in agent's metadata", 1) test.add_cmd('stonith_admin', '-R true1 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s"' % (our_uname)) test.add_cmd('stonith_admin', '-R true2 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s"' % (our_uname)) test.add_cmd("stonith_admin", "-U %s -t 3" % (our_uname)) # both devices should be executed test.add_stonith_log_pattern("with device 'true1' returned: 0 (OK)") test.add_stonith_log_pattern("with device 'true2' returned: 0 (OK)") ### verify unfencing using automatic unfencing fails if any of the required agents fail test = self.new_test("cpg_unfence_required_2", "Verify require unfencing on all devices when automatic=true in agent's metadata", 1) test.add_cmd('stonith_admin', '-R true1 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s"' % (our_uname)) test.add_cmd('stonith_admin', '-R true2 -a fence_dummy_auto_unfence -o "mode=fail" -o "pcmk_host_list=%s"' % (our_uname)) test.add_expected_fail_cmd("stonith_admin", "-U %s -t 6" % (our_uname), CrmExit.ERROR) ### verify unfencing using automatic devices with topology test = self.new_test("cpg_unfence_required_3", "Verify require unfencing on all devices even when at different topology levels", 1) test.add_cmd('stonith_admin', '-R true1 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd('stonith_admin', '-R true2 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 1 -v true1" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 2 -v true2" % (our_uname)) test.add_cmd("stonith_admin", "-U %s -t 3" % (our_uname)) test.add_stonith_log_pattern("with device 'true1' returned: 0 (OK)") test.add_stonith_log_pattern("with device 'true2' returned: 0 (OK)") ### verify unfencing using automatic devices with topology test = self.new_test("cpg_unfence_required_4", "Verify all required devices are executed even with topology levels fail.", 1) test.add_cmd('stonith_admin', '-R true1 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd('stonith_admin', '-R true2 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd('stonith_admin', '-R true3 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd('stonith_admin', '-R true4 -a fence_dummy_auto_unfence -o "mode=pass" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd('stonith_admin', '-R false1 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd('stonith_admin', '-R false2 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd('stonith_admin', '-R false3 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd('stonith_admin', '-R false4 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=%s node3"' % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 1 -v true1" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 1 -v false1" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 2 -v false2" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 2 -v true2" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 2 -v false3" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 2 -v true3" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 3 -v false4" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 4 -v true4" % (our_uname)) test.add_cmd("stonith_admin", "-U %s -t 3" % (our_uname)) test.add_stonith_log_pattern("with device 'true1' returned: 0 (OK)") test.add_stonith_log_pattern("with device 'true2' returned: 0 (OK)") test.add_stonith_log_pattern("with device 'true3' returned: 0 (OK)") test.add_stonith_log_pattern("with device 'true4' returned: 0 (OK)") def build_unfence_on_target_tests(self): """ Register tests that verify unfencing that runs on the target """ our_uname = localname() ### verify unfencing using on_target device test = self.new_test("cpg_unfence_on_target_1", "Verify unfencing with on_target = true", 1) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=%s\"" % (our_uname)) test.add_cmd("stonith_admin", "-U %s -t 3" % (our_uname)) test.add_stonith_log_pattern("(on) to be executed on the target node") ### verify failure of unfencing using on_target device test = self.new_test("cpg_unfence_on_target_2", "Verify failure unfencing with on_target = true", 1) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=%s node_fake_1234\"" % (our_uname)) test.add_expected_fail_cmd("stonith_admin", "-U node_fake_1234 -t 3", CrmExit.ERROR) test.add_stonith_log_pattern("(on) to be executed on the target node") ### verify unfencing using on_target device with topology test = self.new_test("cpg_unfence_on_target_3", "Verify unfencing with on_target = true using topology", 1) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=%s node3\"" % (our_uname)) test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=%s node3\"" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 1 -v true1" % (our_uname)) test.add_cmd("stonith_admin", "-r %s -i 2 -v true2" % (our_uname)) test.add_cmd("stonith_admin", "-U %s -t 3" % (our_uname)) test.add_stonith_log_pattern("(on) to be executed on the target node") ### verify unfencing using on_target device with topology fails when victim node doesn't exist test = self.new_test("cpg_unfence_on_target_4", "Verify unfencing failure with on_target = true using topology", 1) test.add_cmd("stonith_admin", "-R true1 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=%s node_fake\"" % (our_uname)) test.add_cmd("stonith_admin", "-R true2 -a fence_dummy -o \"mode=pass\" -o \"pcmk_host_list=%s node_fake\"" % (our_uname)) test.add_cmd("stonith_admin", "-r node_fake -i 1 -v true1") test.add_cmd("stonith_admin", "-r node_fake -i 2 -v true2") test.add_expected_fail_cmd("stonith_admin", "-U node_fake -t 3", CrmExit.ERROR) test.add_stonith_log_pattern("(on) to be executed on the target node") def build_remap_tests(self): """ Register tests that verify remapping of reboots to off-on """ test = self.new_test("cpg_remap_simple", "Verify sequential topology reboot is remapped to all-off-then-all-on", 1) test.add_cmd("stonith_admin", """-R true1 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node_fake" """ """-o "pcmk_off_timeout=1" -o "pcmk_reboot_timeout=10" """) test.add_cmd("stonith_admin", """-R true2 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node_fake" """ """-o "pcmk_off_timeout=2" -o "pcmk_reboot_timeout=20" """) test.add_cmd("stonith_admin", "-r node_fake -i 1 -v true1 -v true2") test.add_cmd("stonith_admin", "-B node_fake -t 5") test.add_stonith_log_pattern("Remapping multiple-device reboot of node_fake") # timeout should be sum of off timeouts (1+2=3), not reboot timeouts (10+20=30) test.add_stonith_log_pattern("Total timeout set to 3 for peer's fencing of node_fake") test.add_stonith_log_pattern("perform op 'node_fake off' with 'true1'") test.add_stonith_log_pattern("perform op 'node_fake off' with 'true2'") test.add_stonith_log_pattern("Remapped off of node_fake complete, remapping to on") # fence_dummy sets "on" as an on_target action test.add_stonith_log_pattern("Ignoring true1 'on' failure (no capable peers) for node_fake") test.add_stonith_log_pattern("Ignoring true2 'on' failure (no capable peers) for node_fake") test.add_stonith_log_pattern("Undoing remap of reboot of node_fake") test = self.new_test("cpg_remap_automatic", "Verify remapped topology reboot skips automatic 'on'", 1) test.add_cmd("stonith_admin", """-R true1 -a fence_dummy_auto_unfence """ """-o "mode=pass" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", """-R true2 -a fence_dummy_auto_unfence """ """-o "mode=pass" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", "-r node_fake -i 1 -v true1 -v true2") test.add_cmd("stonith_admin", "-B node_fake -t 5") test.add_stonith_log_pattern("Remapping multiple-device reboot of node_fake") test.add_stonith_log_pattern("perform op 'node_fake off' with 'true1'") test.add_stonith_log_pattern("perform op 'node_fake off' with 'true2'") test.add_stonith_log_pattern("Remapped off of node_fake complete, remapping to on") test.add_stonith_log_pattern("Undoing remap of reboot of node_fake") test.add_stonith_neg_log_pattern("perform op 'node_fake on' with") test.add_stonith_neg_log_pattern("'on' failure") test = self.new_test("cpg_remap_complex_1", "Verify remapped topology reboot in second level works if non-remapped first level fails", 1) test.add_cmd("stonith_admin", """-R false1 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", """-R true1 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", """-R true2 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", "-r node_fake -i 1 -v false1") test.add_cmd("stonith_admin", "-r node_fake -i 2 -v true1 -v true2") test.add_cmd("stonith_admin", "-B node_fake -t 5") test.add_stonith_log_pattern("perform op 'node_fake reboot' with 'false1'") test.add_stonith_log_pattern("Remapping multiple-device reboot of node_fake") test.add_stonith_log_pattern("perform op 'node_fake off' with 'true1'") test.add_stonith_log_pattern("perform op 'node_fake off' with 'true2'") test.add_stonith_log_pattern("Remapped off of node_fake complete, remapping to on") test.add_stonith_log_pattern("Ignoring true1 'on' failure (no capable peers) for node_fake") test.add_stonith_log_pattern("Ignoring true2 'on' failure (no capable peers) for node_fake") test.add_stonith_log_pattern("Undoing remap of reboot of node_fake") test = self.new_test("cpg_remap_complex_2", "Verify remapped topology reboot failure in second level proceeds to third level", 1) test.add_cmd("stonith_admin", """-R false1 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", """-R false2 -a fence_dummy -o "mode=fail" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", """-R true1 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", """-R true2 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", """-R true3 -a fence_dummy -o "mode=pass" -o "pcmk_host_list=node_fake" """) test.add_cmd("stonith_admin", "-r node_fake -i 1 -v false1") test.add_cmd("stonith_admin", "-r node_fake -i 2 -v true1 -v false2 -v true3") test.add_cmd("stonith_admin", "-r node_fake -i 3 -v true2") test.add_cmd("stonith_admin", "-B node_fake -t 5") test.add_stonith_log_pattern("perform op 'node_fake reboot' with 'false1'") test.add_stonith_log_pattern("Remapping multiple-device reboot of node_fake") test.add_stonith_log_pattern("perform op 'node_fake off' with 'true1'") test.add_stonith_log_pattern("perform op 'node_fake off' with 'false2'") test.add_stonith_log_pattern("Attempted to execute agent fence_dummy (off) the maximum number of times") test.add_stonith_log_pattern("Undoing remap of reboot of node_fake") test.add_stonith_log_pattern("perform op 'node_fake reboot' with 'true2'") test.add_stonith_neg_log_pattern("node_fake with true3") def setup_environment(self, use_corosync): """ Prepare the host before executing any tests """ if self.autogen_corosync_cfg and use_corosync: corosync_conf = (""" totem { version: 2 crypto_cipher: none crypto_hash: none nodeid: 101 secauth: off interface { ttl: 1 ringnumber: 0 mcastport: 6666 mcastaddr: 226.94.1.1 bindnetaddr: 127.0.0.1 } } logging { debug: off fileline: off to_syslog: no to_stderr: no syslog_facility: daemon timestamp: on to_logfile: yes logfile: @CRM_LOG_DIR@/corosync.test.log logfile_priority: info } """) os.system("cat <<-END >>/etc/corosync/corosync.conf\n%s\nEND" % (corosync_conf)) if use_corosync: ### make sure we are in control ### killall("corosync") self.start_corosync() subprocess.call(["cts-support", "install"]) def cleanup_environment(self, use_corosync): """ Clean up the host after executing desired tests """ if use_corosync: killall("corosync") if self.verbose and os.path.exists('@CRM_LOG_DIR@/corosync.test.log'): print("Corosync output") logfile = io.open('@CRM_LOG_DIR@/corosync.test.log', 'rt') for line in logfile.readlines(): print(line.strip()) os.remove('@CRM_LOG_DIR@/corosync.test.log') if self.autogen_corosync_cfg: os.system("rm -f /etc/corosync/corosync.conf") subprocess.call(["cts-support", "uninstall"]) 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['invalid-arg'] = "" self.options['cpg-only'] = 0 self.options['no-cpg'] = 0 self.options['show-usage'] = 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] == "-n" or args[i] == "--no-cpg": self.options['no-cpg'] = 1 elif args[i] == "-c" or args[i] == "--cpg-only": self.options['cpg-only'] = 1 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(self): """ 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 [--cpg-only | -c] Only run tests that require corosync.") print("\t [--no-cpg | -n] Only run tests that do not require corosync") print("\t [--run-only | -r 'testname'] Run a specific test") print("\t [--verbose | -V] Verbose output") 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 fencing regression tests as specified by arguments """ update_path() opts = TestOptions() opts.build_options(argv) use_corosync = 1 tests = Tests(opts.options['verbose']) tests.build_standalone_tests() tests.build_custom_timeout_tests() tests.build_api_sanity_tests() tests.build_fence_merge_tests() tests.build_fence_no_merge_tests() tests.build_unfence_tests() tests.build_unfence_on_target_tests() tests.build_nodeid_tests() tests.build_remap_tests() if opts.options['list-tests']: tests.print_list() sys.exit(CrmExit.OK) elif opts.options['show-usage']: opts.show_usage() sys.exit(CrmExit.OK) print("Starting ...") if opts.options['no-cpg']: use_corosync = 0 tests.setup_environment(use_corosync) 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() elif opts.options['no-cpg']: tests.run_no_cpg() tests.print_results() elif opts.options['cpg-only']: tests.run_cpg_only() tests.print_results() else: tests.run_tests() tests.print_results() tests.cleanup_environment(use_corosync) tests.exit() if __name__ == "__main__": main(sys.argv) diff --git a/daemons/fenced/fence_legacy.in b/daemons/fenced/fence_legacy.in index ae39e558a9..7324757e35 100755 --- a/daemons/fenced/fence_legacy.in +++ b/daemons/fenced/fence_legacy.in @@ -1,275 +1,275 @@ #!@PYTHON@ # Pacemaker targets compatibility with Python 2.7 and 3.2+ from __future__ import print_function, unicode_literals, absolute_import, division __copyright__ = "Copyright 2018 Andrew Beekhof " __license__ = "GNU General Public License version 2 or later (GPLv2+) WITHOUT ANY WARRANTY" import os import sys import argparse import subprocess VERSION = "1.1.0" USAGE = """Helper that presents a Pacemaker-style interface for Linux-HA stonith plugins Should never be invoked by the user directly Usage: fence_legacy [options] Options: -h usage -t sub agent -n nodename -o Action: on | off | reset (default) | stat | hostlist -s stonith command -q quiet mode -V version""" META_DATA = """ This agent should never be invoked by the user directly. https://www.clusterlabs.org/ Fencing Action Physical plug number or name of virtual machine Display help and exit """ ACTIONS = [ "on", "off", "reset", "reboot", "stat", "status", "metadata", "monitor", "list", "hostlist", "poweroff", "poweron" ] # These values must be kept in sync with include/crm/crm.h -class CrmExit: +class CrmExit(object): OK = 0 ERROR = 1 INVALID_PARAM = 2 def parse_cli_options(): """ Return parsed command-line options (as argparse namespace) """ # Don't add standard help option, so we can format it how we want parser = argparse.ArgumentParser(add_help=False) parser.add_argument("-t", metavar="SUBAGENT", dest="subagent", nargs=1, default="none", help="sub-agent") parser.add_argument("-n", metavar="NODE", dest="node", nargs=1, default="", help="name of target node") # The help text here is consistent with the original version, though # perhaps all actions should be listed. parser.add_argument("-o", metavar="ACTION", dest="action", nargs=1, choices=ACTIONS, default="reset", help="action: on | off | reset (default) | stat | hostlist") parser.add_argument("-s", metavar="COMMAND", dest="command", nargs=1, default="stonith", help="stonith command") parser.add_argument("-q", dest="quiet", action="store_true", help="quiet mode") parser.add_argument("-h", "--help", action="store_true", help="show usage and exit") # Don't use action="version", because that printed to stderr before # Python 3.4, and help2man doesn't like that. parser.add_argument("-V", "--version", action="store_true", help="show version and exit") return parser.parse_args() def parse_stdin_options(options): """ Update options namespace with options parsed from stdin """ nlines = 0 for line in sys.stdin: # Remove leading and trailing whitespace line = line.strip() # Skip blank lines and comments if line == "" or line[0] == "#": continue nlines = nlines + 1 # Parse option name and value (allow whitespace around equals sign) try: (name, value) = line.split("=", 1) name = name.rstrip() if name == "": raise ValueError except ValueError: print("parse error: illegal name in option %d" % nlines, file=sys.stderr) sys.exit(CrmExit.INVALID_PARAM) value = value.lstrip() if name == "plugin": options.subagent = value elif name in [ "option", "action" ]: options.action = value elif name == "nodename": options.node = value os.environ[name] = value elif name == "stonith": options.command = value elif name != "agent": # agent is used by fenced os.environ[name] = value def normalize_options(options): """ Use string rather than list of one string """ if not hasattr(options.subagent, "strip"): options.subagent = options.subagent[0] if not hasattr(options.node, "strip"): options.node = options.node[0] if not hasattr(options.action, "strip"): options.action = options.action[0] if not hasattr(options.command, "strip"): options.command = options.command[0] def build_command(options): """ Return command to execute (as list of arguments) """ if options.action in [ "hostlist", "list" ]: extra_args = [ "-l" ] elif options.action in [ "monitor", "stat", "status" ]: extra_args = [ "-S" ] else: if options.node == "": if not options.quiet: print("failed: no plug number") sys.exit(CrmExit.ERROR) extra_args = [ "-T", options.action, options.node ] return [ options.command, "-t", options.subagent, "-E" ] + extra_args def handle_local_options(options): """ Handle options that don't require the fence agent """ if options.help: print(USAGE) sys.exit(CrmExit.OK) if options.version: print(VERSION) sys.exit(CrmExit.OK) def remap_action(options): """ Pre-process requested action """ options.action = options.action.lower() if options.action == "metadata": print(META_DATA) sys.exit(CrmExit.OK) elif options.action in [ "hostlist", "list" ]: options.quiet = True # Remap accepted aliases to their actual commands elif options.action == "reboot": options.action = "reset" elif options.action == "poweron": options.action = "on" elif options.action == "poweroff": options.action = "off" def execute_command(options, cmd): """ Execute command and return its exit status """ if not options.quiet: print("Performing: " + " ".join(cmd)) return subprocess.call(cmd) def handle_result(options, status): """ Process fence agent result """ if status == 0: message = "success" exitcode = CrmExit.OK else: message = "failed" exitcode = CrmExit.ERROR if not options.quiet: print("%s: %s %d" % (message, options.node, status)) sys.exit(exitcode) def main(): """ Execute an LHA-style fence agent """ options = parse_cli_options() handle_local_options(options) normalize_options(options) parse_stdin_options(options) remap_action(options) cmd = build_command(options) status = execute_command(options, cmd) handle_result(options, status) if __name__ == "__main__": main() diff --git a/extra/logrotate/pacemaker.in b/extra/logrotate/pacemaker.in index 3ae810ed9a..fb361c6f3d 100644 --- a/extra/logrotate/pacemaker.in +++ b/extra/logrotate/pacemaker.in @@ -1,11 +1,12 @@ @CRM_LOG_DIR@/pacemaker.log @CRM_BUNDLE_DIR@/*/pacemaker.log { compress dateext weekly su hacluster haclient rotate 99 maxage 365 + maxsize 100M notifempty missingok copytruncate } diff --git a/pacemaker.spec.in b/pacemaker.spec.in index 4ee79b1032..fd7a06a877 100644 --- a/pacemaker.spec.in +++ b/pacemaker.spec.in @@ -1,812 +1,805 @@ # Globals and defines to control package behavior (configure these as desired) ## User and group to use for nonprivileged services %global uname hacluster %global gname haclient ## Where to install Pacemaker documentation %global pcmk_docdir %{_docdir}/%{name} ## GitHub entity that distributes source (for ease of using a fork) %global github_owner ClusterLabs ## Upstream pacemaker version, and its package version (specversion ## can be incremented to build packages reliably considered "newer" ## than previously built packages with the same pcmkversion) %global pcmkversion 2.0.0 %global specversion 1 ## Upstream commit (or git tag, such as "Pacemaker-" plus the ## {pcmkversion} macro for an official release) to use for this package %global commit HEAD ## Since git v2.11, the extent of abbreviation is autoscaled by default ## (used to be constant of 7), so we need to convey it for non-tags, too. %global commit_abbrev 7 ## Python major version to use (2, 3, or 0 for auto-detect) %global python_major 0 # Define globals for convenient use later ## Workaround to use parentheses in other globals %global lparen ( %global rparen ) ## Short version of git commit %define shortcommit %(c=%{commit}; case ${c} in Pacemaker-*%{rparen} echo ${c:10};; *%{rparen} echo ${c:0:%{commit_abbrev}};; esac) ## Whether this is a tagged release %define tag_release %([ %{commit} != Pacemaker-%{shortcommit} ]; echo $?) ## Whether this is a release candidate (in case of a tagged release) %define pre_release %([ "%{tag_release}" -eq 0 ] || { case "%{shortcommit}" in *-rc[[:digit:]]*%{rparen} false;; esac; }; echo $?) ## Heuristic used to infer bleeding-edge deployments that are ## less likely to have working versions of the documentation tools %define bleeding %(test ! -e /etc/yum.repos.d/fedora-rawhide.repo; echo $?) ## Whether this platform defaults to using systemd as an init system ## (needs to be evaluated prior to BuildRequires being enumerated and ## installed as it's intended to conditionally select some of these, and ## for that there are only few indicators with varying reliability: ## - presence of systemd-defined macros (when building in a full-fledged ## environment, which is not the case with ordinary mock-based builds) ## - systemd-aware rpm as manifested with the presence of particular ## macro (rpm itself will trivially always be present when building) ## - existence of /usr/lib/os-release file, which is something heavily ## propagated by systemd project ## - when not good enough, there's always a possibility to check ## particular distro-specific macros (incl. version comparison) %define systemd_native (%{?_unitdir:1}%{!?_unitdir:0}%{nil \ } || %{?__transaction_systemd_inhibit:1}%{!?__transaction_systemd_inhibit:0}%{nil \ } || %(test -f /usr/lib/os-release; test $? -ne 0; echo $?)) %if 0%{?fedora} > 20 || 0%{?rhel} > 7 ## Base GnuTLS cipher priorities (presumably only the initial, required keyword) ## overridable with "rpmbuild --define 'pcmk_gnutls_priorities PRIORITY-SPEC'" %define gnutls_priorities %{?pcmk_gnutls_priorities}%{!?pcmk_gnutls_priorities:@SYSTEM} %endif # Python-related definitions ## Use Python 3 on certain platforms if major version not specified %if %{?python_major} == 0 %if 0%{?fedora} > 26 || 0%{?rhel} > 7 %global python_major 3 %endif %endif ## Turn off auto-compilation of Python files outside Python specific paths, ## so there's no risk that unexpected "__python" macro gets picked to do the ## RPM-native byte-compiling there (only "{_datadir}/pacemaker/tests" affected) ## -- distro-dependent tricks or automake's fallback to be applied there %if %{defined _python_bytecompile_extra} %global _python_bytecompile_extra 0 %else ### the statement effectively means no RPM-native byte-compiling will occur at ### all, so distro-dependent tricks for Python-specific packages to be applied %global __os_install_post %(echo '%{__os_install_post}' | { sed -e 's!/usr/lib[^[:space:]]*/brp-python-bytecompile[[:space:]].*$!!g'; }) %endif ## Values that differ by Python major version %if 0%{?python_major} > 2 -%if 0%{?rhel} > 7 -%global python_path %{__python3} -%else -%global python_path /usr/bin/python%{?python3_pkgversion}%{!?python3_pkgversion:3} -%endif -%global python_pkg python3 -%global python_min 3.2 -%define py_site %{?python3_sitelib}%{!?python3_sitelib:%( - python3 -c 'from distutils.sysconfig import get_python_lib as gpl; print(gpl(1))' 2>/dev/null)} +%global python_name python3 +%global python_path %{?__python3}%{!?__python3:/usr/bin/python%{?python3_pkgversion}%{!?python3_pkgversion:3}} +%define python_site %{?python3_sitelib}%{!?python3_sitelib:%( + %{python_path} -c 'from distutils.sysconfig import get_python_lib as gpl; print(gpl(1))' 2>/dev/null)} %else %if 0%{?python_major} > 1 -%global python_path /usr/bin/python%{?python2_pkgversion}%{!?python2_pkgversion:2} -%global python_pkg python2 -%global python_min 2.7 -%define py_site %{?python2_sitelib}%{!?python2_sitelib:%( - python2 -c 'from distutils.sysconfig import get_python_lib as gpl; print(gpl(1))' 2>/dev/null)} +%global python_name python2 +%global python_path %{?__python2}%{!?__python2:/usr/bin/python%{?python2_pkgversion}%{!?python2_pkgversion:2}} +%define python_site %{?python2_sitelib}%{!?python2_sitelib:%( + %{python_path} -c 'from distutils.sysconfig import get_python_lib as gpl; print(gpl(1))' 2>/dev/null)} %else -%global python_min 2.7 -%global python_pkg python -%define py_site %{?python_sitelib}%{!?python_sitelib:%( +%global python_name python +%global python_path %{?__python}%{!?__python:/usr/bin/python%{?python_pkgversion}} +%define python_site %{?python_sitelib}%{!?python_sitelib:%( python -c 'from distutils.sysconfig import get_python_lib as gpl; print(gpl(1))' 2>/dev/null)} %endif %endif # Definitions for backward compatibility with older RPM versions ## Ensure the license macro behaves consistently (older RPM will otherwise ## overwrite it once it encounters "License:"). Courtesy Jason Tibbitts: ## https://pkgs.fedoraproject.org/cgit/rpms/epel-rpm-macros.git/tree/macros.zzz-epel?h=el6&id=e1adcb77 %if !%{defined _licensedir} %define description %{lua: rpm.define("license %doc") print("%description") } %endif # Define conditionals so that "rpmbuild --with " and # "rpmbuild --without " can enable and disable specific features ## Add option to enable support for stonith/external fencing agents %bcond_with stonithd ## Add option to create binaries suitable for use with profiling tools %bcond_with profiling ## Add option to create binaries with coverage analysis %bcond_with coverage ## Add option to skip generating documentation ## (the build tools aren't available everywhere) %bcond_without doc ## Add option to prefix package version with "0." ## (so later "official" packages will be considered updates) %bcond_with pre_release ## Add option to ship Upstart job files %bcond_with upstart_job ## Add option to turn off hardening of libraries and daemon executables %bcond_without hardening ## Add option to disable links for legacy daemon names %bcond_without legacy_links # Keep sane profiling data if requested %if %{with profiling} ## Disable -debuginfo package and stripping binaries/libraries %define debug_package %{nil} %endif # Define the release version # (do not look at externally enforced pre-release flag for tagged releases # as only -rc tags, captured with the second condition, implies that then) %if (!%{tag_release} && %{with pre_release}) || 0%{pre_release} %if 0%{pre_release} %define pcmk_release 0.%{specversion}.%(s=%{shortcommit}; echo ${s: -3}) %else %define pcmk_release 0.%{specversion}.%{shortcommit}.git %endif %else %if 0%{tag_release} %define pcmk_release %{specversion} %else %define pcmk_release %{specversion}.%{shortcommit}.git %endif %endif Name: pacemaker Summary: Scalable High-Availability cluster resource manager Version: %{pcmkversion} Release: %{pcmk_release}%{?dist} %if %{defined _unitdir} License: GPLv2+ and LGPLv2+ %else # initscript is Revised BSD License: GPLv2+ and LGPLv2+ and BSD %endif Url: http://www.clusterlabs.org Group: System Environment/Daemons # Hint: use "spectool -s 0 pacemaker.spec" (rpmdevtools) to check the final URL: # https://github.com/ClusterLabs/pacemaker/archive/e91769e5a39f5cb2f7b097d3c612368f0530535e/pacemaker-e91769e.tar.gz Source0: https://github.com/%{github_owner}/%{name}/archive/%{commit}/%{name}-%{shortcommit}.tar.gz Requires: resource-agents Requires: %{name}-libs = %{version}-%{release} Requires: %{name}-cluster-libs = %{version}-%{release} Requires: %{name}-cli = %{version}-%{release} %if !%{defined _unitdir} Requires: procps-ng Requires: psmisc %endif %{?systemd_requires} -# Pacemaker requires a minimum Python functionality -Requires: %{python_pkg} >= %{python_min} -BuildRequires: %{python_pkg}-devel >= %{python_min} +Requires: %{python_path} +BuildRequires: %{python_name}-devel # Pacemaker requires a minimum libqb functionality Requires: libqb >= 0.13.0 BuildRequires: libqb-devel >= 0.13.0 # Basics required for the build (even if usually satisfied through other BRs) BuildRequires: coreutils findutils grep sed # Required for core functionality BuildRequires: automake autoconf gcc libtool pkgconfig libtool-ltdl-devel BuildRequires: pkgconfig(glib-2.0) >= 2.16 BuildRequires: libxml2-devel libxslt-devel libuuid-devel BuildRequires: bzip2-devel # Enables optional functionality BuildRequires: ncurses-devel docbook-style-xsl BuildRequires: help2man gnutls-devel pam-devel pkgconfig(dbus-1) %if %{systemd_native} BuildRequires: pkgconfig(systemd) %endif Requires: corosync >= 2.0.0 BuildRequires: corosynclib-devel >= 2.0.0 %if %{with stonithd} BuildRequires: cluster-glue-libs-devel %endif ## (note no avoiding effect when building through non-customized mock) %if !%{bleeding} %if %{with doc} BuildRequires: inkscape asciidoc publican %endif %endif %description Pacemaker is an advanced, scalable High-Availability cluster resource manager. It supports more than 16 node clusters with significant capabilities for managing resources and dependencies. It will run scripts at initialization, when machines go up or down, when related resources fail and can be configured to periodically check resource health. Available rpmbuild rebuild options: --with(out) : coverage doc stonithd hardening pre_release profiling upstart_job %package cli License: GPLv2+ and LGPLv2+ Summary: Command line tools for controlling Pacemaker clusters Group: System Environment/Daemons Requires: %{name}-libs = %{version}-%{release} Requires: perl-TimeDate Requires: procps-ng Requires: psmisc Requires(post):coreutils %description cli Pacemaker is an advanced, scalable High-Availability cluster resource manager. The %{name}-cli package contains command line tools that can be used to query and control the cluster from machines that may, or may not, be part of the cluster. %package libs License: GPLv2+ and LGPLv2+ Summary: Core Pacemaker libraries Group: System Environment/Daemons Requires(pre): shadow-utils %description libs Pacemaker is an advanced, scalable High-Availability cluster resource manager. The %{name}-libs package contains shared libraries needed for cluster nodes and those just running the CLI tools. %package cluster-libs License: GPLv2+ and LGPLv2+ Summary: Cluster Libraries used by Pacemaker Group: System Environment/Daemons Requires: %{name}-libs = %{version}-%{release} %description cluster-libs Pacemaker is an advanced, scalable High-Availability cluster resource manager. The %{name}-cluster-libs package contains cluster-aware shared libraries needed for nodes that will form part of the cluster nodes. %package remote %if %{defined _unitdir} License: GPLv2+ and LGPLv2+ %else # initscript is Revised BSD License: GPLv2+ and LGPLv2+ and BSD %endif Summary: Pacemaker remote daemon for non-cluster nodes Group: System Environment/Daemons Requires: %{name}-libs = %{version}-%{release} Requires: %{name}-cli = %{version}-%{release} Requires: resource-agents %if !%{defined _unitdir} Requires: procps-ng %endif # -remote can be fully independent of systemd %{?systemd_ordering}%{!?systemd_ordering:%{?systemd_requires}} %description remote Pacemaker is an advanced, scalable High-Availability cluster resource manager. The %{name}-remote package contains the Pacemaker Remote daemon which is capable of extending pacemaker functionality to remote nodes not running the full corosync/cluster stack. %package libs-devel License: GPLv2+ and LGPLv2+ Summary: Pacemaker development package Group: Development/Libraries Requires: %{name}-libs%{?_isa} = %{version}-%{release} Requires: %{name}-cluster-libs%{?_isa} = %{version}-%{release} Requires: libuuid-devel%{?_isa} libtool-ltdl-devel%{?_isa} Requires: libxml2-devel%{?_isa} libxslt-devel%{?_isa} Requires: bzip2-devel%{?_isa} glib2-devel%{?_isa} Requires: libqb-devel%{?_isa} Requires: corosynclib-devel%{?_isa} >= 2.0.0 %description libs-devel Pacemaker is an advanced, scalable High-Availability cluster resource manager. The %{name}-libs-devel package contains headers and shared libraries for developing tools for Pacemaker. %package cts License: GPLv2+ and LGPLv2+ Summary: Test framework for cluster-related technologies like Pacemaker Group: System Environment/Daemons -Requires: %{python_pkg} >= %{python_min} +Requires: %{python_path} Requires: %{name}-libs = %{version}-%{release} Requires: procps-ng Requires: psmisc BuildArch: noarch # systemd python bindings are separate package in some distros %if %{defined systemd_requires} %if 0%{?fedora} > 22 || 0%{?rhel} > 7 -Requires: %{python_pkg}-systemd +Requires: %{python_name}-systemd %else %if 0%{?fedora} > 20 || 0%{?rhel} > 6 Requires: systemd-python %endif %endif %endif %description cts Test framework for cluster-related technologies like Pacemaker %package doc License: CC-BY-SA-4.0 Summary: Documentation for Pacemaker Group: Documentation BuildArch: noarch %description doc Documentation for Pacemaker. Pacemaker is an advanced, scalable High-Availability cluster resource manager. %prep %setup -q -n %{name}-%{commit} %build # Early versions of autotools (e.g. RHEL <= 5) do not support --docdir export docdir=%{pcmk_docdir} export systemdunitdir=%{?_unitdir}%{!?_unitdir:no} %if %{with hardening} # prefer distro-provided hardening flags in case they are defined # through _hardening_{c,ld}flags macros, configure script will # use its own defaults otherwise; if such hardenings are completely # undesired, rpmbuild using "--without hardening" # (or "--define '_without_hardening 1'") export CFLAGS_HARDENED_EXE="%{?_hardening_cflags}" export CFLAGS_HARDENED_LIB="%{?_hardening_cflags}" export LDFLAGS_HARDENED_EXE="%{?_hardening_ldflags}" export LDFLAGS_HARDENED_LIB="%{?_hardening_ldflags}" %endif ./autogen.sh %{configure} \ - %{?python_path: PYTHON=%{python_path}} \ + PYTHON=%{python_path} \ %{!?with_hardening: --disable-hardening} \ %{!?with_legacy_links: --disable-legacy-links} \ %{?with_profiling: --with-profiling} \ %{?with_coverage: --with-coverage} \ %{!?with_doc: --with-brand=} \ %{?gnutls_priorities: --with-gnutls-priorities="%{gnutls_priorities}"} \ --with-initdir=%{_initrddir} \ --localstatedir=%{_var} \ --with-version=%{version}-%{release} %if 0%{?suse_version} >= 1200 # Fedora handles rpath removal automagically sed -i 's|^hardcode_libdir_flag_spec=.*|hardcode_libdir_flag_spec=""|g' libtool sed -i 's|^runpath_var=LD_RUN_PATH|runpath_var=DIE_RPATH_DIE|g' libtool %endif make %{_smp_mflags} V=1 all %check { cts/cts-scheduler --run one-or-more-unrunnable-instances \ && cts/cts-cli \ && touch .CHECKED } 2>&1 | sed 's/[fF]ail/faiil/g' # prevent false positives in rpmlint [ -f .CHECKED ] && rm -f -- .CHECKED exit $? # TODO remove when rpm<4.14 compatibility irrelevant %install # skip automake-native Python byte-compilation, since RPM-native one (possibly # distro-confined to Python-specific directories, which is currently the only # relevant place, anyway) assures proper intrinsic alignment with wider system # (such as with py_byte_compile macro, which is concurrent Fedora/EL specific) make install \ DESTDIR=%{buildroot} V=1 docdir=%{pcmk_docdir} \ %{?_python_bytecompile_extra:%{?py_byte_compile:am__py_compile=true}} mkdir -p ${RPM_BUILD_ROOT}%{_sysconfdir}/sysconfig install -m 644 daemons/pacemakerd/pacemaker.sysconfig ${RPM_BUILD_ROOT}%{_sysconfdir}/sysconfig/pacemaker install -m 644 tools/crm_mon.sysconfig ${RPM_BUILD_ROOT}%{_sysconfdir}/sysconfig/crm_mon %if %{with upstart_job} mkdir -p ${RPM_BUILD_ROOT}%{_sysconfdir}/init install -m 644 pacemakerd/pacemaker.upstart ${RPM_BUILD_ROOT}%{_sysconfdir}/init/pacemaker.conf install -m 644 pacemakerd/pacemaker.combined.upstart ${RPM_BUILD_ROOT}%{_sysconfdir}/init/pacemaker.combined.conf install -m 644 tools/crm_mon.upstart ${RPM_BUILD_ROOT}%{_sysconfdir}/init/crm_mon.conf %endif %if %{defined _unitdir} mkdir -p ${RPM_BUILD_ROOT}%{_localstatedir}/lib/rpm-state/%{name} %endif # Don't package static libs find %{buildroot} -name '*.a' -type f -print0 | xargs -0 rm -f find %{buildroot} -name '*.la' -type f -print0 | xargs -0 rm -f # For now, don't package the servicelog-related binaries built only for # ppc64le when certain dependencies are installed. If they get more exercise by # advanced users, we can reconsider. rm -f %{buildroot}/%{_sbindir}/notifyServicelogEvent rm -f %{buildroot}/%{_sbindir}/ipmiservicelogd # Don't ship init scripts for systemd based platforms %if %{defined _unitdir} rm -f %{buildroot}/%{_initrddir}/pacemaker rm -f %{buildroot}/%{_initrddir}/pacemaker_remote %endif # Byte-compile Python sources where suitable and the distro procedures known -%if %{defined py_byte_compile} && %{defined python_path} +%if %{defined py_byte_compile} %{py_byte_compile %{python_path} %{buildroot}%{_datadir}/pacemaker/tests} %if !%{defined _python_bytecompile_extra} -%{py_byte_compile %{python_path} %{buildroot}%{py_site}/cts} +%{py_byte_compile %{python_path} %{buildroot}%{python_site}/cts} %endif %endif %if %{with coverage} GCOV_BASE=%{buildroot}/%{_var}/lib/pacemaker/gcov mkdir -p $GCOV_BASE find . -name '*.gcno' -type f | while read F ; do D=`dirname $F` mkdir -p ${GCOV_BASE}/$D cp $F ${GCOV_BASE}/$D done %endif %post %if %{defined _unitdir} %systemd_post pacemaker.service %else /sbin/chkconfig --add pacemaker || : %endif %preun %if %{defined _unitdir} %systemd_preun pacemaker.service %else /sbin/service pacemaker stop >/dev/null 2>&1 || : if [ "$1" -eq 0 ]; then # Package removal, not upgrade /sbin/chkconfig --del pacemaker || : fi %endif %postun %if %{defined _unitdir} %systemd_postun_with_restart pacemaker.service %endif %pre remote %if %{defined _unitdir} # Stop the service before anything is touched, and remember to restart # it as one of the last actions (compared to using systemd_postun_with_restart, # this avoids suicide when sbd is in use) systemctl --quiet is-active pacemaker_remote if [ $? -eq 0 ] ; then mkdir -p %{_localstatedir}/lib/rpm-state/%{name} touch %{_localstatedir}/lib/rpm-state/%{name}/restart_pacemaker_remote systemctl stop pacemaker_remote >/dev/null 2>&1 else rm -f %{_localstatedir}/lib/rpm-state/%{name}/restart_pacemaker_remote fi %endif %post remote %if %{defined _unitdir} %systemd_post pacemaker_remote.service %else /sbin/chkconfig --add pacemaker_remote || : %endif %preun remote %if %{defined _unitdir} %systemd_preun pacemaker_remote.service %else /sbin/service pacemaker_remote stop >/dev/null 2>&1 || : if [ "$1" -eq 0 ]; then # Package removal, not upgrade /sbin/chkconfig --del pacemaker_remote || : fi %endif %postun remote %if %{defined _unitdir} # This next line is a no-op, because we stopped the service earlier, but # we leave it here because it allows us to revert to the standard behavior # in the future if desired %systemd_postun_with_restart pacemaker_remote.service # Explicitly take care of removing the flag-file(s) upon final removal if [ "$1" -eq 0 ] ; then rm -f %{_localstatedir}/lib/rpm-state/%{name}/restart_pacemaker_remote fi %endif %posttrans remote %if %{defined _unitdir} if [ -e %{_localstatedir}/lib/rpm-state/%{name}/restart_pacemaker_remote ] ; then systemctl start pacemaker_remote >/dev/null 2>&1 rm -f %{_localstatedir}/lib/rpm-state/%{name}/restart_pacemaker_remote fi %endif %post cli %if %{defined _unitdir} %systemd_post crm_mon.service %endif if [ "$1" -eq 2 ]; then # Package upgrade, not initial install: # Move any pre-2.0 logs to new location to ensure they get rotated { mv -fbS.rpmsave %{_var}/log/pacemaker.log* %{_var}/log/pacemaker \ || mv -f %{_var}/log/pacemaker.log* %{_var}/log/pacemaker } >/dev/null 2>/dev/null || : fi %preun cli %if %{defined _unitdir} %systemd_preun crm_mon.service %endif %postun cli %if %{defined _unitdir} %systemd_postun_with_restart crm_mon.service %endif %pre libs getent group %{gname} >/dev/null || groupadd -r %{gname} -g 189 getent passwd %{uname} >/dev/null || useradd -r -g %{gname} -u 189 -s /sbin/nologin -c "cluster user" %{uname} exit 0 %if %{defined ldconfig_scriptlets} %ldconfig_scriptlets libs %ldconfig_scriptlets cluster-libs %else %post libs -p /sbin/ldconfig %postun libs -p /sbin/ldconfig %post cluster-libs -p /sbin/ldconfig %postun cluster-libs -p /sbin/ldconfig %endif %files ########################################################### %config(noreplace) %{_sysconfdir}/sysconfig/pacemaker %{_sbindir}/pacemakerd %if %{defined _unitdir} %{_unitdir}/pacemaker.service %else %{_initrddir}/pacemaker %endif %exclude %{_libexecdir}/pacemaker/cts-log-watcher %exclude %{_libexecdir}/pacemaker/cts-support %exclude %{_sbindir}/pacemaker-remoted %if %{with legacy_links} %exclude %{_sbindir}/pacemaker_remoted %endif %{_libexecdir}/pacemaker/* %{_sbindir}/crm_attribute %{_sbindir}/crm_master %{_sbindir}/fence_legacy %{_sbindir}/stonith_admin %doc %{_mandir}/man7/pacemaker-controld.* %doc %{_mandir}/man7/pacemaker-schedulerd.* %doc %{_mandir}/man7/pacemaker-fenced.* %doc %{_mandir}/man7/ocf_pacemaker_controld.* %doc %{_mandir}/man7/ocf_pacemaker_o2cb.* %doc %{_mandir}/man7/ocf_pacemaker_remote.* %doc %{_mandir}/man8/crm_attribute.* %doc %{_mandir}/man8/crm_master.* %doc %{_mandir}/man8/fence_legacy.* %doc %{_mandir}/man8/pacemakerd.* %doc %{_mandir}/man8/stonith_admin.* %doc %{_datadir}/pacemaker/alerts %license licenses/GPLv2 %doc COPYING %doc ChangeLog %dir %attr (750, %{uname}, %{gname}) %{_var}/lib/pacemaker/cib %dir %attr (750, %{uname}, %{gname}) %{_var}/lib/pacemaker/pengine /usr/lib/ocf/resource.d/pacemaker/controld /usr/lib/ocf/resource.d/pacemaker/o2cb /usr/lib/ocf/resource.d/pacemaker/remote %if %{with upstart_job} %config(noreplace) %{_sysconfdir}/init/pacemaker.conf %config(noreplace) %{_sysconfdir}/init/pacemaker.combined.conf %endif %files cli %dir %attr (750, root, %{gname}) %{_sysconfdir}/pacemaker %config(noreplace) %{_sysconfdir}/logrotate.d/pacemaker %config(noreplace) %{_sysconfdir}/sysconfig/crm_mon %if %{defined _unitdir} %{_unitdir}/crm_mon.service %endif %if %{with upstart_job} %config(noreplace) %{_sysconfdir}/init/crm_mon.conf %endif %{_sbindir}/attrd_updater %{_sbindir}/cibadmin %{_sbindir}/crm_diff %{_sbindir}/crm_error %{_sbindir}/crm_failcount %{_sbindir}/crm_mon %{_sbindir}/crm_node %{_sbindir}/crm_resource %{_sbindir}/crm_standby %{_sbindir}/crm_verify %{_sbindir}/crmadmin %{_sbindir}/iso8601 %{_sbindir}/crm_shadow %{_sbindir}/crm_simulate %{_sbindir}/crm_report %{_sbindir}/crm_ticket %exclude %{_datadir}/pacemaker/alerts %exclude %{_datadir}/pacemaker/tests %{_datadir}/pacemaker %{_datadir}/snmp/mibs/PCMK-MIB.txt %exclude /usr/lib/ocf/resource.d/pacemaker/controld %exclude /usr/lib/ocf/resource.d/pacemaker/o2cb %exclude /usr/lib/ocf/resource.d/pacemaker/remote %dir /usr/lib/ocf %dir /usr/lib/ocf/resource.d /usr/lib/ocf/resource.d/pacemaker %doc %{_mandir}/man7/* %exclude %{_mandir}/man7/pacemaker-controld.* %exclude %{_mandir}/man7/pacemaker-schedulerd.* %exclude %{_mandir}/man7/pacemaker-fenced.* %exclude %{_mandir}/man7/ocf_pacemaker_controld.* %exclude %{_mandir}/man7/ocf_pacemaker_o2cb.* %exclude %{_mandir}/man7/ocf_pacemaker_remote.* %doc %{_mandir}/man8/* %exclude %{_mandir}/man8/crm_attribute.* %exclude %{_mandir}/man8/crm_master.* %exclude %{_mandir}/man8/fence_legacy.* %exclude %{_mandir}/man8/pacemakerd.* %exclude %{_mandir}/man8/pacemaker-remoted.* %exclude %{_mandir}/man8/stonith_admin.* %license licenses/GPLv2 %doc COPYING %doc ChangeLog %dir %attr (750, %{uname}, %{gname}) %{_var}/lib/pacemaker %dir %attr (750, %{uname}, %{gname}) %{_var}/lib/pacemaker/blackbox %dir %attr (750, %{uname}, %{gname}) %{_var}/lib/pacemaker/cores %dir %attr (770, %{uname}, %{gname}) %{_var}/log/pacemaker %dir %attr (770, %{uname}, %{gname}) %{_var}/log/pacemaker/bundles %files libs %{_libdir}/libcib.so.* %{_libdir}/liblrmd.so.* %{_libdir}/libcrmservice.so.* %{_libdir}/libcrmcommon.so.* %{_libdir}/libpe_status.so.* %{_libdir}/libpe_rules.so.* %{_libdir}/libpengine.so.* %{_libdir}/libstonithd.so.* %{_libdir}/libtransitioner.so.* %license licenses/LGPLv2.1 %doc COPYING %doc ChangeLog %files cluster-libs %{_libdir}/libcrmcluster.so.* %license licenses/LGPLv2.1 %doc COPYING %doc ChangeLog %files remote %config(noreplace) %{_sysconfdir}/sysconfig/pacemaker %if %{defined _unitdir} # state directory is shared between the subpackets # let rpm take care of removing it once it isn't # referenced anymore and empty %ghost %dir %{_localstatedir}/lib/rpm-state/%{name} %{_unitdir}/pacemaker_remote.service %else %{_initrddir}/pacemaker_remote %endif %{_sbindir}/pacemaker-remoted %if %{with legacy_links} %{_sbindir}/pacemaker_remoted %endif %{_mandir}/man8/pacemaker-remoted.* %license licenses/GPLv2 %doc COPYING %doc ChangeLog %files doc %doc %{pcmk_docdir} %license licenses/CC-BY-SA-4.0 %files cts -%{py_site}/cts +%{python_site}/cts %{_datadir}/pacemaker/tests %{_libexecdir}/pacemaker/cts-log-watcher %{_libexecdir}/pacemaker/cts-support %license licenses/GPLv2 %doc COPYING %doc ChangeLog %files libs-devel %{_includedir}/pacemaker %{_libdir}/*.so %if %{with coverage} %{_var}/lib/pacemaker/gcov %endif %{_libdir}/pkgconfig/*.pc %license licenses/LGPLv2.1 %doc COPYING %doc ChangeLog %changelog diff --git a/tools/report.collector.in b/tools/report.collector.in index ff78ac1ee6..352eb37901 100644 --- a/tools/report.collector.in +++ b/tools/report.collector.in @@ -1,866 +1,869 @@ # # Originally based on hb_report # Copyright 2007 Dejan Muhamedagic # # Later changes copyright 2010-2018 Andrew Beekhof # # This source code is licensed under the GNU General Public License version 2 # or later (GPLv2+) WITHOUT ANY WARRANTY. # if echo $REPORT_HOME | grep -qs '^/' then debug "Using full path to working directory: $REPORT_HOME" else REPORT_HOME="$HOME/$REPORT_HOME" debug "Canonicalizing working directory path: $REPORT_HOME" fi detect_host # # find files newer than a and older than b # isnumber() { echo "$*" | grep -qs '^[0-9][0-9]*$' } touchfile() { t=`mktemp` && perl -e "\$file=\"$t\"; \$tm=$1;" -e 'utime $tm, $tm, $file;' && echo $t } find_files_clean() { [ -z "$from_stamp" ] || rm -f "$from_stamp" [ -z "$to_stamp" ] || rm -f "$to_stamp" from_stamp="" to_stamp="" } find_files() { dirs= from_time=$2 to_time=$3 for d in $1; do if [ -d $d ]; then dirs="$dirs $d" fi done if [ x"$dirs" = x ]; then return fi isnumber "$from_time" && [ "$from_time" -gt 0 ] || { warning "sorry, can't find files in [ $1 ] based on time if you don't supply time" return } trap find_files_clean 0 if ! from_stamp=`touchfile $from_time`; then warning "sorry, can't create temporary file for find_files" return fi findexp="-newer $from_stamp" if isnumber "$to_time" && [ "$to_time" -gt 0 ]; then if ! to_stamp=`touchfile $to_time`; then warning "sorry, can't create temporary file for find_files" find_files_clean return fi findexp="$findexp ! -newer $to_stamp" fi find $dirs -type f $findexp find_files_clean trap "" 0 } # # check permissions of files/dirs # pl_checkperms() { perl -e ' # check permissions and ownership # uid and gid are numeric # everything must match exactly # no error checking! (file should exist, etc) ($filename, $perms, $in_uid, $in_gid) = @ARGV; ($mode,$uid,$gid) = (stat($filename))[2,4,5]; $p=sprintf("%04o", $mode & 07777); $p ne $perms and exit(1); $uid ne $in_uid and exit(1); $gid ne $in_gid and exit(1); ' $* } num_id() { getent $1 $2 | awk -F: '{print $3}' } chk_id() { [ "$2" ] && return 0 echo "$1: id not found" return 1 } check_perms() { while read type f p uid gid; do if [ ! -e "$f" ]; then echo "$f doesn't exist" continue elif [ ! -$type "$f" ]; then echo "$f has wrong type" continue fi n_uid=`num_id passwd $uid` chk_id "$uid" "$n_uid" || continue n_gid=`num_id group $gid` chk_id "$gid" "$n_gid" || continue pl_checkperms $f $p $n_uid $n_gid || { echo "wrong permissions or ownership for $f:" ls -ld $f } done } # # coredumps # findbinary() { random_binary=`which cat 2>/dev/null` # suppose we are lucky binary=`gdb $random_binary $1 < /dev/null 2>/dev/null | grep 'Core was generated' | awk '{print $5}' | sed "s/^.//;s/[.':]*$//"` if [ x = x"$binary" ]; then debug "Could not detect the program name for core $1 from the gdb output; will try with file(1)" binary=$(file $1 | awk '/from/{ for( i=1; i<=NF; i++ ) if( $i == "from" ) { print $(i+1) break } }') binary=`echo $binary | tr -d "'"` binary=$(echo $binary | tr -d '`') if [ "$binary" ]; then binary=`which $binary 2>/dev/null` fi fi if [ x = x"$binary" ]; then warning "Could not find the program path for core $1" return fi fullpath=`which $binary 2>/dev/null` if [ x = x"$fullpath" ]; then if [ -x $CRM_DAEMON_DIR/$binary ]; then echo $CRM_DAEMON_DIR/$binary debug "Found the program at $CRM_DAEMON_DIR/$binary for core $1" else warning "Could not find the program path for core $1" fi else echo $fullpath debug "Found the program at $fullpath for core $1" fi } getbt() { which gdb > /dev/null 2>&1 || { warning "Please install gdb to get backtraces" return } for corefile; do absbinpath=`findbinary $corefile` [ x = x"$absbinpath" ] && continue echo "====================== start backtrace ======================" ls -l $corefile # Summary first... gdb -batch -n -quiet -ex ${BT_OPTS:-"thread apply all bt"} -ex quit \ $absbinpath $corefile 2>/dev/null echo "====================== start detail ======================" # Now the unreadable details... gdb -batch -n -quiet -ex ${BT_OPTS:-"thread apply all bt full"} -ex quit \ $absbinpath $corefile 2>/dev/null echo "======================= end backtrace =======================" done } dump_status_and_config() { crm_mon -1 2>&1 | grep -v '^Last upd' > $target/$CRM_MON_F cibadmin -Ql 2>/dev/null > $target/${CIB_F}.live } getconfig() { cluster=$1; shift; target=$1; shift; for cf in $*; do if [ -e "$cf" ]; then cp -a "$cf" $target/ fi done if is_running pacemaker-controld; then dump_status_and_config crm_node -p > "$target/$MEMBERSHIP_F" 2>&1 echo "$host" > $target/RUNNING elif is_running pacemaker-remoted; then dump_status_and_config echo "$host" > $target/RUNNING # Pre-2.0.0 daemon name in case we're collecting on a mixed-version cluster elif is_running pacemaker_remoted; then dump_status_and_config echo "$host" > $target/RUNNING else echo "$host" > $target/STOPPED fi } get_readable_cib() { target="$1"; shift; if [ -f "$target/$CIB_F" ]; then crm_verify -V -x "$target/$CIB_F" >"$target/$CRM_VERIFY_F" 2>&1 if which crm >/dev/null 2>&1 ; then CIB_file="$target/$CIB_F" crm configure show >"$target/$CIB_TXT_F" 2>&1 elif which pcs >/dev/null 2>&1 ; then pcs config -f "$target/$CIB_F" >"$target/$CIB_TXT_F" 2>&1 fi fi } # # remove values of sensitive attributes # # this is not proper xml parsing, but it will work under the # circumstances sanitize_xml_attrs() { sed $( for patt in $SANITIZE; do echo "-e /name=\"$patt\"/s/value=\"[^\"]*\"/value=\"****\"/" done ) } sanitize_hacf() { awk ' $1=="stonith_host"{ for( i=5; i<=NF; i++ ) $i="****"; } {print} ' } sanitize_one_clean() { [ -z "$tmp" ] || rm -f "$tmp" tmp="" [ -z "$ref" ] || rm -f "$ref" ref="" } sanitize() { file=$1 compress="" if [ -z "$SANITIZE" ]; then return fi echo $file | grep -qs 'gz$' && compress=gzip echo $file | grep -qs 'bz2$' && compress=bzip2 if [ "$compress" ]; then decompress="$compress -dc" else compress=cat decompress=cat fi trap sanitize_one_clean 0 tmp=`mktemp` ref=`mktemp` if [ -z "$tmp" -o -z "$ref" ]; then sanitize_one_clean fatal "cannot create temporary files" fi touch -r $file $ref # save the mtime if [ "`basename $file`" = ha.cf ]; then sanitize_hacf else $decompress | sanitize_xml_attrs | $compress fi < $file > $tmp mv $tmp $file # note: cleaning $tmp up is still needed even after it's renamed # because its temp directory is still there. touch -r $ref $file sanitize_one_clean trap "" 0 } # # get some system info # distro() { if which lsb_release >/dev/null 2>&1 then lsb_release -d | sed -e 's/^Description:\s*//' debug "Using lsb_release for distribution info" return fi relf=`ls /etc/debian_version 2>/dev/null` || relf=`ls /etc/slackware-version 2>/dev/null` || relf=`ls -d /etc/*-release 2>/dev/null` && { for f in $relf; do test -f $f && { echo "`ls $f` `cat $f`" debug "Found `echo $relf | tr '\n' ' '` distribution release file(s)" return } done } warning "No lsb_release, no /etc/*-release, no /etc/debian_version: no distro information" } pkg_ver() { if which dpkg >/dev/null 2>&1 ; then pkg_mgr="deb" elif which rpm >/dev/null 2>&1 ; then pkg_mgr="rpm" elif which pkg_info >/dev/null 2>&1 ; then pkg_mgr="pkg_info" elif which pkginfo >/dev/null 2>&1 ; then pkg_mgr="pkginfo" else warning "Unknown package manager" return fi debug "The package manager is: $pkg_mgr" echo "The package manager is: $pkg_mgr" echo "Installed packages:" case $pkg_mgr in deb) dpkg-query -f '${Package} ${Version} ${Architecture}\n' -W | sort echo for pkg in $*; do if dpkg-query -W $pkg 2>/dev/null ; then debug "Verifying installation of: $pkg" echo "Verifying installation of: $pkg" debsums -s $pkg 2>/dev/null fi done ;; rpm) rpm -qa --qf '%{name} %{version}-%{release} - %{distribution} %{arch}\n' | sort echo for pkg in $*; do if rpm -q $pkg >/dev/null 2>&1 ; then debug "Verifying installation of: $pkg" echo "Verifying installation of: $pkg" rpm --verify $pkg 2>&1 fi done ;; pkg_info) pkg_info ;; pkginfo) pkginfo | awk '{print $3}' # format? ;; esac } getbacktraces() { debug "Looking for backtraces: $*" flist=$( for f in `find_files "$CRM_CORE_DIRS" $1 $2`; do bf=`basename $f` test `expr match $bf core` -gt 0 && echo $f done) if [ "$flist" ]; then for core in $flist; do log "Found core file: `ls -al $core`" done # Make a copy of them in case we need more data later # Luckily they compress well mkdir cores >/dev/null 2>&1 cp -a $flist cores/ shrink cores rm -rf cores # Now get as much as we can from them automagically for f in $flist; do getbt $f done fi } getpeinputs() { if [ -n "$PE_STATE_DIR" ]; then flist=$( find_files "$PE_STATE_DIR" "$1" "$2" | sed "s,`dirname $PE_STATE_DIR`/,,g" ) if [ "$flist" ]; then (cd $(dirname "$PE_STATE_DIR") && tar cf - $flist) | (cd "$3" && tar xf -) debug "found `echo $flist | wc -w` scheduler input files in $PE_STATE_DIR" fi fi } getblackboxes() { flist=$( find_files $BLACKBOX_DIR $1 $2 ) for bb in $flist; do bb_short=`basename $bb` qb-blackbox $bb > $3/${bb_short}.blackbox 2>&1 info "Extracting contents of blackbox: $bb_short" done } # # some basic system info and stats # sys_info() { cluster=$1; shift echo "Platform: `uname`" echo "Kernel release: `uname -r`" echo "Architecture: `uname -m`" if [ `uname` = Linux ]; then echo "Distribution: `distro`" fi echo cibadmin --version 2>&1 | head -1 cibadmin -! 2>&1 case $cluster in corosync) /usr/sbin/corosync -v 2>&1 | head -1 ;; esac # Cluster glue version hash (if available) stonith -V 2>/dev/null # Resource agents version hash echo "resource-agents: `grep 'Build version:' /usr/lib/ocf/resource.d/heartbeat/.ocf-shellfuncs`" echo pkg_ver $* } sys_stats() { set -x uname -n uptime ps axf ps auxw top -b -n 1 ifconfig -a ip addr list netstat -i arp -an test -d /proc && { cat /proc/cpuinfo } lsscsi lspci mount df set +x } dlm_dump() { if which dlm_tool >/dev/null 2>&1 ; then if is_running dlm_controld; then echo "--- Lockspace overview:" dlm_tool ls -n echo "---Lockspace history:" dlm_tool dump echo "---Lockspace status:" dlm_tool status dlm_tool status -v echo "---Lockspace config:" dlm_tool dump_config dlm_tool log_plock dlm_tool ls | grep name | while read X N ; do echo "--- Lockspace $N:" dlm_tool lockdump "$N" dlm_tool lockdebug -svw "$N" done fi fi } drbd_info() { test -f /proc/drbd && { echo "--- /proc/drbd:" cat /proc/drbd 2>&1 echo } if which drbd-overview >/dev/null 2>&1; then echo "--- drbd-overview:" drbd-overview 2>&1 echo fi if which drbdsetup >/dev/null 2>&1; then echo "--- drbdsetup status:" drbdsetup status --verbose --statistics 2>&1 echo echo "--- drbdsetup events2:" drbdsetup events2 --timestamps --statistics --now 2>&1 echo fi if which drbdadm >/dev/null 2>&1; then echo "--- drbdadm show-gi:" for res in $(drbdsetup status | grep -e ^\\S | awk '{ print $1 }'); do echo "$res:" drbdadm show-gi $res 2>&1 echo done fi } iscfvarset() { test "`getcfvar $1 $2`" } iscfvartrue() { getcfvar $1 $2 $3 | grep -E -qsi "^(true|y|yes|on|1)" } iscfvarfalse() { getcfvar $1 $2 $3 | grep -E -qsi "^(false|n|no|off|0)" } find_syslog() { priority="$1" # Always include system logs (if we can find them) msg="Mark:pcmk:`perl -e 'print time()'`" logger -p "$priority" "$msg" >/dev/null 2>&1 # Force buffer flush killall -HUP rsyslogd >/dev/null 2>&1 sleep 2 # Give syslog time to catch up in case it's busy findmsg 1 "$msg" } get_logfiles_cs() { if [ ! -f "$cf_file" ]; then return fi debug "Reading $cf_type log settings from $cf_file" # The default value of to_syslog is yes. if ! iscfvarfalse $cf_type to_syslog "$cf_file"; then facility_cs=$(getcfvar $cf_type syslog_facility "$cf_file") if [ -z "$facility_cs" ]; then facility_cs="daemon" fi find_syslog "$facility_cs.info" fi if [ "$SOS_MODE" = "1" ]; then return fi if iscfvartrue $cf_type to_logfile "$cf_file"; then logfile=$(getcfvar $cf_type logfile "$cf_file") if [ -f "$logfile" ]; then debug "Log settings found for cluster type $cf_type: $logfile" echo "$logfile" fi fi } get_logfiles() { cf_type=$1 cf_file="$2" case $cf_type in corosync) get_logfiles_cs;; esac . @CONFIGDIR@/pacemaker facility="$PCMK_logfacility" if [ -z "$facility" ]; then facility="daemon" fi if [ "$facility" != "$facility_cs" ]&&[ "$facility" != none ]; then find_syslog "$facility.notice" fi if [ "$SOS_MODE" = "1" ]; then return fi logfile="$PCMK_logfile" if [ "$logfile" != none ]; then if [ -z "$logfile" ]; then for logfile in "@CRM_LOG_DIR@/pacemaker.log" "/var/log/pacemaker.log"; do if [ -f "$logfile" ]; then debug "Log settings not found for Pacemaker, assuming $logfile" echo "$logfile" break fi done elif [ -f "$logfile" ]; then debug "Log settings found for Pacemaker: $logfile" echo "$logfile" fi fi # Look for detail logs: # - initial pacemakerd logs and tracing might go to a different file pattern="Starting Pacemaker" # - make sure we get something from the scheduler pattern="$pattern\\|Calculated transition" # - cib and pacemaker-execd updates # (helpful on non-DC nodes and when cluster has been up for a long time) pattern="$pattern\\|cib_perform_op\\|process_lrm_event" # - pacemaker_remote might use a different file pattern="$pattern\\|pacemaker[-_]remoted:" findmsg 3 "$pattern" } essential_files() { cat< /dev/null 2>&1 if [ $? -eq 0 ]; then cl_have_journald=1 else cl_have_journald=0 fi cl_lognames="$CL_LOGFILES" if [ $cl_have_journald -eq 1 ]; then cl_lognames="$cl_lognames journalctl" fi cl_lognames=$(trim "$cl_lognames") if [ -z "$cl_lognames" ]; then return fi # YYYY-MM-DD HH:MM:SS cl_start_ymd=$(date -d @${CL_START} +"%F %T") cl_end_ymd=$(date -d @${CL_END} +"%F %T") debug "Gathering logs from $cl_start_ymd to $cl_end_ymd:" debug " $cl_lognames" # Remove our temporary file if we get interrupted here trap '[ -z "$cl_pattfile" ] || rm -f "$cl_pattfile"' 0 # Create a temporary file with patterns to grep for cl_pattfile=$(mktemp) || fatal "cannot create temporary files" for cl_pattern in $LOG_PATTERNS; do echo "$cl_pattern" done > $cl_pattfile echo "Log pattern matches from $REPORT_TARGET:" > $ANALYSIS_F if [ -n "$CL_LOGFILES" ]; then for cl_logfile in $CL_LOGFILES; do cl_extract="$(basename $cl_logfile).extract.txt" if [ ! -f "$cl_logfile" ]; then # Not a file continue elif [ -f "$cl_extract" ]; then # We already have it continue fi dumplogset "$cl_logfile" $LOG_START $LOG_END > "$cl_extract" sanitize "$cl_extract" grep -f "$cl_pattfile" "$cl_extract" >> $ANALYSIS_F done fi # Collect systemd logs if present if [ $cl_have_journald -eq 1 ]; then journalctl --since "$cl_start_ymd" --until "$cl_end_ymd" > journal.log grep -f "$cl_pattfile" journal.log >> $ANALYSIS_F fi rm -f $cl_pattfile trap "" 0 } debug "Initializing $REPORT_TARGET subdir" if [ "$REPORT_MASTER" != "$REPORT_TARGET" ]; then if [ -e $REPORT_HOME/$REPORT_TARGET ]; then warning "Directory $REPORT_HOME/$REPORT_TARGET already exists, using /tmp/$$/$REPORT_TARGET instead" REPORT_HOME=/tmp/$$ fi fi mkdir -p $REPORT_HOME/$REPORT_TARGET cd $REPORT_HOME/$REPORT_TARGET case $CLUSTER in any) cluster=`get_cluster_type`;; *) cluster=$CLUSTER;; esac cluster_cf=`find_cluster_cf $cluster` # If cluster stack is still "any", this might be a Pacemaker Remote node, # so don't complain in that case. if [ -z "$cluster_cf" ] && [ $cluster != "any" ]; then warning "Could not determine the location of your cluster configuration" fi if [ "$SEARCH_LOGS" = "1" ]; then logfiles=$(get_logfiles "$cluster" "$cluster_cf" | sort -u) fi logfiles="$(trim "$logfiles $EXTRA_LOGS")" if [ -z "$logfiles" ]; then which journalctl > /dev/null 2>&1 if [ $? -eq 0 ]; then info "Systemd journal will be only log collected" else info "No logs will be collected" fi info "No log files found or specified with --logfile /some/path" fi debug "Config: $cluster ($cluster_cf) $logfiles" sys_info $cluster $PACKAGES > $SYSINFO_F essential_files $cluster | check_perms > $PERMISSIONS_F 2>&1 getconfig $cluster "$REPORT_HOME/$REPORT_TARGET" "$cluster_cf" "$CRM_CONFIG_DIR/$CIB_F" "/etc/drbd.conf" "/etc/drbd.d" "/etc/booth" getpeinputs $LOG_START $LOG_END $REPORT_HOME/$REPORT_TARGET getbacktraces $LOG_START $LOG_END > $REPORT_HOME/$REPORT_TARGET/$BT_F getblackboxes $LOG_START $LOG_END $REPORT_HOME/$REPORT_TARGET case $cluster in corosync) if is_running corosync; then corosync-blackbox >corosync-blackbox-live.txt 2>&1 # corosync-fplay > corosync-blackbox.txt tool=`pickfirst corosync-objctl corosync-cmapctl` case $tool in *objctl) $tool -a > corosync.dump 2>/dev/null;; *cmapctl) $tool > corosync.dump 2>/dev/null;; esac corosync-quorumtool -s -i > corosync.quorum 2>&1 fi ;; esac dc=`crm_mon -1 2>/dev/null | awk '/Current DC/ {print $3}'` if [ "$REPORT_TARGET" = "$dc" ]; then echo "$REPORT_TARGET" > DC fi dlm_dump > $DLM_DUMP_F 2>&1 sys_stats > $SYSSTATS_F 2>&1 drbd_info > $DRBD_INFO_F 2>&1 debug "Sanitizing files: $SANITIZE" # # replace sensitive info with '****' # cf="" if [ ! -z "$cluster_cf" ]; then cf=`basename $cluster_cf` fi for f in "$cf" "$CIB_F" "$CIB_F.live" pengine/*; do if [ -f "$f" ]; then sanitize "$f" fi done # For convenience, generate human-readable version of CIB and any XML errors -# in it (AFTER sanitizing, so we don't need to sanitize this output) -get_readable_cib "$REPORT_HOME/$REPORT_TARGET" +# in it (AFTER sanitizing, so we don't need to sanitize this output). +# sosreport does this itself, so we do not need to when run by sosreport. +if [ "$SOS_MODE" != "1" ]; then + get_readable_cib "$REPORT_HOME/$REPORT_TARGET" +fi collect_logs "$LOG_START" "$LOG_END" $logfiles # Purge files containing no information for f in `ls -1`; do if [ -d "$f" ]; then continue elif [ ! -s "$f" ]; then case $f in *core*) log "Detected empty core file: $f";; *) debug "Removing empty file: `ls -al $f`" rm -f $f ;; esac fi done # Parse for events for l in $logfiles; do b="$(basename $l).extract.txt" node_events "$b" > $EVENTS_F # Link the first logfile to a standard name if it doesn't yet exist if [ -e "$b" -a ! -e "$HALOG_F" ]; then ln -s "$b" "$HALOG_F" fi done if [ -e $REPORT_HOME/.env ]; then debug "Localhost: $REPORT_MASTER $REPORT_TARGET" elif [ "$REPORT_MASTER" != "$REPORT_TARGET" ]; then debug "Streaming report back to $REPORT_MASTER" (cd $REPORT_HOME && tar cf - $REPORT_TARGET) if [ "$REMOVE" = "1" ]; then cd rm -rf $REPORT_HOME fi fi # vim: set expandtab tabstop=8 softtabstop=4 shiftwidth=4 textwidth=80: diff --git a/tools/report.common.in b/tools/report.common.in index ee5b8b3386..9f48f14fb1 100644 --- a/tools/report.common.in +++ b/tools/report.common.in @@ -1,867 +1,874 @@ # # Originally based on hb_report # Copyright 2007 Dejan Muhamedagic # # Later changes copyright 2010-2018 Andrew Beekhof # # This source code is licensed under the GNU General Public License version 2 # or later (GPLv2+) WITHOUT ANY WARRANTY. # host=`uname -n` shorthost=`echo $host | sed s:\\\\..*::` if [ -z $verbose ]; then verbose=0 fi # Target Files EVENTS_F=events.txt ANALYSIS_F=analysis.txt HALOG_F=cluster-log.txt BT_F=backtraces.txt SYSINFO_F=sysinfo.txt SYSSTATS_F=sysstats.txt DLM_DUMP_F=dlm_dump.txt CRM_MON_F=crm_mon.txt MEMBERSHIP_F=members.txt CRM_VERIFY_F=crm_verify.txt PERMISSIONS_F=permissions.txt CIB_F=cib.xml CIB_TXT_F=cib.txt DRBD_INFO_F=drbd_info.txt EVENT_PATTERNS=" state do_state_transition membership pcmk_peer_update.*(lost|memb): quorum (crmd|pacemaker-controld).*crm_update_quorum pause Process.pause.detected resources (lrmd|pacemaker-execd).*rsc:(start|stop) stonith te_fence_node|fenced.*(requests|(Succeeded|Failed).to.|result=) start_stop shutdown.decision|Corosync.Cluster.Engine|corosync.*Initializing.transport|Executive.Service.RELEASE|crm_shutdown:.Requesting.shutdown|pcmk_shutdown:.Shutdown.complete " # superset of all packages of interest on all distros # (the package manager will be used to validate the installation # of any of these packages that are installed) PACKAGES="pacemaker pacemaker-libs pacemaker-cluster-libs libpacemaker3 pacemaker-remote pacemaker-pygui pacemaker-pymgmt pymgmt-client corosync corosynclib libcorosync4 resource-agents cluster-glue-libs cluster-glue libglue2 ldirectord ocfs2-tools ocfs2-tools-o2cb ocfs2console ocfs2-kmp-default ocfs2-kmp-pae ocfs2-kmp-xen ocfs2-kmp-debug ocfs2-kmp-trace drbd drbd-kmp-xen drbd-kmp-pae drbd-kmp-default drbd-kmp-debug drbd-kmp-trace drbd-pacemaker drbd-utils drbd-bash-completion drbd-xen lvm2 lvm2-clvm cmirrord libdlm libdlm2 libdlm3 hawk ruby lighttpd kernel-default kernel-pae kernel-xen glibc " # Potential locations of system log files SYSLOGS=" /var/log/* /var/logs/* /var/syslog/* /var/adm/* /var/log/ha/* /var/log/cluster/* " # Whether pacemaker-remoted was found (0 = yes, 1 = no, -1 = haven't looked yet) REMOTED_STATUS=-1 # # keep the user posted # record() { if [ x != x"$REPORT_HOME" -a -d "${REPORT_HOME}/$shorthost" ]; then rec="${REPORT_HOME}/$shorthost/report.out" elif [ x != x"${l_base}" -a -d "${l_base}" ]; then rec="${l_base}/report.summary" else rec="/dev/null" fi printf "%-10s $*\n" "$shorthost:" 2>&1 >> "${rec}" } log() { printf "%-10s $*\n" "$shorthost:" 1>&2 record "$*" } debug() { if [ $verbose -gt 0 ]; then log "Debug: $*" else record "Debug: $*" fi } info() { log "$*" } warning() { log "WARN: $*" } fatal() { log "ERROR: $*" exit 1 } # check if process of given substring in its name does exist; # only look for processes originated by user 0 (by UID), "@CRM_DAEMON_USER@" # or effective user running this script, and/or group 0 (by GID), # "@CRM_DAEMON_GROUP@" or one of the groups the effective user belongs to # (there's no business in probing any other processes) is_running() { ps -G "0 $(getent group '@CRM_DAEMON_GROUP@' 2>/dev/null | cut -d: -f3) $(id -G)" \ -u "0 @CRM_DAEMON_USER@ $(id -u)" -f \ | grep -Eqs $(echo "$1" | sed -e 's/^\(.\)/[\1]/') } has_remoted() { if [ $REMOTED_STATUS -eq -1 ]; then REMOTED_STATUS=1 if which pacemaker-remoted >/dev/null 2>&1; then REMOTED_STATUS=0 # Check for pre-2.0.0 daemon name in case we have mixed-version cluster elif which pacemaker_remoted >/dev/null 2>&1; then REMOTED_STATUS=0 elif [ -x "@sbindir@/pacemaker-remoted" ]; then REMOTED_STATUS=0 elif [ -x "@sbindir@/pacemaker_remoted" ]; then REMOTED_STATUS=0 else # @TODO: the binary might be elsewhere, # but a global search is too expensive for d in /{usr,opt}/{local/,}{s,}bin; do if [ -x "${d}/pacemaker-remoted" ]; then REMOTED_STATUS=0 elif [ -x "${d}/pacemaker_remoted" ]; then REMOTED_STATUS=0 fi done fi fi return $REMOTED_STATUS } # found_dir found_dir() { echo "$2" info "Pacemaker $1 found in: $2" } detect_daemon_dir() { info "Searching for where Pacemaker daemons live... this may take a while" for d in \ {/usr,/usr/local,/opt/local,@exec_prefix@}/{libexec,lib64,lib}/pacemaker do # pacemaker and pacemaker-cts packages can install to daemon directory, # so check for a file from each if [ -e $d/pacemaker-schedulerd ] || [ -e $d/cts-exec-helper ]; then found_dir "daemons" "$d" return fi done # Pacemaker Remote nodes don't need to install daemons if has_remoted; then info "Pacemaker daemons not found (this appears to be a Pacemaker Remote node)" return fi for f in $(find / -maxdepth $maxdepth -type f -name pacemaker-schedulerd -o -name cts-exec-helper); do d=$(dirname "$f") found_dir "daemons" "$d" return done fatal "Pacemaker daemons not found (nonstandard installation?)" } detect_cib_dir() { d="${local_state_dir}/lib/pacemaker/cib" if [ -f "$d/cib.xml" ]; then found_dir "config files" "$d" return fi # Pacemaker Remote nodes don't need a CIB if has_remoted; then info "Pacemaker config not found (this appears to be a Pacemaker Remote node)" return fi info "Searching for where Pacemaker keeps config information... this may take a while" # TODO: What about false positives where someone copied the CIB? for f in $(find / -maxdepth $maxdepth -type f -name cib.xml); do d=$(dirname $f) found_dir "config files" "$d" return done warning "Pacemaker config not found (nonstandard installation?)" } detect_state_dir() { if [ -n "$CRM_CONFIG_DIR" ]; then # Assume new layout # $local_state_dir/lib/pacemaker/(cib,pengine,blackbox,cores) dirname "$CRM_CONFIG_DIR" # Pacemaker Remote nodes might not have a CRM_CONFIG_DIR elif [ -d "$local_state_dir/lib/pacemaker" ]; then echo $local_state_dir/lib/pacemaker fi } detect_pe_dir() { config_root="$1" d="$config_root/pengine" if [ -d "$d" ]; then found_dir "scheduler inputs" "$d" return fi if has_remoted; then info "Pacemaker scheduler inputs not found (this appears to be a Pacemaker Remote node)" return fi info "Searching for where Pacemaker keeps scheduler inputs... this may take a while" for d in $(find / -maxdepth $maxdepth -type d -name pengine); do found_dir "scheduler inputs" "$d" return done fatal "Pacemaker scheduler inputs not found (nonstandard installation?)" } detect_host() { local_state_dir=@localstatedir@ if [ -d $local_state_dir/run ]; then CRM_STATE_DIR=$local_state_dir/run/crm else info "Searching for where Pacemaker keeps runtime data... this may take a while" for d in `find / -maxdepth $maxdepth -type d -name run`; do local_state_dir=`dirname $d` CRM_STATE_DIR=$d/crm break done info "Found: $CRM_STATE_DIR" fi debug "Machine runtime directory: $local_state_dir" debug "Pacemaker runtime data located in: $CRM_STATE_DIR" CRM_DAEMON_DIR=$(detect_daemon_dir) CRM_CONFIG_DIR=$(detect_cib_dir) config_root=$(detect_state_dir) # Older versions had none BLACKBOX_DIR=$config_root/blackbox debug "Pacemaker blackboxes (if any) located in: $BLACKBOX_DIR" PE_STATE_DIR=$(detect_pe_dir "$config_root") CRM_CORE_DIRS="" for d in $config_root/cores $local_state_dir/lib/corosync; do if [ -d $d ]; then CRM_CORE_DIRS="$CRM_CORE_DIRS $d" fi done debug "Core files located under: $CRM_CORE_DIRS" } time2str() { perl -e "use POSIX; print strftime('%x %X',localtime($1));" } get_time() { perl -e "\$time=\"$*\";" -e ' $unix_tm = 0; eval "use Date::Parse"; if (index($time, ":") < 0) { } elsif (!$@) { $unix_tm = str2time($time); } else { eval "use Date::Manip"; if (!$@) { $unix_tm = UnixDate(ParseDateString($time), "%s"); } } if ($unix_tm != "") { print int($unix_tm); } else { print ""; } ' } -get_time_() { - warning "Unknown time format used by: $*" -} - get_time_syslog() { awk '{print $1,$2,$3}' } get_time_legacy() { awk '{print $2}' | sed 's/_/ /' } get_time_iso8601() { awk '{print $1}' } get_time_format_for_string() { l="$*" t=$(get_time `echo $l | get_time_syslog`) if [ "x$t" != x ]; then echo syslog return fi t=$(get_time `echo $l | get_time_iso8601`) if [ "x$t" != x ]; then echo iso8601 return fi t=$(get_time `echo $l | get_time_legacy`) if [ "x$t" != x ]; then echo legacy return fi } get_time_format() { t=0 l="" func="" trycnt=10 while [ $trycnt -gt 0 ] && read l; do func=$(get_time_format_for_string $l) if [ "x$func" != x ]; then break fi trycnt=$(($trycnt-1)) done #debug "Logfile uses the $func time format" echo $func } +get_time_from_line() { + GTFL_FORMAT="$1" + shift + if [ "$GTFL_FORMAT" = "" ]; then + GTFL_FORMAT=$(get_time_format_for_string "$@") + fi + case $GTFL_FORMAT in + syslog|legacy|iso8601) + get_time $(echo "$@" | get_time_${GTFL_FORMAT}) + ;; + *) + warning "Unknown time format in: $@" + ;; + esac +} + get_first_time() { l="" format=$1 while read l; do - t=$(echo $l | get_time_$format) - ts=$(get_time $t) + ts=$(get_time_from_line "$format" "$l") if [ "x$ts" != x ]; then echo "$ts" return fi done } get_last_time() { l="" best=`date +%s` # Now format=$1 while read l; do - t=$(echo $l | get_time_$format) - ts=$(get_time $t) + ts=$(get_time_from_line "$format" "$l") if [ "x$ts" != x ]; then best=$ts fi done echo $best } linetime() { - l=`tail -n +$2 $1 | grep -a ":[0-5][0-9]:" | head -n 1` - format=`get_time_format_for_string $l` - t=`echo $l | get_time_$format` - get_time "$t" + get_time_from_line "" $(tail -n +$2 $1 | grep -a ":[0-5][0-9]:" | head -n 1) } # # findmsg # # Print the names of up to system logs that contain , # ordered by most recently modified. # findmsg() { max=$1 pattern="$2" found=0 # List all potential system logs ordered by most recently modified. candidates=$(ls -1td $SYSLOGS 2>/dev/null) if [ -z "$candidates" ]; then debug "No system logs found to search for pattern \'$pattern\'" return fi # Portable way to handle files with spaces in their names. SAVE_IFS=$IFS IFS=" " # Check each log file for matches. logfiles="" for f in $candidates; do local cat="" # We only care about readable files with something in them. if [ ! -f "$f" ] || [ ! -r "$f" ] || [ ! -s "$f" ] ; then continue fi cat=$(find_decompressor "$f") # We want to avoid grepping through potentially huge binary logs such # as lastlog. However, control characters sometimes find their way into # text logs, so we use a heuristic of more than 256 nonprintable # characters in the file's first kilobyte. if [ $($cat "$f" 2>/dev/null | head -c 1024 | tr -d '[:print:][:space:]' | wc -c) -gt 256 ] then continue fi # Our patterns are ASCII, so we can use LC_ALL="C" to speed up grep $cat "$f" 2>/dev/null | LC_ALL="C" grep -q -e "$pattern" if [ $? -eq 0 ]; then # Add this file to the list of hits # (using newline as separator to handle spaces in names). if [ -z "$logfiles" ]; then logfiles="$f" else logfiles="$logfiles $f" fi # If we have enough hits, print them and return. found=$(($found+1)) if [ $found -ge $max ]; then break fi fi done 2>/dev/null IFS=$SAVE_IFS if [ -z "$logfiles" ]; then debug "Pattern \'$pattern\' not found in any system logs" else debug "Pattern \'$pattern\' found in: [ $logfiles ]" echo "$logfiles" fi } node_events() { if [ -e $1 ]; then Epatt=`echo "$EVENT_PATTERNS" | while read title p; do [ -n "$p" ] && echo -n "|$p"; done | sed 's/.//' ` grep -E "$Epatt" $1 fi } pickfirst() { for x; do which $x >/dev/null 2>&1 && { echo $x return 0 } done return 1 } shrink() { olddir=$PWD dir=`dirname $1` base=`basename $1` target=$1.tar tar_options="cf" variant=`pickfirst bzip2 gzip xz false` case $variant in bz*) tar_options="jcf" target="$target.bz2" ;; gz*) tar_options="zcf" target="$target.gz" ;; xz*) tar_options="Jcf" target="$target.xz" ;; *) warning "Could not find a compression program, the resulting tarball may be huge" ;; esac if [ -e $target ]; then fatal "Destination $target already exists, specify an alternate name with --dest" fi cd $dir >/dev/null 2>&1 tar $tar_options $target $base >/dev/null 2>&1 cd $olddir >/dev/null 2>&1 echo $target } findln_by_time() { local logf=$1 local tm=$2 local first=1 # Some logs can be massive (over 1,500,000,000 lines have been seen in the wild) # Even just 'wc -l' on these files can take 10+ minutes local fileSize=`ls -lh | awk '{ print $5 }' | grep -ie G` if [ x$fileSize != x ]; then warning "$logf is ${fileSize} in size and could take many hours to process. Skipping." return fi local last=`wc -l < $logf` while [ $first -le $last ]; do mid=$((($last+$first)/2)) trycnt=10 while [ $trycnt -gt 0 ]; do tmid=`linetime $logf $mid` [ "$tmid" ] && break warning "cannot extract time: $logf:$mid; will try the next one" trycnt=$(($trycnt-1)) # shift the whole first-last segment first=$(($first-1)) last=$(($last-1)) mid=$((($last+$first)/2)) done if [ -z "$tmid" ]; then warning "giving up on log..." return fi if [ $tmid -gt $tm ]; then last=$(($mid-1)) elif [ $tmid -lt $tm ]; then first=$(($mid+1)) else break fi done echo $mid } dumplog() { local logf=$1 local from_line=$2 local to_line=$3 [ "$from_line" ] || return tail -n +$from_line $logf | if [ "$to_line" ]; then head -$(($to_line-$from_line+1)) else cat fi } # # find log/set of logs which are interesting for us # # # find log slices # find_decompressor() { case $1 in *bz2) echo "bzip2 -dc" ;; *gz) echo "gzip -dc" ;; *xz) echo "xz -dc" ;; *) echo "cat" ;; esac } # # check if the log contains a piece of our segment # is_our_log() { local logf=$1 local from_time=$2 local to_time=$3 local cat=`find_decompressor $logf` local format=`$cat $logf | get_time_format` local first_time=`$cat $logf | head -10 | get_first_time $format` local last_time=`$cat $logf | tail -10 | get_last_time $format` if [ x = "x$first_time" -o x = "x$last_time" ]; then warning "Skipping bad logfile '$1': Could not determine log dates" return 0 # skip (empty log?) fi if [ $from_time -gt $last_time ]; then # we shouldn't get here anyway if the logs are in order return 2 # we're past good logs; exit fi if [ $from_time -ge $first_time ]; then return 3 # this is the last good log fi # have to go further back if [ x = "x$to_time" -o $to_time -ge $first_time ]; then return 1 # include this log else return 0 # don't include this log fi } # # go through archived logs (timewise backwards) and see if there # are lines belonging to us # (we rely on untouched log files, i.e. that modify time # hasn't been changed) # arch_logs() { local logf=$1 local from_time=$2 local to_time=$3 # look for files such as: ha-log-20090308 or # ha-log-20090308.gz (.bz2) or ha-log.0, etc ls -t $logf $logf*[0-9z] 2>/dev/null | while read next_log; do is_our_log $next_log $from_time $to_time case $? in 0) ;; # noop, continue 1) echo $next_log # include log and continue debug "Found log $next_log" ;; 2) break;; # don't go through older logs! 3) echo $next_log # include log and continue debug "Found log $next_log" break ;; # don't go through older logs! esac done } # # print part of the log # drop_tmp_file() { [ -z "$tmp" ] || rm -f "$tmp" } print_logseg() { local logf=$1 local from_time=$2 local to_time=$3 # uncompress to a temp file (if necessary) local cat=`find_decompressor $logf` if [ "$cat" != "cat" ]; then tmp=`mktemp` $cat $logf > $tmp trap drop_tmp_file 0 sourcef=$tmp else sourcef=$logf tmp="" fi if [ "$from_time" = 0 ]; then FROM_LINE=1 else FROM_LINE=`findln_by_time $sourcef $from_time` fi if [ -z "$FROM_LINE" ]; then warning "couldn't find line for time $from_time; corrupt log file?" return fi TO_LINE="" if [ "$to_time" != 0 ]; then TO_LINE=`findln_by_time $sourcef $to_time` if [ -z "$TO_LINE" ]; then warning "couldn't find line for time $to_time; corrupt log file?" return fi if [ $FROM_LINE -lt $TO_LINE ]; then dumplog $sourcef $FROM_LINE $TO_LINE log "Including segment [$FROM_LINE-$TO_LINE] from $logf" else debug "Empty segment [$FROM_LINE-$TO_LINE] from $logf" fi else dumplog $sourcef $FROM_LINE $TO_LINE log "Including all logs after line $FROM_LINE from $logf" fi drop_tmp_file trap "" 0 } # # find log/set of logs which are interesting for us # dumplogset() { local logf=$1 local from_time=$2 local to_time=$3 local logf_set=`arch_logs $logf $from_time $to_time` if [ x = "x$logf_set" ]; then return fi local num_logs=`echo "$logf_set" | wc -l` local oldest=`echo $logf_set | awk '{print $NF}'` local newest=`echo $logf_set | awk '{print $1}'` local mid_logfiles=`echo $logf_set | awk '{for(i=NF-1; i>1; i--) print $i}'` # the first logfile: from $from_time to $to_time (or end) # logfiles in the middle: all # the last logfile: from beginning to $to_time (or end) case $num_logs in 1) print_logseg $newest $from_time $to_time;; *) print_logseg $oldest $from_time 0 for f in $mid_logfiles; do `find_decompressor $f` $f debug "including complete $f logfile" done print_logseg $newest 0 $to_time ;; esac } # cut out a stanza getstanza() { awk -v name="$1" ' !in_stanza && NF==2 && /^[a-z][a-z]*[[:space:]]*{/ { # stanza start if ($1 == name) in_stanza = 1 } in_stanza { print } in_stanza && NF==1 && $1 == "}" { exit } ' } # supply stanza in $1 and variable name in $2 # (stanza is optional) getcfvar() { cf_type=$1; shift; cf_var=$1; shift; cf_file=$* [ -f "$cf_file" ] || return case $cf_type in corosync) sed 's/#.*//' < $cf_file | if [ $# -eq 2 ]; then getstanza "$cf_var" shift 1 else cat fi | awk -v varname="$cf_var" ' NF==2 && match($1,varname":$")==1 { print $2; exit; } ' ;; esac } pickfirst() { for x; do which $x >/dev/null 2>&1 && { echo $x return 0 } done return 1 } # # figure out the cluster type, depending on the process list # and existence of configuration files # get_cluster_type() { if is_running corosync; then tool=`pickfirst corosync-objctl corosync-cmapctl` case $tool in *objctl) quorum=`$tool -a | grep quorum.provider | sed 's/.*=\s*//'`;; *cmapctl) quorum=`$tool | grep quorum.provider | sed 's/.*=\s*//'`;; esac stack="corosync" # Now we're guessing... # TODO: Technically these could be anywhere :-/ elif [ -f /etc/corosync/corosync.conf ]; then stack="corosync" else # We still don't know. This might be a Pacemaker Remote node, # or the configuration might be in a nonstandard location. stack="any" fi debug "Detected the '$stack' cluster stack" echo $stack } find_cluster_cf() { case $1 in corosync) best_size=0 best_file="" # TODO: Technically these could be anywhere :-/ for cf in /etc/corosync/corosync.conf; do if [ -f $cf ]; then size=`wc -l $cf | awk '{print $1}'` if [ $size -gt $best_size ]; then best_size=$size best_file=$cf fi fi done if [ -z "$best_file" ]; then debug "Looking for corosync configuration file. This may take a while..." for f in `find / -maxdepth $maxdepth -type f -name corosync.conf`; do best_file=$f break done fi debug "Located corosync config file: $best_file" echo "$best_file" ;; any) # Cluster type is undetermined. Don't complain, because this # might be a Pacemaker Remote node. ;; *) warning "Unknown cluster type: $1" ;; esac } # # check for the major prereq for a) parameter parsing and b) # parsing logs # t=`get_time "12:00"` if [ "$t" = "" ]; then fatal "please install the perl Date::Parse module (perl-DateTime-Format-DateParse on Fedora/Red Hat)" fi # vim: set expandtab tabstop=8 softtabstop=4 shiftwidth=4 textwidth=80: