diff --git a/cts/cts-exec.in b/cts/cts-exec.in index 3f6579013b..451884288b 100644 --- a/cts/cts-exec.in +++ b/cts/cts-exec.in @@ -1,989 +1,989 @@ #!@PYTHON@ """Regression tests for Pacemaker's pacemaker-execd.""" # pylint doesn't like the module name "cts-execd" which is an invalid complaint for this file # but probably something we want to continue warning about elsewhere # pylint: disable=invalid-name # pacemaker imports need to come after we modify sys.path, which pylint will complain about. # pylint: disable=wrong-import-position __copyright__ = "Copyright 2012-2024 the Pacemaker project contributors" __license__ = "GNU General Public License version 2 or later (GPLv2+) WITHOUT ANY WARRANTY" import argparse import os import stat import sys import subprocess import shutil import tempfile # Where to find test binaries # Prefer the source tree if available TEST_DIR = sys.path[0] # These imports allow running from a source checkout after running `make`. # Note that while this doesn't necessarily mean it will successfully run tests, # but being able to see --help output can be useful. if os.path.exists("@abs_top_srcdir@/python"): sys.path.insert(0, "@abs_top_srcdir@/python") # pylint: disable=comparison-of-constants,comparison-with-itself,condition-evals-to-constant if os.path.exists("@abs_top_builddir@/python") and "@abs_top_builddir@" != "@abs_top_srcdir@": sys.path.insert(0, "@abs_top_builddir@/python") from pacemaker.buildoptions import BuildOptions from pacemaker.exitstatus import ExitStatus from pacemaker._cts.process import killall, exit_if_proc_running, stdout_from_command from pacemaker._cts.test import Test, Tests # 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 def update_path(): # pylint: disable=protected-access """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)" % (BuildOptions._BUILD_DIR, TEST_DIR)) # For pacemaker-execd, cts-exec-helper, and pacemaker-remoted new_path = "%s/daemons/execd:%s" % (BuildOptions._BUILD_DIR, new_path) new_path = "%s/tools:%s" % (BuildOptions._BUILD_DIR, new_path) # For crm_resource # For pacemaker-fenced new_path = "%s/daemons/fenced:%s" % (BuildOptions._BUILD_DIR, new_path) # For cts-support new_path = "%s/cts/support:%s" % (BuildOptions._BUILD_DIR, new_path) else: print("Running tests from the install tree: %s (not %s)" % (BuildOptions.DAEMON_DIR, TEST_DIR)) # For cts-exec-helper, cts-support, pacemaker-execd, pacemaker-fenced, # and pacemaker-remoted new_path = "%s:%s" % (BuildOptions.DAEMON_DIR, new_path) print('Using PATH="%s"' % new_path) os.environ['PATH'] = new_path class ExecTest(Test): """Executor for a single pacemaker-execd regression test.""" def __init__(self, name, description, **kwargs): """Create a new ExecTest instance. Arguments: name -- A unique name for this test. This can be used on the command line to specify that only a specific test should be executed. description -- A meaningful description for the test. Keyword arguments: tls -- Enable pacemaker-remoted. """ Test.__init__(self, name, description, **kwargs) self.tls = kwargs.get("tls", False) if self.tls: self._daemon_location = "pacemaker-remoted" else: self._daemon_location = "pacemaker-execd" self._test_tool_location = "cts-exec-helper" # We additionally need to keep track of a stonith process. self._stonith_process = None def _new_cmd(self, cmd, args, exitcode, **kwargs): """Add a command to be executed as part of this test.""" if self.verbose and cmd == self._test_tool_location: args += " -V " if (cmd == self._test_tool_location) and self.tls: args += " -S " kwargs["validate"] = False kwargs["check_rng"] = False kwargs["check_stderr"] = False Test._new_cmd(self, cmd, args, exitcode, **kwargs) def _kill_daemons(self): killall([ "pacemaker-fenced", "lt-pacemaker-fenced", "pacemaker-execd", "lt-pacemaker-execd", "cts-exec-helper", "lt-cts-exec-helper", "pacemaker-remoted", ]) def _start_daemons(self): if not self.tls: # pylint: disable=consider-using-with self._stonith_process = subprocess.Popen(["pacemaker-fenced", "-s"]) cmd = [self._daemon_location, "-l", self.logpath] if self.verbose: cmd += ["-V"] # pylint: disable=consider-using-with self._daemon_process = subprocess.Popen(cmd) def clean_environment(self): """Clean up the host after running a test.""" if self._daemon_process: self._daemon_process.terminate() self._daemon_process.wait() if self.verbose: print("Daemon Output Start") with open(self.logpath, "rt", errors="replace", encoding="utf-8") as logfile: for line in logfile: print(line.strip()) print("Daemon Output End") if self._stonith_process: self._stonith_process.terminate() self._stonith_process.wait() self._daemon_process = None self._stonith_process = None def add_cmd(self, cmd=None, **kwargs): """Add a cts-exec-helper command to be executed as part of this test.""" if cmd is None: cmd = self._test_tool_location self._new_cmd(cmd, kwargs.pop("args"), ExitStatus.OK, **kwargs) def add_cmd_and_kill(self, cmd=None, **kwargs): """Add a cts-exec-helper command and system command to be executed as part of this test.""" if cmd is None: cmd = self._test_tool_location self._new_cmd(cmd, kwargs.pop("args"), ExitStatus.OK, **kwargs) def add_cmd_check_stdout(self, cmd=None, **kwargs): """Add a command with expected output to be executed as part of this test.""" if cmd is None: cmd = self._test_tool_location self._new_cmd(cmd, kwargs.pop("args"), ExitStatus.OK, **kwargs) def add_cmd_expected_fail(self, cmd=None, **kwargs): """Add a cts-exec-helper command to be executed as part of this test and expected to fail.""" if cmd is None: cmd = self._test_tool_location self._new_cmd(cmd, kwargs.pop("args"), kwargs.pop("expected_exitcode", ExitStatus.ERROR), **kwargs) def add_sys_cmd(self, cmd, args): """Add a simple command to be executed as part of this test.""" self._new_cmd(cmd, args, ExitStatus.OK) def run(self): """Execute this test.""" 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 Test.run(self) class ExecTests(Tests): """Collection of all pacemaker-execd regression tests.""" def __init__(self, **kwargs): """ Create a new ExecTests instance. Keyword arguments: tls -- Enable pacemaker-remoted. """ Tests.__init__(self, **kwargs) self.tls = kwargs.get("tls", False) self._action_timeout = " -t 9000 " self._installed_files = [] self._rsc_classes = self._setup_rsc_classes() print("Testing resource classes %r" % 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 _setup_rsc_classes(self): """Determine which resource classes are supported.""" classes = stdout_from_command(["crm_resource", "--list-standards"]) # Strip trailing empty line classes = classes[:-1] if self.tls: classes.remove("stonith") if "nagios" in classes: classes.remove("nagios") if "systemd" in 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. # pylint: disable=import-outside-toplevel,unused-import import systemd.daemon except ImportError: print("Python systemd bindings not found.") print("The tests for systemd class are not going to be run.") classes.remove("systemd") return classes def new_test(self, name, description): """Create a named test.""" test = ExecTest(name, description, verbose=self.verbose, tls=self.tls, timeout=self.timeout, force_wait=self.force_wait, logdir=self.logdir) self._tests.append(test) return test def setup_environment(self): """Prepare the host before executing any tests.""" if BuildOptions.REMOTE_ENABLED: os.system("service pacemaker_remote stop") self.cleanup_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 # pylint: disable=protected-access if os.path.exists("%s/cts/cts-exec.in" % BuildOptions._BUILD_DIR): if not os.path.exists("%s/pacemaker" % BuildOptions.OCF_RA_INSTALL_DIR): # @TODO remember which components were created and remove them os.makedirs("%s/pacemaker" % BuildOptions.OCF_RA_INSTALL_DIR, 0o755) for agent in ["Dummy", "Stateful", "ping"]: agent_source = "%s/extra/resources/%s" % (BuildOptions._BUILD_DIR, agent) agent_dest = "%s/pacemaker/%s" % (BuildOptions.OCF_RA_INSTALL_DIR, 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_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_cmd_str(self, rsc, ty): """Construct a command string for the given resource and type.""" return "%s %s" % (self._common_cmds["%s_%s_line" % (rsc, ty)], self._common_cmds["%s_%s_event" % (rsc, ty)]) 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(args=self._build_cmd_str(rsc, "reg")) test.add_cmd(args=self._build_cmd_str(rsc, "unreg")) # 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(args=self._build_cmd_str(rsc, "reg")) test.add_cmd(args=self._build_cmd_str(rsc, "start")) test.add_cmd(args=self._build_cmd_str(rsc, "stop")) test.add_cmd(args=self._build_cmd_str(rsc, "unreg")) # 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(args=self._build_cmd_str(rsc, "reg")) test.add_cmd(args=self._build_cmd_str(rsc, "start")) test.add_cmd(args=self._build_cmd_str(rsc, "monitor")) # If this fails, that means the monitor may not be getting rescheduled test.add_cmd(args=common_cmds["%s_monitor_event" % rsc]) # If this fails, that means the monitor may not be getting rescheduled test.add_cmd(args=common_cmds["%s_monitor_event" % rsc]) test.add_cmd(args=self._build_cmd_str(rsc, "cancel")) # If this happens the monitor did not actually cancel correctly test.add_cmd_expected_fail(args=common_cmds["%s_monitor_event" % rsc], expected_exitcode=ExitStatus.TIMEOUT) # If this happens the monitor did not actually cancel correctly test.add_cmd_expected_fail(args=common_cmds["%s_monitor_event" % rsc], expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args=self._build_cmd_str(rsc, "stop")) test.add_cmd(args=self._build_cmd_str(rsc, "unreg")) # 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(args=self._build_cmd_str(rsc, "reg")) test.add_cmd(args=self._build_cmd_str(rsc, "start")) test.add_cmd(args=self._build_cmd_str(rsc, "monitor")) # If this fails, that means the monitor may not be getting rescheduled test.add_cmd(args=common_cmds["%s_monitor_event" % rsc]) # If this fails, that means the monitor may not be getting rescheduled test.add_cmd(args=common_cmds["%s_monitor_event" % rsc]) # Add the duplicate monitors test.add_cmd(args=self._build_cmd_str(rsc, "monitor")) test.add_cmd(args=self._build_cmd_str(rsc, "monitor")) test.add_cmd(args=self._build_cmd_str(rsc, "monitor")) test.add_cmd(args=self._build_cmd_str(rsc, "monitor")) # verify we still get update events # If this fails, that means the monitor may not be getting rescheduled test.add_cmd(args=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(args=self._build_cmd_str(rsc, "cancel")) # If this happens the monitor did not actually cancel correctly test.add_cmd_expected_fail(args=common_cmds["%s_monitor_event" % rsc], expected_exitcode=ExitStatus.TIMEOUT) # If this happens the monitor did not actually cancel correctly test.add_cmd_expected_fail(args=common_cmds["%s_monitor_event" % rsc], expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args=self._build_cmd_str(rsc, "stop")) test.add_cmd(args=self._build_cmd_str(rsc, "unreg")) # 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(args=self._build_cmd_str(rsc, "reg")) test.add_cmd(args=self._build_cmd_str(rsc, "start")) test.add_cmd(args=self._build_cmd_str(rsc, "monitor")) # If this fails, that means the monitor may not be getting rescheduled test.add_cmd(args=common_cmds["%s_monitor_event" % rsc]) # If this fails, that means the monitor may not be getting rescheduled test.add_cmd(args=common_cmds["%s_monitor_event" % rsc]) test.add_cmd(args=self._build_cmd_str(rsc, "stop")) # If this happens the monitor did not actually cancel correctly test.add_cmd_expected_fail(args=common_cmds["%s_monitor_event" % rsc], expected_exitcode=ExitStatus.TIMEOUT) # If this happens the monitor did not actually cancel correctly test.add_cmd_expected_fail(args=common_cmds["%s_monitor_event" % rsc], expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args=self._build_cmd_str(rsc, "unreg")) 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(args=self._build_cmd_str(rsc, "reg")) for rsc in self._rsc_classes: test.add_cmd(args=self._build_cmd_str(rsc, "start")) for rsc in self._rsc_classes: test.add_cmd(args=self._build_cmd_str(rsc, "monitor")) for rsc in self._rsc_classes: # If this fails, that means the monitor is not being rescheduled test.add_cmd(args=common_cmds["%s_monitor_event" % rsc]) for rsc in self._rsc_classes: test.add_cmd(args=self._build_cmd_str(rsc, "cancel")) for rsc in self._rsc_classes: test.add_cmd(args=self._build_cmd_str(rsc, "stop")) for rsc in self._rsc_classes: test.add_cmd(args=self._build_cmd_str(rsc, "unreg")) 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(args='-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(args='-c exec -r test_rsc -a start -k op_sleep -v 5 -t 1000 -w') test.add_cmd(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:error op_status:Timed Out" ' + self._action_timeout) test.add_cmd(args='-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(args='-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(args='-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"') # -t must be less than self._action_timeout test.add_cmd(args='-c exec -r test_rsc -a start -k monitor_delay -v 30 -t 1000 -w') test.add_cmd(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:error op_status:Timed Out" ' + self._action_timeout) test.add_cmd(args='-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(args='-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 test = self.new_test("stonith_component_fail", "Kill stonith component after pacemaker-execd connects") test.add_cmd(args=self._build_cmd_str("stonith", "reg")) test.add_cmd(args=self._build_cmd_str("stonith", "start")) test.add_cmd(args='-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(args='-l "NEW_EVENT event_type:exec_complete rsc_id:stonith_test_rsc action:monitor rc:error op_status:error" -t 15000', kill="killall -9 -q pacemaker-fenced lt-pacemaker-fenced") test.add_cmd(args=self._build_cmd_str("stonith", "unreg")) # monitor fail for ocf resources test = self.new_test("monitor_fail_ocf", "Force ocf monitor to fail, verify failure is reported.") test.add_cmd(args='-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(args='-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(args='-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(args='-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(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete"' + self._action_timeout) test.add_cmd(args='-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(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete" ' + self._action_timeout, kill="rm -f %s/run/Dummy-test_rsc.state" % BuildOptions.LOCAL_STATE_DIR) test.add_cmd(args='-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_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args='-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(args='-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(args='-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(args='-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_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd_and_kill(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete"' + self._action_timeout, kill='rm -f %s/run/Dummy-test_rsc.state' % BuildOptions.LOCAL_STATE_DIR) test.add_cmd(args='-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_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args='-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(args='-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(args='-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(args='-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(args='-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(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ' + self._action_timeout) test.add_cmd(args='-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(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete"' + self._action_timeout, kill="pkill -9 -f pacemaker-cts-dummyd") test.add_cmd(args='-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_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args='-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(args='-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(args='-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(args='-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(args='-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(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ' + self._action_timeout) test.add_cmd(args='-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(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete"' + self._action_timeout, kill='killall -9 -q dd') test.add_cmd(args='-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_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:not running op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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_cmd_expected_fail(args='-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_cmd_expected_fail(args='-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(args='-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_cmd_expected_fail(args='-c exec -r test_rsc -a start ' + self._action_timeout + '-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:error op_status:complete" ') test.add_cmd_expected_fail(args='-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_expected_fail(args='-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_cmd_expected_fail(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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_cmd_expected_fail(args='-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_cmd_expected_fail(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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(args='-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 %s" % BuildOptions.RSC_TMP_DIR) test.add_cmd(args='-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(args='-c exec -r test_rsc -a start -t 4000') test.add_sys_cmd("ls", "-al %s" % BuildOptions.RSC_TMP_DIR) test.add_sys_cmd("ls", "%s/Dummy-test_rsc.state" % BuildOptions.RSC_TMP_DIR) test.add_cmd(args='-c exec -r test_rsc -a stop -t 4000') test.add_cmd(args='-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(args='-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(args='-c exec -r test_rsc -s 6000 -a start -w -t 6000') test.add_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete" -t 2000', expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete" -t 6000') test.add_cmd(args='-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(args='-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(args='-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(args='-c exec -r test_rsc -s 5000 -a start -w -t 4000') test.add_cmd(args='-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_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:start rc:ok op_status:complete" -t 5000', expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args='-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(args='-c register_rsc -r rsc1 -C systemd -T pacemaker-cts-dummyd@3 ' + self._action_timeout) test.add_cmd(args='-c get_rsc_info -r rsc1 ') test.add_cmd(args='-c unregister_rsc -r rsc1 ' + self._action_timeout) test.add_cmd_expected_fail(args='-c get_rsc_info -r rsc1 ') if "upstart" in self._rsc_classes: test.add_cmd(args='-c register_rsc -r rsc1 -C upstart -T pacemaker-cts-dummyd ' + self._action_timeout) test.add_cmd(args='-c get_rsc_info -r rsc1 ') test.add_cmd(args='-c unregister_rsc -r rsc1 ' + self._action_timeout) test.add_cmd_expected_fail(args='-c get_rsc_info -r rsc1 ') test.add_cmd(args='-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker ' + self._action_timeout) test.add_cmd(args='-c get_rsc_info -r rsc2 ') test.add_cmd(args='-c unregister_rsc -r rsc2 ' + self._action_timeout) test.add_cmd_expected_fail(args='-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(args='-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker ' + self._action_timeout) test.add_cmd_check_stdout(args="-c get_rsc_info -r rsc2 ", - match="id:rsc2 class:ocf provider:pacemaker type:Dummy") + stdout_match="id:rsc2 class:ocf provider:pacemaker type:Dummy") test.add_cmd(args='-c register_rsc -r rsc2 -C ocf -T Dummy -P pacemaker ' + self._action_timeout) test.add_cmd_check_stdout(args="-c get_rsc_info -r rsc2 ", - match="id:rsc2 class:ocf provider:pacemaker type:Dummy") + stdout_match="id:rsc2 class:ocf provider:pacemaker type:Dummy") test.add_cmd(args='-c register_rsc -r rsc2 -C ocf -T Stateful -P pacemaker ' + self._action_timeout) test.add_cmd_check_stdout(args="-c get_rsc_info -r rsc2 ", - match="id:rsc2 class:ocf provider:pacemaker type:Stateful") + stdout_match="id:rsc2 class:ocf provider:pacemaker type:Stateful") test.add_cmd(args='-c unregister_rsc -r rsc2 ' + self._action_timeout) test.add_cmd_expected_fail(args='-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(args='-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(args='-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(args='-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_cmd_expected_fail(args='-l "NEW_EVENT event_type:exec_complete rsc_id:test_rsc action:monitor rc:ok op_status:complete" ' + self._action_timeout, expected_exitcode=ExitStatus.TIMEOUT) test.add_cmd(args='-c cancel -r test_rsc -a monitor -i 1s -t 6000 ') test.add_cmd(args='-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(args="-c metadata -C ocf -P pacemaker -T Dummy", - match="resource-agent name=\"Dummy\"") + stdout_match="resource-agent name=\"Dummy\"") test.add_cmd(args="-c metadata -C ocf -P pacemaker -T Stateful") test.add_cmd_expected_fail(args="-c metadata -P pacemaker -T Stateful") test.add_cmd_expected_fail(args="-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(args="-c metadata -C lsb -T LSBDummy", - match="resource-agent name='LSBDummy'") + stdout_match="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(args="-c metadata -C stonith -P pacemaker -T fence_dummy", - match="resource-agent name=\"fence_dummy\"") + stdout_match="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(args="-c metadata -C systemd -T pacemaker-cts-dummyd@", - match="resource-agent name=\"pacemaker-cts-dummyd@\"") + stdout_match="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(args="-c metadata -C upstart -T pacemaker-cts-dummyd", - match="resource-agent name=\"pacemaker-cts-dummyd\"") + stdout_match="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(args="-c list_ocf_providers ", match="pacemaker") - test.add_cmd_check_stdout(args="-c list_ocf_providers -T ping", match="pacemaker") + test.add_cmd_check_stdout(args="-c list_ocf_providers ", stdout_match="pacemaker") + test.add_cmd_check_stdout(args="-c list_ocf_providers -T ping", stdout_match="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(args="-c list_agents ", match="Stateful") # ocf - test.add_cmd_check_stdout(args="-c list_agents -C ocf", match="Stateful") - test.add_cmd_check_stdout(args="-c list_agents -C lsb", match="", stdout_no_match="Stateful") # should not exist - test.add_cmd_check_stdout(args="-c list_agents -C service", match="", stdout_no_match="Stateful") # should not exist - test.add_cmd_check_stdout(args="-c list_agents ", match="LSBDummy") # init.d - test.add_cmd_check_stdout(args="-c list_agents -C lsb", match="LSBDummy") - test.add_cmd_check_stdout(args="-c list_agents -C service", match="LSBDummy") - test.add_cmd_check_stdout(args="-c list_agents -C ocf", match="", stdout_no_match="pacemaker-cts-dummyd@") # should not exist - - test.add_cmd_check_stdout(args="-c list_agents -C ocf", match="", stdout_no_match="pacemaker-cts-dummyd@") # should not exist - test.add_cmd_check_stdout(args="-c list_agents -C lsb", match="", stdout_no_match="fence_dummy") # should not exist - test.add_cmd_check_stdout(args="-c list_agents -C service", match="", stdout_no_match="fence_dummy") # should not exist - test.add_cmd_check_stdout(args="-c list_agents -C ocf", match="", stdout_no_match="fence_dummy") # should not exist + test.add_cmd_check_stdout(args="-c list_agents ", stdout_match="Stateful") # ocf + test.add_cmd_check_stdout(args="-c list_agents -C ocf", stdout_match="Stateful") + test.add_cmd_check_stdout(args="-c list_agents -C lsb", stdout_match="", stdout_no_match="Stateful") # should not exist + test.add_cmd_check_stdout(args="-c list_agents -C service", stdout_match="", stdout_no_match="Stateful") # should not exist + test.add_cmd_check_stdout(args="-c list_agents ", stdout_match="LSBDummy") # init.d + test.add_cmd_check_stdout(args="-c list_agents -C lsb", stdout_match="LSBDummy") + test.add_cmd_check_stdout(args="-c list_agents -C service", stdout_match="LSBDummy") + test.add_cmd_check_stdout(args="-c list_agents -C ocf", stdout_match="", stdout_no_match="pacemaker-cts-dummyd@") # should not exist + + test.add_cmd_check_stdout(args="-c list_agents -C ocf", stdout_match="", stdout_no_match="pacemaker-cts-dummyd@") # should not exist + test.add_cmd_check_stdout(args="-c list_agents -C lsb", stdout_match="", stdout_no_match="fence_dummy") # should not exist + test.add_cmd_check_stdout(args="-c list_agents -C service", stdout_match="", stdout_no_match="fence_dummy") # should not exist + test.add_cmd_check_stdout(args="-c list_agents -C ocf", stdout_match="", stdout_no_match="fence_dummy") # should not exist if "systemd" in self._rsc_classes: - test.add_cmd_check_stdout(args="-c list_agents ", match="pacemaker-cts-dummyd@") # systemd - test.add_cmd_check_stdout(args="-c list_agents -C service", match="LSBDummy") - test.add_cmd_check_stdout(args="-c list_agents -C systemd", match="", stdout_no_match="Stateful") # should not exist - test.add_cmd_check_stdout(args="-c list_agents -C systemd", match="pacemaker-cts-dummyd@") - test.add_cmd_check_stdout(args="-c list_agents -C systemd", match="", stdout_no_match="fence_dummy") # should not exist + test.add_cmd_check_stdout(args="-c list_agents ", stdout_match="pacemaker-cts-dummyd@") # systemd + test.add_cmd_check_stdout(args="-c list_agents -C service", stdout_match="LSBDummy") + test.add_cmd_check_stdout(args="-c list_agents -C systemd", stdout_match="", stdout_no_match="Stateful") # should not exist + test.add_cmd_check_stdout(args="-c list_agents -C systemd", stdout_match="pacemaker-cts-dummyd@") + test.add_cmd_check_stdout(args="-c list_agents -C systemd", stdout_match="", stdout_no_match="fence_dummy") # should not exist if "upstart" in self._rsc_classes: - test.add_cmd_check_stdout(args="-c list_agents ", match="pacemaker-cts-dummyd") # upstart - test.add_cmd_check_stdout(args="-c list_agents -C service", match="LSBDummy") - test.add_cmd_check_stdout(args="-c list_agents -C upstart", match="", stdout_no_match="Stateful") # should not exist - test.add_cmd_check_stdout(args="-c list_agents -C upstart", match="pacemaker-cts-dummyd") - test.add_cmd_check_stdout(args="-c list_agents -C upstart", match="", stdout_no_match="fence_dummy") # should not exist + test.add_cmd_check_stdout(args="-c list_agents ", stdout_match="pacemaker-cts-dummyd") # upstart + test.add_cmd_check_stdout(args="-c list_agents -C service", stdout_match="LSBDummy") + test.add_cmd_check_stdout(args="-c list_agents -C upstart", stdout_match="", stdout_no_match="Stateful") # should not exist + test.add_cmd_check_stdout(args="-c list_agents -C upstart", stdout_match="pacemaker-cts-dummyd") + test.add_cmd_check_stdout(args="-c list_agents -C upstart", stdout_match="", stdout_no_match="fence_dummy") # should not exist if "stonith" in self._rsc_classes: - test.add_cmd_check_stdout(args="-c list_agents -C stonith", match="fence_dummy") # stonith - test.add_cmd_check_stdout(args="-c list_agents -C stonith", match="", # should not exist + test.add_cmd_check_stdout(args="-c list_agents -C stonith", stdout_match="fence_dummy") # stonith + test.add_cmd_check_stdout(args="-c list_agents -C stonith", stdout_match="", # should not exist stdout_no_match="pacemaker-cts-dummyd@") - test.add_cmd_check_stdout(args="-c list_agents -C stonith", match="", stdout_no_match="Stateful") # should not exist - test.add_cmd_check_stdout(args="-c list_agents ", match="fence_dummy") + test.add_cmd_check_stdout(args="-c list_agents -C stonith", stdout_match="", stdout_no_match="Stateful") # should not exist + test.add_cmd_check_stdout(args="-c list_agents ", stdout_match="fence_dummy") def build_options(): """Handle command line arguments.""" parser = argparse.ArgumentParser(formatter_class=argparse.RawDescriptionHelpFormatter, description="Run pacemaker-execd regression tests", epilog="Example: Run only the test 'start_stop'\n" "\t " + sys.argv[0] + " --run-only start_stop\n\n" "Example: Run only the tests with the string 'systemd' present in them\n" "\t " + sys.argv[0] + " --run-only-pattern systemd") parser.add_argument("-l", "--list-tests", action="store_true", help="Print out all registered tests") parser.add_argument("-p", "--run-only-pattern", metavar='PATTERN', help="Run only tests matching the given pattern") parser.add_argument("-r", "--run-only", metavar='TEST', help="Run a specific test") parser.add_argument("-t", "--timeout", type=float, default=2, help="Up to how many seconds each test case waits for the daemon to " "be initialized. Defaults to 2. The value 0 means no limit.") parser.add_argument("-w", "--force-wait", action="store_true", help="Each test case waits the default/specified --timeout for the " "daemon without tracking the log") if BuildOptions.REMOTE_ENABLED: parser.add_argument("-R", "--pacemaker-remote", action="store_true", help="Test pacemaker-remoted binary instead of pacemaker-execd") parser.add_argument("-V", "--verbose", action="store_true", help="Verbose output") args = parser.parse_args() return args def main(): """Run pacemaker-execd regression tests as specified by arguments.""" update_path() # Ensure all command output is in portable locale for comparison os.environ['LC_ALL'] = "C" opts = build_options() if opts.pacemaker_remote: daemon_name = "pacemaker-remoted" else: daemon_name = "pacemaker-execd" exit_if_proc_running(daemon_name) # Create a temporary directory for log files (the directory will # automatically be erased when done) with tempfile.TemporaryDirectory(prefix="cts-exec-") as logdir: tests = ExecTests(verbose=opts.verbose, tls=opts.pacemaker_remote, timeout=opts.timeout, force_wait=opts.force_wait, logdir=logdir) tests.build_generic_tests() tests.build_multi_rsc_tests() tests.build_negative_tests() tests.build_custom_tests() tests.build_stress_tests() if opts.list_tests: tests.print_list() sys.exit(ExitStatus.OK) print("Starting ...") tests.setup_environment() if opts.run_only_pattern: tests.run_tests_matching(opts.run_only_pattern) tests.print_results() elif opts.run_only: tests.run_single(opts.run_only) tests.print_results() else: tests.run_tests() tests.print_results() tests.cleanup_environment() tests.exit() if __name__ == "__main__": main() diff --git a/python/pacemaker/_cts/test.py b/python/pacemaker/_cts/test.py index a6b5e96a7d..eaf6faf218 100644 --- a/python/pacemaker/_cts/test.py +++ b/python/pacemaker/_cts/test.py @@ -1,638 +1,638 @@ """ A module providing base classes. These classes are used for defining regression tests and groups of regression tests. Everything exported here should be considered an abstract class that needs to be subclassed in order to do anything useful. Various functions will raise NotImplementedError if not overridden by a subclass. """ __copyright__ = "Copyright 2009-2024 the Pacemaker project contributors" __license__ = "GNU General Public License version 2 or later (GPLv2+)" __all__ = ["Test", "Tests"] import io import os import re import shlex import signal import subprocess import sys import time from pacemaker._cts.errors import ExitCodeError, OutputFoundError, OutputNotFoundError, XmlValidationError from pacemaker._cts.process import pipe_communicate from pacemaker.buildoptions import BuildOptions from pacemaker.exitstatus import ExitStatus def find_validator(rng_file): """ Return the command line used to validate XML output. If no validator is found, return None. """ if os.access("/usr/bin/xmllint", os.X_OK): if rng_file is None: return ["xmllint", "-"] return ["xmllint", "--relaxng", rng_file, "-"] return None def rng_directory(): """Return the directory containing RNG schema files.""" if "PCMK_schema_directory" in os.environ: return os.environ["PCMK_schema_directory"] if os.path.exists("%s/cts-fencing.in" % sys.path[0]): return "xml" return BuildOptions.SCHEMA_DIR class Pattern: """A class for checking log files for a given pattern.""" def __init__(self, pat, negative=False, regex=False): """ Create a new Pattern instance. Arguments: pat -- The string to search for negative -- If True, pat must not be found in any input regex -- If True, pat is a regex and not a substring """ self._pat = pat self.negative = negative self.regex = regex def __str__(self): return self._pat def match(self, line): """Return True if this pattern is found in the given line.""" if self.regex: return re.search(self._pat, line) is not None return self._pat in line class Test: """ The base class for a single regression test. A single regression test may still run multiple commands as part of its execution. """ def __init__(self, name, description, **kwargs): """ Create a new Test instance. This method must be provided by all subclasses, which must call Test.__init__ first. Arguments: description -- A user-readable description of the test, helpful in identifying what test is running or has failed. name -- The name of the test. Command line tools use this attribute to allow running only tests with the exact name, or tests whose name matches a given pattern. This should be unique among all tests. Keyword arguments: force_wait -- logdir -- The base directory under which to create a directory to store output and temporary data. timeout -- How long to wait for the test to complete. verbose -- Whether to print additional information, including verbose command output and daemon log files. """ self.description = description self.executed = False self.name = name self.force_wait = kwargs.get("force_wait", False) self.logdir = kwargs.get("logdir", "/tmp") self.timeout = kwargs.get("timeout", 2) self.verbose = kwargs.get("verbose", False) self._cmds = [] self._patterns = [] self._daemon_location = None self._daemon_output = "" self._daemon_process = None self._result_exitcode = ExitStatus.OK self._result_txt = "" # # PROPERTIES # @property def exitcode(self): """ Return the final exitcode of the Test. If all commands pass, this property will be ExitStatus.OK. Otherwise, this property will be the exitcode of the first command to fail. """ return self._result_exitcode @exitcode.setter def exitcode(self, value): self._result_exitcode = value @property def logpath(self): """ Return the path to the log for whatever daemon is being tested. Note that this requires all subclasses to set self._daemon_location before accessing this property or an exception will be raised. """ return os.path.join(self.logdir, "%s.log" % self._daemon_location) # # PRIVATE METHODS # def _kill_daemons(self): """Kill any running daemons in preparation for executing the test.""" raise NotImplementedError("_kill_daemons not provided by subclass") def _match_log_patterns(self): """ Check test output for expected patterns. Set self.exitcode and self._result_txt as appropriate. Not all subclass will need to do this. """ if len(self._patterns) == 0: return n_failed_matches = 0 n_negative_matches = 0 output = self._daemon_output.split("\n") for pat in self._patterns: positive_match = False for line in output: if pat.match(line): if pat.negative: n_negative_matches += 1 if self.verbose: print("This pattern should not have matched = '%s" % pat) break positive_match = True break if not pat.negative and not positive_match: n_failed_matches += 1 print("Pattern Not Matched = '%s'" % pat) if n_failed_matches > 0 or n_negative_matches > 0: msg = "FAILURE - '%s' failed. %d patterns out of %d not matched. %d negative matches." self._result_txt = msg % (self.name, n_failed_matches, len(self._patterns), n_negative_matches) self.exitcode = ExitStatus.ERROR def _new_cmd(self, cmd, args, exitcode, **kwargs): """ Add a command to be executed as part of this test. Arguments: cmd -- The program to run. args -- Commands line arguments to pass to cmd, as a string. exitcode -- The expected exit code of cmd. This can be used to run a command that is expected to fail. Keyword arguments: stdout_match -- If not None, a string that is expected to be present in the stdout of cmd. This can be a regular expression. no_wait -- Do not wait for cmd to complete. stdout_no_match -- If not None, a string that is expected to be missing in the stdout of cmd. This can be a regualr expression. kill -- A command to be run after cmd, typically in order to kill a failed process. This should be the entire command line including arguments as a single string. validate -- If True, the output of cmd will be passed to xmllint for validation. If validation fails, XmlValidationError will be raised. check_rng -- If True and validate is True, command output will additionally be checked against the api-result.rng file. check_stderr -- If True, the stderr of cmd will be included in output. env -- If not None, variables to set in the environment """ if cmd is None: raise ValueError("cmd cannot be None") self._cmds.append( { "args": args, "check_rng": kwargs.get("check_rng", True), "check_stderr": kwargs.get("check_stderr", True), "cmd": cmd, "expected_exitcode": exitcode, "kill": kwargs.get("kill"), "no_wait": kwargs.get("no_wait", False), "stdout_match": kwargs.get("stdout_match"), "stdout_no_match": kwargs.get("stdout_no_match"), "validate": kwargs.get("validate", True), "env": kwargs.get("env"), } ) def _start_daemons(self): """Start any necessary daemons in preparation for executing the test.""" raise NotImplementedError("_start_daemons not provided by subclass") # # PUBLIC METHODS # def add_cmd(self, cmd=None, **kwargs): """ Add a command to be executed as part of this test. Arguments: cmd -- The program to run. Keyword arguments: args -- Commands line arguments to pass to cmd, as a string. check_rng -- If True and validate is True, command output will additionally be checked against the api-result.rng file. check_stderr -- If True, the stderr of cmd will be included in output. env -- If not None, variables to set in the environment expected_exitcode -- The expected exit code of cmd. This can be used to run a command that is expected to fail. kill -- A command to be run after cmd, typically in order to kill a failed process. This should be the entire command line including arguments as a single string. no_wait -- Do not wait for cmd to complete. stdout_match -- If not None, a string that is expected to be present in the stdout of cmd. This can be a regular expression. stdout_no_match -- If not None, a string that is expected to be missing in the stdout of cmd. This can be a regular expression. validate -- If True, the output of cmd will be passed to xmllint for validation. If validation fails, XmlValidationError will be raised. """ if cmd is None: raise ValueError("cmd cannot be None") self._cmds.append( { "args": kwargs.get("args", ""), "check_rng": kwargs.get("check_rng", True), "check_stderr": kwargs.get("check_stderr", True), "cmd": cmd, "expected_exitcode": kwargs.get("expected_exitcode", ExitStatus.OK), "kill": kwargs.get("kill"), "no_wait": kwargs.get("no_wait", False), "stdout_match": kwargs.get("stdout_match"), "stdout_no_match": kwargs.get("stdout_no_match"), "validate": kwargs.get("validate", True), "env": kwargs.get("env"), } ) def add_cmd_check_stdout(self, cmd=None, **kwargs): """Add a simple command with expected output to be executed as part of this test.""" self._new_cmd(cmd, kwargs.pop("args", ""), ExitStatus.OK, - stdout_match=kwargs.get("match"), + stdout_match=kwargs.get("stdout_match"), stdout_no_match=kwargs.get("stdout_no_match"), env=kwargs.get("env")) def add_cmd_expected_fail(self, cmd=None, **kwargs): """Add a command that is expected to fail to be executed as part of this test.""" self._new_cmd(cmd, kwargs.pop("args", ""), kwargs.get("expected_exitcode", ExitStatus.ERROR)) def add_cmd_no_wait(self, cmd=None, **kwargs): """Add a simple command to be executed (without waiting) as part of this test.""" self._new_cmd(cmd, kwargs.pop("args", ""), ExitStatus.OK, no_wait=kwargs.get("no_wait", True)) def add_log_pattern(self, pattern, negative=False, regex=False): """Add a pattern that should appear in the test's logs.""" self._patterns.append(Pattern(pattern, negative=negative, regex=regex)) def _signal_dict(self): """Return a dictionary mapping signal numbers to their names.""" # FIXME: When we support python >= 3.5, this function can be replaced with: # signal.Signals(self.daemon_process.returncode).name return { getattr(signal, _signame): _signame for _signame in dir(signal) if _signame.startswith("SIG") and not _signame.startswith("SIG_") } def clean_environment(self): """Clean up the host after executing a test.""" if self._daemon_process: if self._daemon_process.poll() is None: self._daemon_process.terminate() self._daemon_process.wait() else: rc = self._daemon_process.returncode signame = self._signal_dict().get(-rc, "RET=%s" % rc) msg = "FAILURE - '%s' failed. %s abnormally exited during test (%s)." self._result_txt = msg % (self.name, self._daemon_location, signame) self.exitcode = ExitStatus.ERROR self._daemon_process = None self._daemon_output = "" # the default for utf-8 encoding would error out if e.g. memory corruption # makes fenced output any kind of 8 bit value - while still interesting # for debugging and we'd still like the regression-test to go over the # full set of test-cases with open(self.logpath, 'rt', encoding="ISO-8859-1") as logfile: for line in logfile.readlines(): self._daemon_output += line if self.verbose: print("Daemon Output Start") print(self._daemon_output) print("Daemon Output End") def print_result(self, filler): """Print the result of the last test execution.""" print("%s%s" % (filler, self._result_txt)) def run(self): """Execute this test.""" i = 1 self.start_environment() if self.verbose: print("\n--- START TEST - %s" % self.name) self._result_txt = "SUCCESS - '%s'" % (self.name) self.exitcode = ExitStatus.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_no_match'], e)) self.set_error(i, cmd) break except XmlValidationError as e: print("Step %d FAILED - xmllint failed: %s" % (i, e)) self.set_error(i, cmd) break if self.verbose: print("Step %d SUCCESS" % (i)) i += 1 self.clean_environment() if self.exitcode == ExitStatus.OK: self._match_log_patterns() print(self._result_txt) if self.verbose: print("--- END TEST - %s\n" % self.name) self.executed = True 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: %s" % " ".join(cmd)) # FIXME: Using "with" here breaks fencing merge tests. # pylint: disable=consider-using-with if args['env']: new_env = os.environ.copy() new_env.update(args['env']) test = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, env=new_env) else: test = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE) if args['kill']: if self.verbose: print("Also running: %s" % 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 not args['no_wait']: test.wait() else: return ExitStatus.OK output = pipe_communicate(test, check_stderr=args['check_stderr']) if self.verbose: print(output) if test.returncode != args['expected_exitcode']: raise ExitCodeError(test.returncode) if args['stdout_match'] is not None and \ re.search(args['stdout_match'], output) is None: raise OutputNotFoundError(output) if args['stdout_no_match'] is not None and \ re.search(args['stdout_no_match'], output) is not None: raise OutputFoundError(output) if args['validate']: if args['check_rng']: rng_file = "%s/api/api-result.rng" % rng_directory() else: rng_file = None cmd = find_validator(rng_file) if not cmd: raise XmlValidationError("Could not find validator for %s" % rng_file) if self.verbose: print("\nRunning: %s" % " ".join(cmd)) with subprocess.Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE) as validator: output = pipe_communicate(validator, check_stderr=True, stdin=output) if self.verbose: print(output) if validator.returncode != 0: raise XmlValidationError(output) return ExitStatus.OK 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.exitcode = ExitStatus.ERROR def start_environment(self): """Prepare the host for executing a test.""" if os.path.exists(self.logpath): os.remove(self.logpath) self._kill_daemons() self._start_daemons() logfile = None init_time = time.time() update_time = init_time while True: # FIXME: Eventually use 'with' here, which seems complicated given # everything happens in a loop. # pylint: disable=consider-using-with time.sleep(0.1) if not self.force_wait and logfile is None \ and os.path.exists(self.logpath): logfile = io.open(self.logpath, 'rt', encoding="ISO-8859-1") if not self.force_wait and logfile is not None: for line in logfile.readlines(): if "successfully started" in line: return now = time.time() if self.timeout > 0 and (now - init_time) >= self.timeout: if not self.force_wait: print("\tDaemon %s doesn't seem to have been initialized within %fs." "\n\tConsider specifying a longer '--timeout' value." % (self._daemon_location, self.timeout)) return if self.verbose and (now - update_time) >= 5: print("Waiting for %s to be initialized: %fs ..." % (self._daemon_location, now - init_time)) update_time = now class Tests: """The base class for a collection of regression tests.""" def __init__(self, **kwargs): """ Create a new Tests instance. This method must be provided by all subclasses, which must call Tests.__init__ first. Keywork arguments: force_wait -- logdir -- The base directory under which to create a directory to store output and temporary data. timeout -- How long to wait for the test to complete. verbose -- Whether to print additional information, including verbose command output and daemon log files. """ self.force_wait = kwargs.get("force_wait", False) self.logdir = kwargs.get("logdir", "/tmp") self.timeout = kwargs.get("timeout", 2) self.verbose = kwargs.get("verbose", False) self._tests = [] def exit(self): """Exit (with error status code if any test failed).""" for test in self._tests: if not test.executed: continue if test.exitcode != ExitStatus.OK: sys.exit(ExitStatus.ERROR) sys.exit(ExitStatus.OK) 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 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 not test.executed: continue if test.exitcode != ExitStatus.OK: failures += 1 test.print_result(" ") else: success += 1 if failures == 0: print(" None") print("\n--- TOTALS\n Pass:%d\n Fail:%d\n" % (success, failures)) 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(self): """Run all tests.""" for test in self._tests: test.run() 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()