def _get_local_services_status(
service_manager: ServiceManagerInterface,
) -> List[_ServiceStatus]:
service_def = [
# (service name, display even if not enabled nor running)
("corosync", True),
("pacemaker", True),
("pacemaker_remote", False),
("pcsd", True),
(get_sbd_service_name(service_manager), False),
]
service_status_list = []
for service, display_always in service_def:
try:
service_status_list.append(
_ServiceStatus(
service,
display_always,
service_manager.is_enabled(service),
service_manager.is_running(service),
)
)
except LibraryError:
pass
return service_status_list
def status_stonith_check():
# We should read the default value from pacemaker. However that may slow
# pcs down as we need to run 'pengine metadata' to get it.
stonith_enabled = True
stonith_devices = []
stonith_devices_id_action = []
stonith_devices_id_method_cycle = []
sbd_running = False
cib = utils.get_cib_dom()
for conf in cib.getElementsByTagName("configuration"):
for crm_config in conf.getElementsByTagName("crm_config"):
for nvpair in crm_config.getElementsByTagName("nvpair"):
if (nvpair.getAttribute("name") == "stonith-enabled"
and is_false(nvpair.getAttribute("value"))):
stonith_enabled = False
break
if not stonith_enabled:
break
for resource in conf.getElementsByTagName("primitive"):
if resource.getAttribute("class") == "stonith":
stonith_devices.append(resource)
for attribs in resource.getElementsByTagName(
"instance_attributes"):
for nvpair in attribs.getElementsByTagName("nvpair"):
if (nvpair.getAttribute("name") == "action"
and nvpair.getAttribute("value")):
stonith_devices_id_action.append(
resource.getAttribute("id"))
if (nvpair.getAttribute("name") == "method"
and nvpair.getAttribute("value") == "cycle"):
stonith_devices_id_method_cycle.append(
resource.getAttribute("id"))
if not utils.usefile:
# check if SBD daemon is running
try:
sbd_running = utils.is_service_running(utils.cmd_runner(),
get_sbd_service_name())
except LibraryError:
pass
if stonith_enabled and not stonith_devices and not sbd_running:
print("WARNING: no stonith devices and stonith-enabled is not false")
if stonith_devices_id_action:
print(
"WARNING: following stonith devices have the 'action' option set, "
"it is recommended to set {0} instead: {1}".format(
", ".join(
["'{0}'".format(x) for x in _STONITH_ACTION_REPLACED_BY]),
", ".join(sorted(stonith_devices_id_action))))
if stonith_devices_id_method_cycle:
print(
"WARNING: following stonith devices have the 'method' option set "
"to 'cycle' which is potentially dangerous, please consider using "
"'onoff': {0}".format(", ".join(
sorted(stonith_devices_id_method_cycle))))
def full_cluster_status_plaintext(
env: LibraryEnvironment,
hide_inactive_resources: bool = False,
verbose: bool = False,
) -> str:
"""
Return full cluster status as plaintext
env -- LibraryEnvironment
hide_inactive_resources -- if True, do not display non-running resources
verbose -- if True, display more info
"""
# pylint: disable=too-many-branches
# pylint: disable=too-many-locals
# pylint: disable=too-many-statements
# validation
if not env.is_cib_live and env.is_corosync_conf_live:
raise LibraryError(
ReportItem.error(
reports.messages.LiveEnvironmentNotConsistent(
[file_type_codes.CIB], [file_type_codes.COROSYNC_CONF],
)
)
)
if env.is_cib_live and not env.is_corosync_conf_live:
raise LibraryError(
ReportItem.error(
reports.messages.LiveEnvironmentNotConsistent(
[file_type_codes.COROSYNC_CONF], [file_type_codes.CIB],
)
)
)
# initialization
runner = env.cmd_runner()
report_processor = env.report_processor
live = env.is_cib_live and env.is_corosync_conf_live
is_sbd_running = False
# load status, cib, corosync.conf
status_text, warning_list = get_cluster_status_text(
runner, hide_inactive_resources, verbose
)
corosync_conf = None
# If we are live on a remote node, we have no corosync.conf.
# TODO Use the new file framework so the path is not exposed.
if not live or os.path.exists(settings.corosync_conf_file):
corosync_conf = env.get_corosync_conf()
cib = env.get_cib()
if verbose:
(
ticket_status_text,
ticket_status_stderr,
ticket_status_retval,
) = get_ticket_status_text(runner)
# get extra info if live
if live:
try:
is_sbd_running = is_service_running(runner, get_sbd_service_name())
except LibraryError:
pass
local_services_status = _get_local_services_status(runner)
if verbose and corosync_conf:
node_name_list, node_names_report_list = get_existing_nodes_names(
corosync_conf
)
report_processor.report_list(node_names_report_list)
node_reachability = _get_node_reachability(
env.get_node_target_factory(),
env.get_node_communicator(),
report_processor,
node_name_list,
)
# check stonith configuration
warning_list = list(warning_list)
warning_list.extend(_stonith_warnings(cib, is_sbd_running))
# put it all together
if report_processor.has_errors:
raise LibraryError()
cluster_name = (
corosync_conf.get_cluster_name()
if corosync_conf
else nvpair.get_value(
"cluster_property_set", get_crm_config(cib), "cluster-name", ""
)
)
parts = []
parts.append(f"Cluster name: {cluster_name}")
if warning_list:
parts.extend(["", "WARNINGS:"] + warning_list + [""])
parts.append(status_text)
if verbose:
parts.extend(["", "Tickets:"])
if ticket_status_retval != 0:
ticket_warning_parts = [
"WARNING: Unable to get information about tickets"
]
if ticket_status_stderr:
ticket_warning_parts.extend(
indent(ticket_status_stderr.splitlines())
)
parts.extend(indent(ticket_warning_parts))
else:
parts.extend(indent(ticket_status_text.splitlines()))
if live:
if verbose and corosync_conf:
parts.extend(["", "PCSD Status:"])
parts.extend(
indent(
_format_node_reachability(node_name_list, node_reachability)
)
)
parts.extend(["", "Daemon Status:"])
parts.extend(
indent(_format_local_services_status(local_services_status))
)
return "\n".join(parts)
def test_not_systemctl(self, mock_is_systemctl):
mock_is_systemctl.return_value = False
self.assertEqual("sbd_helper", lib_sbd.get_sbd_service_name())
mock_is_systemctl.assert_called_once_with()
def test_not_systemctl(self, mock_is_systemctl):
mock_is_systemctl.return_value = False
self.assertEqual("sbd_helper", lib_sbd.get_sbd_service_name())
mock_is_systemctl.assert_called_once_with()
def full_cluster_status_plaintext(
env: LibraryEnvironment,
hide_inactive_resources: bool = False,
verbose: bool = False,
) -> str:
"""
Return full cluster status as plaintext
env -- LibraryEnvironment
hide_inactive_resources -- if True, do not display non-running resources
verbose -- if True, display more info
"""
# pylint: disable=too-many-branches
# pylint: disable=too-many-locals
# validation
if not env.is_cib_live and env.is_corosync_conf_live:
raise LibraryError(
reports.live_environment_not_consistent(
[file_type_codes.CIB],
[file_type_codes.COROSYNC_CONF],
))
if env.is_cib_live and not env.is_corosync_conf_live:
raise LibraryError(
reports.live_environment_not_consistent(
[file_type_codes.COROSYNC_CONF],
[file_type_codes.CIB],
))
# initialization
runner = env.cmd_runner()
report_processor = SimpleReportProcessor(env.report_processor)
live = env.is_cib_live and env.is_corosync_conf_live
is_sbd_running = False
# load status, cib, corosync.conf
status_text, warning_list = get_cluster_status_text(
runner, hide_inactive_resources, verbose)
corosync_conf = env.get_corosync_conf()
cib = env.get_cib()
if verbose:
ticket_status_text, ticket_status_stderr, ticket_status_retval = (
get_ticket_status_text(runner))
# get extra info if live
if live:
try:
is_sbd_running = is_service_running(runner, get_sbd_service_name())
except LibraryError:
pass
local_services_status = _get_local_services_status(runner)
if verbose:
node_name_list, node_names_report_list = get_existing_nodes_names(
corosync_conf)
report_processor.report_list(node_names_report_list)
node_reachability = _get_node_reachability(
env.get_node_target_factory(),
env.get_node_communicator(),
report_processor,
node_name_list,
)
# check stonith configuration
warning_list = list(warning_list)
warning_list.extend(_stonith_warnings(cib, is_sbd_running))
# put it all together
if report_processor.has_errors:
raise LibraryError()
parts = []
parts.append(f"Cluster name: {corosync_conf.get_cluster_name()}")
if warning_list:
parts.extend(["", "WARNINGS:"] + warning_list + [""])
parts.append(status_text)
if verbose:
parts.extend(["", "Tickets:"])
if ticket_status_retval != 0:
ticket_warning_parts = [
"WARNING: Unable to get information about tickets"
]
if ticket_status_stderr:
ticket_warning_parts.extend(
indent(ticket_status_stderr.splitlines()))
parts.extend(indent(ticket_warning_parts))
else:
parts.extend(indent(ticket_status_text.splitlines()))
if live:
if verbose:
parts.extend(["", "PCSD Status:"])
parts.extend(
indent(
_format_node_reachability(node_name_list,
node_reachability)))
parts.extend(["", "Daemon Status:"])
parts.extend(
indent(_format_local_services_status(local_services_status)))
return "\n".join(parts)
def status_stonith_check(modifiers):
"""
Commandline options:
* -f - CIB file, to get stonith devices and cluster property
stonith-enabled from CIB, to determine whenever we are working with
files or cluster
"""
# pylint: disable=too-many-nested-blocks
# We should read the default value from pacemaker. However that may slow
# pcs down as we need to run 'pacemaker-schedulerd metadata' to get it.
warnings = []
stonith_enabled = True
stonith_devices = []
stonith_devices_id_action = []
stonith_devices_id_method_cycle = []
sbd_running = False
cib = utils.get_cib_dom()
for conf in cib.getElementsByTagName("configuration"):
for crm_config in conf.getElementsByTagName("crm_config"):
for nvpair in crm_config.getElementsByTagName("nvpair"):
if (nvpair.getAttribute("name") == "stonith-enabled"
and is_false(nvpair.getAttribute("value"))):
stonith_enabled = False
break
if not stonith_enabled:
break
for resource_el in conf.getElementsByTagName("primitive"):
if resource_el.getAttribute("class") == "stonith":
stonith_devices.append(resource_el)
for attribs in resource_el.getElementsByTagName(
"instance_attributes"):
for nvpair in attribs.getElementsByTagName("nvpair"):
if (nvpair.getAttribute("name") == "action"
and nvpair.getAttribute("value")):
stonith_devices_id_action.append(
resource_el.getAttribute("id"))
if (nvpair.getAttribute("name") == "method"
and nvpair.getAttribute("value") == "cycle"):
stonith_devices_id_method_cycle.append(
resource_el.getAttribute("id"))
if not modifiers.is_specified("-f"):
# check if SBD daemon is running
try:
sbd_running = utils.is_service_running(utils.cmd_runner(),
get_sbd_service_name())
except LibraryError:
pass
if stonith_enabled and not stonith_devices and not sbd_running:
warnings.append("No stonith devices and stonith-enabled is not false")
if stonith_devices_id_action:
warnings.append(
"Following stonith devices have the 'action' option set, "
"it is recommended to set {0} instead: {1}".format(
", ".join(
["'{0}'".format(x) for x in _STONITH_ACTION_REPLACED_BY]),
", ".join(sorted(stonith_devices_id_action))))
if stonith_devices_id_method_cycle:
warnings.append(
"Following stonith devices have the 'method' option set "
"to 'cycle' which is potentially dangerous, please consider using "
"'onoff': {0}".format(", ".join(
sorted(stonith_devices_id_method_cycle))))
return warnings
def test_systemctl(self, mock_is_systemd):
mock_is_systemd.return_value = True
self.assertEqual("sbd",
lib_sbd.get_sbd_service_name(self.service_manager))
mock_is_systemd.assert_called_once_with(self.service_manager)
def cluster_destroy(argv):
if argv:
raise CmdLineInputError()
if "--all" in utils.pcs_options:
# destroy remote and guest nodes
cib = None
lib_env = utils.get_lib_env()
try:
cib = lib_env.get_cib()
except LibraryError as e:
warn("Unable to load CIB to get guest and remote nodes from it, "
"those nodes will not be deconfigured.")
if cib is not None:
try:
all_remote_nodes = get_existing_nodes_names(cib=cib)
if len(all_remote_nodes) > 0:
_destroy_pcmk_remote_env(lib_env,
all_remote_nodes,
skip_offline_nodes=True,
allow_fails=True)
except LibraryError as e:
utils.process_library_reports(e.args)
# destroy full-stack nodes
destroy_cluster(utils.get_corosync_conf_facade().get_nodes_names())
else:
print("Shutting down pacemaker/corosync services...")
for service in ["pacemaker", "corosync-qdevice", "corosync"]:
# Returns an error if a service is not running. It is safe to
# ignore it since we want it not to be running anyways.
utils.stop_service(service)
print("Killing any remaining services...")
kill_local_cluster_services()
try:
utils.disableServices()
except:
# previously errors were suppressed in here, let's keep it that way
# for now
pass
try:
disable_service(utils.cmd_runner(), lib_sbd.get_sbd_service_name())
except:
# it's not a big deal if sbd disable fails
pass
print("Removing all cluster configuration files...")
dummy_output, dummy_retval = utils.run([
"rm",
"-f",
settings.corosync_conf_file,
settings.corosync_authkey_file,
settings.pacemaker_authkey_file,
])
state_files = [
"cib.xml*", "cib-*", "core.*", "hostcache", "cts.*", "pe*.bz2",
"cib.*"
]
for name in state_files:
dummy_output, dummy_retval = utils.run([
"find", "/var/lib/pacemaker", "-name", name, "-exec", "rm",
"-f", "{}", ";"
])
try:
qdevice_net.client_destroy()
except:
# errors from deleting other files are suppressed as well
# we do not want to fail if qdevice was not set up
pass