def _GroupWatcher(opts):
"""Main function for per-group watcher process.
"""
group_uuid = opts.nodegroup.lower()
if not utils.UUID_RE.match(group_uuid):
raise errors.GenericError(
"Node group parameter (%s) must be given a UUID,"
" got '%s'" % (cli.NODEGROUP_OPT_NAME, group_uuid))
logging.info("Watcher for node group '%s'", group_uuid)
known_groups = _LoadKnownGroups()
# Check if node group is known
if group_uuid not in known_groups:
raise errors.GenericError("Node group '%s' is not known by ssconf" %
group_uuid)
# Group UUID has been verified and should not contain any dangerous
# characters
state_path = pathutils.WATCHER_GROUP_STATE_FILE % group_uuid
inst_status_path = pathutils.WATCHER_GROUP_INSTANCE_STATUS_FILE % group_uuid
logging.debug("Using state file %s", state_path)
# Global watcher
statefile = state.OpenStateFile(state_path) # pylint: disable=E0602
if not statefile:
return constants.EXIT_FAILURE
notepad = state.WatcherState(statefile) # pylint: disable=E0602
try:
# Connect to master daemon
client = GetLuxiClient(False)
_CheckMaster(client)
(nodes, instances, locks) = _GetGroupData(client, group_uuid)
# Update per-group instance status file
_UpdateInstanceStatus(inst_status_path, instances.values())
_MergeInstanceStatus(pathutils.INSTANCE_STATUS_FILE,
pathutils.WATCHER_GROUP_INSTANCE_STATUS_FILE,
known_groups)
restart_needed, started = _CheckInstances(client, notepad, instances,
locks)
_CheckDisks(client, notepad, nodes, instances, started)
if not opts.no_verify_disks:
_VerifyDisks(client, group_uuid, nodes, instances)
except Exception, err:
logging.info("Not updating status file due to failure: %s", err)
raise
def _GroupWatcher(opts):
"""Main function for per-group watcher process.
"""
group_uuid = opts.nodegroup.lower()
if not utils.UUID_RE.match(group_uuid):
raise errors.GenericError(
"Node group parameter (%s) must be given a UUID,"
" got '%s'" % (cli.NODEGROUP_OPT_NAME, group_uuid))
logging.info("Watcher for node group '%s'", group_uuid)
known_groups = _LoadKnownGroups()
# Check if node group is known
if group_uuid not in known_groups:
raise errors.GenericError("Node group '%s' is not known by ssconf" %
group_uuid)
# Group UUID has been verified and should not contain any dangerous
# characters
state_path = pathutils.WATCHER_GROUP_STATE_FILE % group_uuid
inst_status_path = pathutils.WATCHER_GROUP_INSTANCE_STATUS_FILE % group_uuid
logging.debug("Using state file %s", state_path)
# Global watcher
statefile = state.OpenStateFile(state_path) # pylint: disable=E0602
if not statefile:
return constants.EXIT_FAILURE
notepad = state.WatcherState(statefile) # pylint: disable=E0602
try:
# Connect to master daemon
client = GetLuxiClient(False)
_CheckMaster(client)
(nodes, instances, locks) = _GetGroupData(client, group_uuid)
# Update per-group instance status file
_UpdateInstanceStatus(inst_status_path, list(instances.values()))
_MergeInstanceStatus(pathutils.INSTANCE_STATUS_FILE,
pathutils.WATCHER_GROUP_INSTANCE_STATUS_FILE,
known_groups)
started = _CheckInstances(client, notepad, instances, locks)
_CheckDisks(client, notepad, nodes, instances, started)
# Check if the nodegroup only has ext storage type
only_ext = compat.all(i.disk_template == constants.DT_EXT
for i in instances.values())
# We skip current NodeGroup verification if there are only external storage
# devices. Currently we provide an interface for external storage provider
# for disk verification implementations, however current ExtStorageDevice
# does not provide an API for this yet.
#
# This check needs to be revisited if ES_ACTION_VERIFY on ExtStorageDevice
# is implemented.
if not opts.no_verify_disks and not only_ext:
_VerifyDisks(client, group_uuid, nodes, instances)
except Exception as err:
logging.info("Not updating status file due to failure: %s", err)
raise
else:
# Save changes for next run
notepad.Save(state_path)
return constants.EXIT_SUCCESS