async def receive(request):
Log.debug(f"Received GET request for component " \
f"{request.rel_url.query['component']}")
try:
component = request.rel_url.query['component']
EventMessage.subscribe(component=component)
alert = EventMessage.receive()
except EventMessageError as e:
status_code = e.rc
error_message = e.desc
Log.error(f"Unable to receive event message for component: " \
f"{component}, status code: {status_code}," \
f" error: {error_message}")
response_obj = {'error_code': status_code, 'exception': \
['EventMessageError', {'message': error_message}]}
except Exception as e:
exception_key = type(e).__name__
exception = RestServerError(exception_key).http_error()
status_code = exception[0]
error_message = exception[1]
Log.error(f"Internal error while receiving event messages for " \
f"component: {component}, status code: " \
f"{status_code}, error: {error_message}")
response_obj = {'error_code': status_code, 'exception': \
[exception_key, {'message': error_message}]}
raise EventMessageError(status_code, error_message) from e
else:
status_code = 200 # No exception, Success
response_obj = {'alert': alert}
Log.debug(f"GET method finished with status code: {status_code}" \
f"for component {component} and received event message " \
f"alert info. - {alert['iem']['info']}.")
finally:
return web.Response(text=json.dumps(response_obj), \
status=status_code)
def run_cmd(self, cmd, check_error=True):
"""
Run command and throw error if cmd failed
Args:
cmd ([string]): Command to execute on system.
Raises:
Exception: raise command failed exception.
HACommandTerminated: Command termineted exception
Returns:
string: Command output.
"""
try:
_err = ""
_proc = SimpleProcess(cmd)
_output, _err, _rc = _proc.run(universal_newlines=True)
Log.debug(f"cmd: {cmd}, output: {_output}, err: {_err}, rc: {_rc}")
if _rc != 0 and check_error:
Log.error(
f"cmd: {cmd}, output: {_output}, err: {_err}, rc: {_rc}")
raise Exception(f"Failed to execute {cmd}")
return _output, _err, _rc
except Exception as e:
Log.error("Failed to execute %s Error: %s %s" % (cmd, e, _err))
raise HACommandTerminated("Failed to execute %s Error: %s %s" %
(cmd, e, _err))
def __init__(self, rc=1, desc=None, message_id=HA_BASIC_ERROR, message_args=None):
"""
Parent class for the HA error classes
"""
super(HAError, self).__init__(rc=rc, desc=desc, message_id=message_id,
message_args=message_args)
Log.error(f"error({self._message_id}):rc({self._rc}):{self._desc}:{self._message_args}")
def _monitor_action(self, callback_ack, state, **args):
"""
Return action on status
"""
Log.debug(str(args))
if args[const.CURRENT_NODE_STATUS] == Action.FAILED and args[const.OTHER_NODE_STATUS] == Action.FAILED:
return const.OCF_SUCCESS
elif args[const.CURRENT_NODE_STATUS] == Action.FAILED:
return const.OCF_ERR_GENERIC
elif args[const.CURRENT_NODE_STATUS] == Action.OK:
return const.OCF_SUCCESS
elif args[const.CURRENT_NODE_STATUS] == Action.RESOLVED:
Log.info(f"Ack for {args[const.FILENAME_KEY]} with key {args[const.PATH_KEY]}"
f" node {args[const.CURRENT_NODE]}")
return callback_ack(args[const.PATH_KEY]+'_'+args[const.CURRENT_NODE])
elif args[const.CURRENT_NODE_STATUS] == Action.RESTART:
Log.info(f"Restart action taken for {args[const.FILENAME_KEY]} on {args[const.CURRENT_NODE]}")
if state == const.STATE_START:
return const.OCF_SUCCESS
elif state == const.STATE_RUNNING:
return const.OCF_ERR_GENERIC
elif state == const.STATE_STOP:
callback_ack(args[const.PATH_KEY]+'_'+args[const.CURRENT_NODE])
Log.info(f"Ack for {args[const.FILENAME_KEY]} with key {args[const.PATH_KEY]} "
f" node {args[const.CURRENT_NODE]}")
return Action.RESTART
return const.OCF_SUCCESS
else:
Log.error(f"Unimplemented value for status {args[const.CURRENT_NODE_STATUS]}")
return const.OCF_ERR_UNIMPLEMENTED
def main(resource: DynamicFidServiceRA, action: str = '') -> int:
"""
Main function acts as switch case for DynamicFidServiceRA resource agent.
Args:
resource (DynamicFidServiceRA): Resource agent
action (str): Resource agent action called by Pacemaker. Defaults to ''.
Returns:
int: Provide output as int code provided by pacemaker.
"""
try:
if action == "meta-data":
return resource.metadata()
ConfigManager.init("resource_agent")
Log.debug(f"{resource} initialized for action {action}")
if action == "monitor":
return resource_agent.monitor()
elif action == "start":
return resource_agent.start()
elif action == "stop":
return resource_agent.stop()
else:
print(f"Usage {sys.argv[0]} [monitor] [start] [stop] [meta-data]")
exit(0)
except Exception as e:
Log.error(
f"systemd_fid_wrapper_ra failed to perform {action}. Error: {e}")
return const.OCF_ERR_GENERIC
def remove_node(self, node):
"""
Remove node from pcs cluster
"""
# TODO: Limitation for node remove (in cluster node cannot remove it self)
# Check if node already removed
_output, _err, _rc = self._execute.run_cmd(const.PCS_STATUS)
Log.info(
f"Cluster status output before remove node: {_output}, {_err}, {_rc}"
)
_rc, status = self.node_status(node)
if _rc != 1:
self._execute.run_cmd(f"pcs cluster node remove {node} --force")
_rc, status = self.node_status(node)
Log.debug(f"For node {node} status: {status}, rc: {_rc}")
if _rc != 1:
Log.error(f"Failed to remove {node}")
raise Exception(f"Failed to remove {node}")
else:
Log.info(f"Node {node} removed from cluster")
else:
Log.info(f"Node {node} already removed from cluster")
_output, _err, _rc = self._execute.run_cmd(const.PCS_STATUS)
Log.info(
f"Cluster status output after remove node: {_output}, {_err}, {_rc}"
)
def addattribute(conn, dn, attr_to_add, value):
mod_attrs = [(ldap.MOD_ADD, attr_to_add, bytes(str(value), 'utf-8'))]
try:
conn.modify_s(dn, mod_attrs)
except Exception:
Log.error('Exception while adding '+ attr_to_add + ' to dn '+ dn)
raise Exception('Exception while adding '+ attr_to_add + ' to dn '+ dn)
def read_ldap_credentials(self):
"""Read ldap credentials (rootdn, sgiam) from the openldap_config file."""
try:
# Load the openldap config file
index_id = 'openldap_config_file_read_index'
Conf.load(index_id, f'yaml://{self.openldap_config_file}')
# Read the cluster id from openldap_config file
self.cluster_id = Conf.get(index_id, f'{self.cluster_id_key}')
cipher_key = Cipher.generate_key(
self.cluster_id,
self.get_confkey('CONFSTORE_OPENLDAP_CONST_KEY'))
# rootdn username/password
self.ldap_root_user = Conf.get(index_id, f'{self.rootdn_user_key}')
encrypted_rootdn_pass = Conf.get(index_id,
f'{self.rootdn_pass_key}')
if encrypted_rootdn_pass != None:
self.rootdn_passwd = Cipher.decrypt(
cipher_key, bytes(str(encrypted_rootdn_pass), 'utf-8'))
except Exception as e:
Log.error(f'read ldap credentials failed, error: {e}')
raise e
def update_ldap_credentials(self):
"""Update ldap credentials (rootdn, sgiam) to openldap_config file."""
try:
# Load the openldap config file
index_id = 'openldap_config_file_write_index'
Conf.load(index_id, f'yaml://{self.openldap_config_file}')
# get the rootdn credentials from provisoner config file
# set the rootdn credentials in to openldap_config file
ldap_root_user = self.get_confvalue(
self.get_confkey('CONFIG>CONFSTORE_ROOTDN_USER_KEY'))
encrypted_rootdn_pass = self.get_confvalue(
self.get_confkey('CONFIG>CONFSTORE_ROOTDN_PASSWD_KEY'))
if encrypted_rootdn_pass is None:
Log.error('rootdn password cannot be None.')
raise Exception('rootdn password cannot be None.')
Conf.set(index_id, f'{self.rootdn_user_key}', f'{ldap_root_user}')
Conf.set(index_id, f'{self.rootdn_pass_key}',
f'{encrypted_rootdn_pass}')
# save openldap config file
Conf.save(index_id)
except Exception as e:
Log.error(f'update rootdn credentials failed, error: {e}')
raise Exception(f'update rootdn credentials failed, error: {e}')
def register_message_type(self, admin_id: str, message_types: list, \
partitions: int):
"""
Creates a list of message types.
Parameters:
admin_id A String that represents Admin client ID.
message_types This is essentially equivalent to the list of
queue/topic name. For e.g. ["Alert"]
partitions Integer that represents number of partitions to be
created.
"""
Log.debug(f"Register message type {message_types} using {admin_id}" \
f" with {partitions} partitions")
admin = self._clients['admin'][admin_id]
new_message_type = [NewTopic(each_message_type, \
num_partitions=partitions) for each_message_type in message_types]
created_message_types = admin.create_topics(new_message_type)
self._task_status(created_message_types, method='register_message_type')
for each_message_type in message_types:
for list_retry in range(1, self._max_list_message_type_count+2):
if each_message_type not in \
list(self._get_metadata(admin).keys()):
if list_retry > self._max_list_message_type_count:
Log.error(f"MessageBusError: Timed out after retry " \
f"{list_retry} while creating message_type " \
f"{each_message_type}")
raise MessageBusError(errno.ETIMEDOUT, "Timed out " +\
"after retry %d while creating message_type %s.", \
list_retry, each_message_type)
time.sleep(list_retry*1)
continue
else:
break
def init(cls, component: str, source: str, cluster_id: str, \
message_server_endpoints: str, **message_server_kwargs):
"""
Set the Event Message context
Parameters:
component Component that generates the IEM. For e.g. 'S3', 'SSPL'
source Single character that indicates the type of component.
For e.g. H-Hardware, S-Software, F-Firmware, O-OS
"""
cls._component = component
cls._source = source
cls._site_id = 1
cls._rack_id = 1
cls._node_id = Conf.machine_id
cls._cluster_id = cluster_id
if cls._component is None:
Log.error("Invalid component type: %s" % cls._component )
raise EventMessageError(errno.EINVAL, "Invalid component type: %s", \
cls._component)
if cls._source not in cls._SOURCE.keys():
Log.error("Invalid source type: %s" % cls._source)
raise EventMessageError(errno.EINVAL, "Invalid source type: %s", \
cls._source)
MessageBus.init(message_server_endpoints, **message_server_kwargs)
cls._producer = MessageProducer(producer_id='event_producer', \
message_type='IEM', method='sync')
Log.info("IEM Producer initialized for component %s and source %s" % \
(cls._component, cls._source))
def get_system_health(
self,
element: CLUSTER_ELEMENTS = CLUSTER_ELEMENTS.CLUSTER.value,
depth: int = 1,
**kwargs) -> json:
"""
Return health status for the requested elements.
Args:
element ([CLUSTER_ELEMENTS]): The element whose health status is to be returned.
depth ([int]): A depth of elements starting from the input "element" that the health status
is to be returned.
**kwargs([dict]): Variable number of arguments that are used as filters,
e.g. "id" of the input "element".
Returns:
([dict]): Returns dictionary. {"status": "Succeeded"/"Failed"/"Partial", "output": "", "error": ""}
status: Succeeded, Failed, Partial
output: Dictionary with element health status
error: Error information if the request "Failed"
"""
try:
# Fetch the health status
system_health_controller = SystemHealthController(self._confstore)
return system_health_controller.get_status(component=element,
depth=depth,
version=self._version,
**kwargs)
except Exception as e:
Log.error(f"Failed returning system health . Error: {e}")
raise ClusterManagerError(
"Failed returning system health, internal error")
def _validate_permissions(self) -> None:
# confirm that user is root or part of haclient group"
user = getpass.getuser()
group_id = os.getgid()
group_list = os.getgrouplist(user, group_id)
try:
# find group id for root and haclient
id_ha = grp.getgrnam(const.USER_GROUP_HACLIENT)
id_root = grp.getgrnam(const.USER_GROUP_ROOT)
# if group not root or haclient return error
if id_ha.gr_gid not in group_list and id_root.gr_gid not in group_list:
Log.error(
f"User {user} does not have necessary permissions to execute this CLI"
)
raise HAInvalidPermission(
f"User {user} does not have necessary permissions to execute this CLI"
)
# TODO : Check if user needs to be changed to hauser
except KeyError:
Log.error("haclient group is not present on the system")
raise HAInvalidPermission(
"haclient group is not present on the system")
def main(action: str = '') -> int:
"""
Main function acts as switch case for IPHealthChecker resource agent.
Args:
action (str): Resource agent action called by Pacemaker. Defaults to ''.
Returns:
int: Provide output as int code provided by pacemaker.
"""
try:
if action == "meta-data":
return VipHealthMonitor.metadata()
ConfigManager.init("resource_agent")
resource_agent = VipHealthMonitor()
Log.debug(f"{resource_agent} initialized for action {action}")
if action == "monitor":
return resource_agent.monitor()
elif action == "start":
return resource_agent.start()
elif action == "stop":
return resource_agent.stop()
else:
print(f"Usage {sys.argv[0]} [monitor] [start] [stop] [meta-data]")
exit(0)
except Exception as e:
Log.error(
f"vip health check failed to perform {action}. Error: {traceback.format_exc()} {e}"
)
return const.OCF_ERR_GENERIC
async def _exc_components_cmd(commands: List, bundle_id: str, path: str, \
component: str, node_name: str, comment: str, config_url:str,
services:str, binlogs:bool, coredumps:bool, stacktrace:bool,
duration:str, size_limit:str):
"""
Executes the Command for Bundle Generation of Every Component.
commands: Command of the component :type:str
bundle_id: Unique Bundle ID of the generation process. :type:str
path: Path to create the tar by components :type:str
component: Name of Component to be executed :type: str
node_name: Name of Node where the Command is being Executed :type:str
comment: User Comment: type:str
"""
for command in commands:
# SB Framework will not parse additional filters until all the components
# accept filters in their respective support bundle scripts.
cli_cmd = f"{command} -b {bundle_id} -t {path} -c {config_url}"\
f" -s {services} --duration {duration} --size_limit {size_limit}"\
f" --binlogs {binlogs} --coredumps {coredumps} --stacktrace {stacktrace}"
Log.info(f"Executing command -> {cli_cmd}")
cmd_proc = SimpleProcess(cli_cmd)
output, err, return_code = cmd_proc.run()
Log.debug(f"Command Output -> {output} {err}, {return_code}")
if return_code != 0:
Log.error(f"Command Output -> {output} {err}, {return_code}")
else:
Log.debug(f"Command Output -> {output} {err}, {return_code}")
return component, return_code
def main(argv: list):
try:
if len(sys.argv) == 1:
Cmd.usage("ha_setup")
sys.exit(1)
if sys.argv[1] == "cleanup":
if not os.path.exists(const.HA_CONFIG_FILE):
a_str = f'Cleanup can not be proceed as \
HA config file: {const.HA_CONFIG_FILE} \
is missing. Either cleanup is already done or there \
is some other problem'
sys.stdout.write(a_str)
return 0
ConfigManager.init("ha_setup")
desc = "HA Setup command"
command = Cmd.get_command(desc, argv[1:])
command.process()
except Exception as err:
Log.error("%s\n" % traceback.format_exc())
sys.stderr.write(
f"Setup command:{argv[1]} failed for cortx-ha. Error: {err}\n")
return errno.EINVAL
def _exc_components_cmd(commands: List, bundle_id: str, path: str,
component: str, node_name: str, comment: str):
"""
Executes the Command for Bundle Generation of Every Component.
:param commands: Command of the component :type:str
:param bundle_id: Unique Bundle ID of the generation process. :type:str
:param path: Path to create the tar by components :type:str
:param component: Name of Component to be executed :type: str
:param node_name:Name of Node where the Command is being Executed
:type:str
:param comment: :User Comment: type:str
:return:
"""
for command in commands:
Log.info(f"Executing command -> {command} {bundle_id} {path}")
cmd_proc = SimpleProcess(f"{command} {bundle_id} {path}")
output, err, return_code = cmd_proc.run()
Log.debug(f"Command Output -> {output} {err}, {return_code}")
if return_code != 0:
Log.error(f"Command Output -> {output} {err}, {return_code}")
ComponentsBundle._publish_log(
f"Bundle generation failed for '{component}'", ERROR,
bundle_id, node_name, comment)
else:
ComponentsBundle._publish_log(
f"Bundle generation started for '{component}'", INFO,
bundle_id, node_name, comment)
def validate_permissions(self) -> None:
# confirm that user is root or part of haclient group"
user = getpass.getuser()
group_id = os.getgid()
try:
# find group id for root and haclient
id_ha = grp.getgrnam(const.USER_GROUP_HACLIENT)
id_root = grp.getgrnam(const.USER_GROUP_ROOT)
# if group not root or haclient return error
if group_id != id_ha.gr_gid and group_id != id_root.gr_gid:
Log.error(
f"User {user} does not have necessary permissions to execute this CLI"
)
raise HAInvalidPermission(
f"User {user} does not have necessary permissions to execute this CLI"
)
# The user name "hauser"; which is part of the "haclient" group;
# is used by HA.
# internal commands are allowed only if the user is "hauser"
# As of now, every HA CLI will be internal command. So, we
# do not need this change. We can revisit this if needed in future
#if user == const.USER_HA_INTERNAL:
# self._is_hauser = True
# TBD : If required raise seperate exception for root and haclient
except KeyError:
Log.error("Group root / haclient is not defined")
raise HAInvalidPermission("Group root / haclient is not defined ")
async def get_bundle_status(command):
"""
Initializes the process for Displaying the Status for Support Bundle.
:param command: Csm_cli Command Object :type: command
:return: None
"""
try:
bundle_id = command.options.get(const.SB_BUNDLE_ID)
conf = GeneralConfig(database.DATABASE)
db = DataBaseProvider(conf)
repo = SupportBundleRepository(db)
all_nodes_status = await repo.retrieve_all(bundle_id)
response = {
"status": [
each_status.to_primitive()
for each_status in all_nodes_status
]
}
return Response(output=response, rc=OPERATION_SUCESSFUL)
except DataAccessExternalError as e:
Log.warn(f"Failed to connect to elasticsearch: {e}")
return Response(
output=("Support Bundle status is not available currently"
" as required services are not running."
" Please wait and check the /tmp/support_bundle"
" folder for newly generated support bundle."),
rc=str(errno.ECONNREFUSED))
except Exception as e:
Log.error(f"Failed to get bundle status: {e}")
return Response(
output=("Support Bundle status is not available currently"
" as required services are not running."
" Failed to get status of bundle."),
rc=str(errno.ENOENT))
def get_status_raw(self,
component: str,
component_id: str = None,
**kwargs):
"""
get status method. This is a generic method which can return status of any component(s).
"""
status = None
try:
# Prepare key and read the health value.
if component_id != None:
key = ElementHealthEvaluator.prepare_key(component,
comp_id=component_id,
**kwargs)
status = self.healthmanager.get_key(key)
else:
key = ElementHealthEvaluator.prepare_key(component, **kwargs)
status = self.healthmanager.get_key(key, just_value=False)
# Remove any keys which are not for the health status.
ignore_keys = []
for key in status:
if "health" not in key:
ignore_keys.append(key)
for key in ignore_keys:
del status[key]
return status
except Exception as e:
Log.error(
f"Failed reading status for component: {component} with Error: {e}"
)
raise HaSystemHealthException("Failed reading status")
def cleanup(forceclean):
# Removing schemas
filelist = glob.glob(
'/etc/openldap/slapd.d/cn=config/cn=schema/*ppolicy.ldif')
for policyfile in filelist:
BaseConfig.safe_remove(policyfile)
module_files = [
"cn=module{0}.ldif", "cn=module{1}.ldif", "cn=module{2}.ldif"
]
for module_file in module_files:
module_file = '/etc/openldap/slapd.d/cn=config/' + str(module_file)
BaseConfig.safe_remove(module_file)
mdb_directory = '/etc/openldap/slapd.d/cn=config/olcDatabase={2}mdb'
try:
files = glob.glob(
'/etc/openldap/slapd.d/cn=config/olcDatabase={2}mdb/*')
for f in files:
BaseConfig.safe_remove(f)
except Exception:
Log.error('Error while deleting ' + mdb_directory)
mdbfile = '/etc/openldap/slapd.d/cn=config/olcDatabase={2}mdb.ldif'
BaseConfig.safe_remove(mdbfile)
#Data Cleanup
if forceclean == 'True':
files = glob.glob('/var/lib/ldap/*')
for f in files:
BaseConfig.safe_remove(f)
#Clear the password set in config
conn = ldap.initialize("ldapi:///")
conn.sasl_non_interactive_bind_s('EXTERNAL')
Replication.deleteattribute(conn, "olcDatabase={0}config,cn=config",
"olcRootPW")
conn.unbind_s()
def get_hierarchy(component: str) -> dict:
"""
This method returns a component health update hierarchy.
"""
try:
if component in SystemHealthHierarchy._server_hw:
return SystemHealthHierarchy._server_hw[
SystemHealthHierarchy._server_hw.index(component):]
elif component in SystemHealthHierarchy._server_service:
return SystemHealthHierarchy._server_service[
SystemHealthHierarchy._server_service.index(component):]
elif component in SystemHealthHierarchy._storage:
return SystemHealthHierarchy._storage[
SystemHealthHierarchy._storage.index(component):]
else:
Log.error(
f"Health update hierarchy not present for component: {component}"
)
raise HaSystemHealthHierarchyException(
"Health update hierarchy not present for the component")
except Exception as e:
Log.error(
f"Failed to get health update hierarchy for component with Error: {e}"
)
raise HaSystemHealthHierarchyException(
"Failed to get health update hierarchy")
def main(resource, action=''):
try:
if action == 'meta-data':
return resource.metadata()
Conf.load(const.HA_GLOBAL_INDEX, Yaml(const.HA_CONFIG_FILE))
log_path = Conf.get(const.HA_GLOBAL_INDEX, f"LOG{_DELIM}path")
log_level = Conf.get(const.HA_GLOBAL_INDEX, f"LOG{_DELIM}level")
Log.init(service_name='resource_agent', log_path=log_path, level=log_level)
with open(const.RESOURCE_SCHEMA, 'r') as f:
resource_schema = json.load(f)
os.makedirs(const.RA_LOG_DIR, exist_ok=True)
resource_agent = resource(DecisionMonitor(), resource_schema)
Log.debug(f"{resource_agent} initialized for action {action}")
if action == 'monitor':
return resource_agent.monitor()
elif action == 'start':
return resource_agent.start()
elif action == 'stop':
return resource_agent.stop()
else:
print('Usage %s [monitor] [start] [stop] [meta-data]' % sys.argv[0])
exit()
except Exception as e:
Log.error(f"{traceback.format_exc()}")
return const.OCF_ERR_GENERIC
def deleteattribute(conn, dn, attr_to_delete):
mod_attrs = [(ldap.MOD_DELETE, attr_to_delete, None)]
try:
conn.modify_s(dn, mod_attrs)
except Exception:
Log.error('Exception while deleting ' + attr_to_delete +
' from dn ' + dn)
def publish(self, event: RecoveryActionEvent) -> None:
"""
Publish event.
Args:
event (RecoveryActionEvent): Action event.
"""
try:
#TODO: Use Transactional producer in future.
component_list = []
# Run through list of components subscribed for this event and send event to each of them
component_list_key = EVENT_MANAGER_KEYS.EVENT_KEY.value.replace(
"<resource>",
event.resource_type).replace("<state>", event.event_type)
component_list_key_val = self._confstore.get(component_list_key)
if component_list_key_val:
_, value = component_list_key_val.popitem()
component_list = json.loads(value)
for component in component_list:
message_producer = self._get_producer(component)
event_to_send = str(event)
Log.info(
f"Sending action event {event_to_send} to component {component}"
)
message_producer.publish(event_to_send)
except Exception as e:
Log.error(
f"Failed sending message for {event.resource_type}, Error: {e}"
)
raise PublishException(
f"Failed sending message for {event.resource_type}, Error: {e}"
)
def deregister_message_type(self, admin_id: str, message_types: list):
"""
Deletes a list of message types.
Parameters:
admin_id A String that represents Admin client ID.
message_types This is essentially equivalent to the list of
queue/topic name. For e.g. ["Alert"]
"""
Log.debug(f"Deregister message type {message_types} using {admin_id}")
admin = self._clients['admin'][admin_id]
deleted_message_types = admin.delete_topics(message_types)
self._task_status(deleted_message_types,\
method='deregister_message_type')
for each_message_type in message_types:
for list_retry in range(1, self._max_list_message_type_count + 2):
if each_message_type in list(self._get_metadata(admin).keys()):
if list_retry > self._max_list_message_type_count:
Log.error(f"MessageBusError: Timed out after "\
f"{list_retry} retry to delete message_type "\
f"{each_message_type}")
raise MessageBusError(errno.ETIMEDOUT,\
"Timed out after %d retry to delete message_type" +\
"%s.", list_retry, each_message_type)
time.sleep(list_retry * 1)
continue
else:
break
def __init__(self):
""" Initialize a MessageBus and load its configurations """
Conf.load('config_file', 'json:///etc/cortx/cortx.conf',
skip_reload=True)
# if Log.logger is already initialized by some parent process
# the same file will be used to log all the messagebus related
# logs, else standard message_bus.log will be used.
if not Log.logger:
log_level = Conf.get('config_file', 'utils>log_level', 'INFO')
Log.init('message_bus', '/var/log/cortx/utils/message_bus', \
level=log_level, backup_count=5, file_size_in_mb=5)
try:
Conf.load('message_bus', self.conf_file, skip_reload=True)
self._broker_conf = Conf.get('message_bus', 'message_broker')
broker_type = self._broker_conf['type']
Log.info(f"MessageBus initialized as {broker_type}")
except ConfError as e:
Log.error(f"MessageBusError: {e.rc} Error while parsing" \
f" configuration file {self.conf_file}. {e}.")
raise MessageBusError(e.rc, "Error while parsing " + \
"configuration file %s. %s.", self.conf_file, e)
except Exception as e:
Log.error(f"MessageBusError: {e.rc} Error while parsing" \
f" configuration file {self.conf_file}. {e}.")
raise MessageBusError(errno.ENOENT, "Error while parsing " + \
"configuration file %s. %s.", self.conf_file, e)
self._broker = MessageBrokerFactory.get_instance(broker_type, \
self._broker_conf)
def send(self, producer_id: str, message_type: str, method: str,\
messages: list, timeout=0.1):
"""
Sends list of messages to Kafka cluster(s).
Parameters:
producer_id A String that represents Producer client ID.
message_type This is essentially equivalent to the
queue/topic name. For e.g. "Alert"
method Can be set to "sync" or "async"(default).
messages A list of messages sent to Kafka Message Server
"""
Log.debug(f"Producer {producer_id} sending list of messages "\
f"{messages} of message type {message_type} to kafka server"\
f" with method {method}")
producer = self._clients['producer'][producer_id]
if producer is None:
Log.error(f"MessageBusError: "\
f"{errors.ERR_SERVICE_NOT_INITIALIZED}. Producer: "\
f"{producer_id} is not initialized")
raise MessageBusError(errors.ERR_SERVICE_NOT_INITIALIZED,\
"Producer %s is not initialized", producer_id)
for message in messages:
producer.produce(message_type, bytes(message, 'utf-8'),\
callback=self.delivery_callback)
if method == 'sync':
producer.flush()
else:
producer.poll(timeout=timeout)
Log.debug("Successfully Sent list of messages to Kafka cluster")
def post_install(self):
"""
Performs post install operations.
Raises exception on error
"""
try:
# Remove opendistro_security plugins.
Elasticsearch.delete_path(self.opendistro_security_plugin)
# Delete opendistro_security configuration entries from
# elasticsearch.yml file.
file_contents = Elasticsearch.read_file_contents(
self.elasticsearch_config_path)
flag = 1
start_line = "######## Start OpenDistro"
end_line = "######## End OpenDistro"
with open(self.elasticsearch_config_path, "w") as f:
for line in file_contents:
if line.startswith(start_line):
flag = 0
if line.startswith(end_line):
flag = 1
if flag and not line.startswith(end_line):
f.write(line)
f.close()
except OSError as e:
msg = f"Exception in in post_install {e}."
Log.error(msg)
raise
Log.info("Post_install done.")
return 0
def filter_event(self):
"""
Filter event based on Node, Resource and Fencing type.
Args:
msg (str): Msg
"""
ha_required_alert_module = None
ha_required_alert_module_operation = None
if self.crm_env is None:
raise AlertEventFilterError(f"Identified empty events: {str(self.crm_env)}")
try:
alert_filter_module = self.crm_env["CRM_alert_kind"]
alert_filter_module_operation = self.crm_env["CRM_alert_desc"]
if alert_filter_module.lower() in self.alert_filter_modules.get(ALERTS.REQUIRED_COMPONENT):
ha_required_alert_module = alert_filter_module
if ha_required_alert_module and alert_filter_module_operation.lower() in self.alert_filter_module_operations.get(
ha_required_alert_module):
ha_required_alert_module_operation = alert_filter_module_operation
return ha_required_alert_module, ha_required_alert_module_operation
except Exception as e:
Log.error(f"Error occurred in alert event filter: {str(e)}")
return ha_required_alert_module, ha_required_alert_module_operation