def __init__(self, agentConfig):
try:
self.config_store = get_config_store(agentConfig=agentConfig)
except Exception as e:
log.error('Failed to instantiate the config store client. '
'Auto-config only will be used. %s' % str(e))
agentConfig['sd_config_backend'] = None
self.config_store = get_config_store(agentConfig=agentConfig)
self.dockerutil = DockerUtil(config_store=self.config_store)
self.kubeutil = None
if Platform.is_k8s():
try:
self.kubeutil = KubeUtil()
except Exception as ex:
log.error(
"Couldn't instantiate the kubernetes client, "
"subsequent kubernetes calls will fail as well. Error: %s"
% str(ex))
self.metadata_collector = MetadataCollector()
self.VAR_MAPPING = {
'host': self._get_host_address,
'pid': self._get_container_pid,
'port': self._get_port,
'container-name': self._get_container_name,
'tags': self._get_additional_tags,
}
AbstractSDBackend.__init__(self, agentConfig)
def init(self):
try:
instance = self.instances[0]
# Getting custom tags for service checks when docker is down
self.custom_tags = instance.get("tags", [])
self.docker_util = DockerUtil()
if not self.docker_util.client:
raise Exception("Failed to initialize Docker client.")
self.docker_gateway = DockerUtil.get_gateway()
self.metadata_collector = MetadataCollector()
self.kubeutil = None
if Platform.is_k8s():
try:
self.kubeutil = KubeUtil()
except Exception as ex:
self.log.error("Couldn't instantiate the kubernetes client, "
"subsequent kubernetes calls will fail as well. Error: %s" % str(ex))
# We configure the check with the right cgroup settings for this host
# Just needs to be done once
self._mountpoints = self.docker_util.get_mountpoints(CGROUP_METRICS)
self._latest_size_query = 0
self._filtered_containers = set()
self._disable_net_metrics = False
# Set tagging options
# The collect_labels_as_tags is legacy, only tagging docker metrics.
# It is replaced by docker_labels_as_tags in config.cfg.
# We keep this line for backward compatibility.
if "collect_labels_as_tags" in instance:
self.collect_labels_as_tags = instance.get("collect_labels_as_tags")
self.kube_pod_tags = {}
self.use_histogram = _is_affirmative(instance.get('use_histogram', False))
performance_tags = instance.get("performance_tags", DEFAULT_PERFORMANCE_TAGS)
self.tag_names = {
CONTAINER: instance.get("container_tags", DEFAULT_CONTAINER_TAGS),
PERFORMANCE: performance_tags,
IMAGE: instance.get('image_tags', DEFAULT_IMAGE_TAGS)
}
# Set filtering settings
if self.docker_util.filtering_enabled:
self.tag_names[FILTERED] = self.docker_util.filtered_tag_names
# Container network mapping cache
self.network_mappings = {}
# get the health check whitelist
self.whitelist_patterns = None
health_scs_whitelist = instance.get('health_service_check_whitelist', [])
if health_scs_whitelist:
patterns, whitelist_tags = compile_filter_rules(health_scs_whitelist)
self.whitelist_patterns = set(patterns)
self.tag_names[HEALTHCHECK] = set(whitelist_tags)
# Other options
self.collect_image_stats = _is_affirmative(instance.get('collect_images_stats', False))
self.collect_container_size = _is_affirmative(instance.get('collect_container_size', False))
self.collect_container_count = _is_affirmative(instance.get('collect_container_count', False))
self.collect_volume_count = _is_affirmative(instance.get('collect_volume_count', False))
self.collect_events = _is_affirmative(instance.get('collect_events', True))
self.event_attributes_as_tags = instance.get('event_attributes_as_tags', [])
self.collect_image_size = _is_affirmative(instance.get('collect_image_size', False))
self.collect_disk_stats = _is_affirmative(instance.get('collect_disk_stats', False))
self.collect_exit_codes = _is_affirmative(instance.get('collect_exit_codes', False))
self.collect_ecs_tags = _is_affirmative(instance.get('ecs_tags', True)) and Platform.is_ecs_instance()
self.filtered_event_types = tuple(instance.get("filtered_event_types", DEFAULT_FILTERED_EVENT_TYPES))
self.capped_metrics = instance.get('capped_metrics')
except Exception as e:
self.log.critical(e)
self.warning("Initialization failed. Will retry at next iteration")
else:
self.init_success = True
def _populate_payload_metadata(self,
payload,
check_statuses,
start_event=True):
"""
Periodically populate the payload with metadata related to the system, host, and/or checks.
"""
now = time.time()
# Include system stats on first postback
if start_event and self._is_first_run():
payload['systemStats'] = self.agentConfig.get('system_stats', {})
# Also post an event in the newsfeed
payload['events']['System'] = [{
'api_key':
self.agentConfig['api_key'],
'host':
self.hostname,
'timestamp':
now,
'event_type':
'Agent Startup',
'msg_text':
'Version %s' % get_version()
}]
# Periodically send the host metadata.
if self._should_send_additional_data('host_metadata'):
# gather metadata with gohai
gohai_metadata = self._run_gohai_metadata()
if gohai_metadata:
payload['gohai'] = gohai_metadata
payload['systemStats'] = get_system_stats(
proc_path=self.agentConfig.get('procfs_path', '/proc').rstrip(
'/'))
if self.agentConfig['collect_orchestrator_tags']:
host_container_metadata = MetadataCollector(
).get_host_metadata()
if host_container_metadata:
payload['container-meta'] = host_container_metadata
payload['meta'] = self._get_hostname_metadata()
self.hostname_metadata_cache = payload['meta']
# Add static tags from the configuration file
host_tags = []
if self.agentConfig['tags'] is not None:
host_tags.extend([
unicode(tag.strip())
for tag in self.agentConfig['tags'].split(",")
])
if self.agentConfig['collect_ec2_tags']:
host_tags.extend(EC2.get_tags(self.agentConfig))
if self.agentConfig['collect_orchestrator_tags']:
host_docker_tags = MetadataCollector().get_host_tags()
if host_docker_tags:
host_tags.extend(host_docker_tags)
if host_tags:
payload['host-tags']['system'] = host_tags
# If required by the user, let's create the dd_check:xxx host tags
if self.agentConfig['create_dd_check_tags']:
app_tags_list = [
DD_CHECK_TAG.format(c.name)
for c in self.initialized_checks_d
]
app_tags_list.extend([
DD_CHECK_TAG.format(cname)
for cname in JMXFiles.get_jmx_appnames()
])
if 'system' not in payload['host-tags']:
payload['host-tags']['system'] = []
payload['host-tags']['system'].extend(app_tags_list)
GCE_tags = GCE.get_tags(self.agentConfig)
if GCE_tags is not None:
payload['host-tags'][GCE.SOURCE_TYPE_NAME] = GCE_tags
# Log the metadata on the first run
if self._is_first_run():
log.info(
"Hostnames: %s, tags: %s" %
(repr(self.hostname_metadata_cache), payload['host-tags']))
# Periodically send extra hosts metadata (vsphere)
# Metadata of hosts that are not the host where the agent runs, not all the checks use
# that
external_host_tags = []
if self._should_send_additional_data('external_host_tags'):
for check in self.initialized_checks_d:
try:
getter = getattr(check, 'get_external_host_tags')
check_tags = getter()
external_host_tags.extend(check_tags)
except AttributeError:
pass
if external_host_tags:
payload['external_host_tags'] = external_host_tags
# Periodically send agent_checks metadata
if self._should_send_additional_data('agent_checks'):
# Add agent checks statuses and error/warning messages
agent_checks = []
for check in check_statuses:
if check.instance_statuses is not None:
for i, instance_status in enumerate(
check.instance_statuses):
agent_checks.append((
check.name,
check.source_type_name,
instance_status.instance_id,
instance_status.status,
# put error message or list of warning messages in the same field
# it will be handled by the UI
instance_status.error or instance_status.warnings
or "",
check.service_metadata[i]))
else:
agent_checks.append(
(check.name, check.source_type_name, "initialization",
check.status, repr(check.init_failed_error)))
payload['agent_checks'] = agent_checks
payload[
'meta'] = self.hostname_metadata_cache # add hostname metadata
