def api_escalation_alarm(self, request, id):
alarm = get_alarm(id)
if alarm.status == "A":
AlarmEscalation.watch_escalations(alarm)
return {"status": True}
else:
return {"status": False, "error": "The alarm is not active at the moment"}
def api_escalation_alarm(self, request, id):
alarm = get_alarm(id)
if alarm.status == "A":
AlarmEscalation.watch_escalations(alarm)
return {'status': True}
else:
return {
'status': False,
'error': 'The alarm is not active at the moment'
}
def check_close_consequence(alarm_id):
logger.info("[%s] Checking close", alarm_id)
alarm = get_alarm(alarm_id)
if alarm is None:
logger.info("[%s] Missing alarm, skipping", alarm_id)
return
if alarm.status == "C":
logger.info("[%s] Alarm is closed. Check passed", alarm_id)
return
# Detach root
logger.info("[%s] Alarm is active. Detaching root", alarm_id)
alarm.root = None
alarm.log_message("Detached from root for not recovered", to_save=True)
metrics["detached_root"] += 1
# Trigger escalations
AlarmEscalation.watch_escalations(alarm)
def alarm_escalation(alarm, mo, ctx):
now = datetime.datetime.now()
for esc in AlarmEscalation.get_class_escalations(alarm['alarm_class']):
for e_item in esc.escalations:
# Check administrative domain
if (e_item.administrative_domain and
e_item.administrative_domain.id not in mo.object.administrative_domain.get_path()):
continue
# Check severity
if e_item.min_severity and alarm['severity'] < e_item.min_severity:
continue
# Check selector
if e_item.selector and not SelectorCache.is_in_selector(mo.object, e_item.selector):
continue
logger.info(
"%s Watch for %s after %s seconds",
alarm['alarm_class'], esc.name, e_item.delay
)
et = now + datetime.timedelta(seconds=e_item.delay)
if et > now:
delay = (et - now).total_seconds()
else:
delay = None
if e_item.notification_group:
a = ActiveAlarm.objects.filter(managed_object=mo.object, vars__path=alarm['vars']["path"]).first()
if a:
logger.info("Alarm already sending")
break
subject = e_item.template.render_subject(**ctx)
body = e_item.template.render_body(**ctx)
logger.debug("Notification message:\nSubject: %s\n%s", subject, body)
call_later(
"noc.custom.handlers.thresholds.thresholdsnotification.threshold_escalation",
delay=delay,
scheduler="scheduler",
notification_group_id=e_item.notification_group.id,
subject=subject,
body=body
)
return {"notification_group": e_item.notification_group,
"clear_template": e_item.clear_template,
"message": "Sending message to : %s" % e_item.notification_group.name}
#
if e_item.stop_processing:
logger.debug("Stopping processing")
break
def escalate(alarm_id, escalation_id, escalation_delay, *args, **kwargs):
def log(message, *args):
msg = message % args
logger.info("[%s] %s", alarm_id, msg)
alarm.log_message(msg, to_save=True)
def summary_to_list(summary, model):
r = []
for k in summary:
p = model.get_by_id(k.profile)
if not p or getattr(p, "show_in_summary", True) is False:
continue
r += [{
"profile": p.name,
"summary": k.summary,
"order": (getattr(p, "display_order", 100), -k.summary),
}]
return sorted(r, key=operator.itemgetter("order"))
logger.info("[%s] Performing escalations", alarm_id)
alarm = get_alarm(alarm_id)
if alarm is None:
logger.info("[%s] Missing alarm, skipping", alarm_id)
metrics["escalation_missed_alarm"] += 1
return
if alarm.status == "C":
logger.info("[%s] Alarm is closed, skipping", alarm_id)
metrics["escalation_already_closed"] += 1
return
if alarm.root:
log("[%s] Alarm is not root cause, skipping", alarm_id)
metrics["escalation_alarm_is_not_root"] += 1
return
#
escalation = AlarmEscalation.get_by_id(escalation_id)
if not escalation:
log("Escalation %s is not found, skipping", escalation_id)
metrics["escalation_not_found"] += 1
return
if alarm.managed_object.tt_system:
sample = alarm.managed_object.tt_system.telemetry_sample
else:
sample = PARENT_SAMPLE
with Span(client="escalator", sample=sample) as ctx:
alarm.set_escalation_context()
# Evaluate escalation chain
mo = alarm.managed_object
for a in escalation.escalations:
if a.delay != escalation_delay:
continue # Try other type
# Check administrative domain
if a.administrative_domain and a.administrative_domain.id not in alarm.adm_path:
continue
# Check severity
if a.min_severity and alarm.severity < a.min_severity:
continue
# Check selector
if a.selector and not SelectorCache.is_in_selector(mo, a.selector):
continue
# Check time pattern
if a.time_pattern and not a.time_pattern.match(alarm.timestamp):
continue
# Render escalation message
if not a.template:
log("No escalation template, skipping")
continue
# Check global limits
# @todo: Move into escalator service
# @todo: Process per-ttsystem limits
ets = datetime.datetime.now() - datetime.timedelta(
seconds=config.escalator.ets)
ae = ActiveAlarm._get_collection().count_documents(
{"escalation_ts": {
"$gte": ets
}})
ae += ArchivedAlarm._get_collection().count_documents(
{"escalation_ts": {
"$gte": ets
}})
if ae >= config.escalator.tt_escalation_limit:
logger.error(
"Escalation limit exceeded (%s/%s). Skipping",
ae,
config.escalator.tt_escalation_limit,
)
metrics["escalation_throttled"] += 1
alarm.set_escalation_error(
"Escalation limit exceeded (%s/%s). Skipping" %
(ae, config.escalator.tt_escalation_limit))
return
# Check whether consequences has escalations
cons_escalated = sorted(alarm.iter_escalated(),
key=operator.attrgetter("timestamp"))
affected_objects = sorted(alarm.iter_affected(),
key=operator.attrgetter("name"))
#
segment = alarm.managed_object.segment
if segment.is_redundant:
uplinks = alarm.managed_object.data.uplinks
lost_redundancy = len(uplinks) > 1
affected_subscribers = summary_to_list(
segment.total_subscribers, SubscriberProfile)
affected_services = summary_to_list(segment.total_services,
ServiceProfile)
else:
lost_redundancy = False
affected_subscribers = []
affected_services = []
#
ctx = {
"alarm":
alarm,
"affected_objects":
affected_objects,
"cons_escalated":
cons_escalated,
"total_objects":
summary_to_list(alarm.total_objects, ManagedObjectProfile),
"total_subscribers":
summary_to_list(alarm.total_subscribers, SubscriberProfile),
"total_services":
summary_to_list(alarm.total_services, ServiceProfile),
"tt":
None,
"lost_redundancy":
lost_redundancy,
"affected_subscribers":
affected_subscribers,
"affected_services":
affected_services,
}
# Escalate to TT
if a.create_tt and mo.can_escalate():
tt_id = None
if alarm.escalation_tt:
log("Already escalated with TT #%s", alarm.escalation_tt)
else:
pre_reason = escalation.get_pre_reason(mo.tt_system)
active_maintenance = Maintenance.get_object_maintenance(mo)
if active_maintenance:
for m in active_maintenance:
log(
"Object is under maintenance: %s (%s-%s)",
m.subject,
m.start,
m.stop,
)
metrics["escalation_stop_on_maintenance"] += 1
elif pre_reason is not None:
subject = a.template.render_subject(**ctx)
body = a.template.render_body(**ctx)
logger.debug(
"[%s] Escalation message:\nSubject: %s\n%s",
alarm_id, subject, body)
log("Creating TT in system %s", mo.tt_system.name)
tts = mo.tt_system.get_system()
try:
try:
tt_id = tts.create_tt(
queue=mo.tt_queue,
obj=mo.tt_system_id,
reason=pre_reason,
subject=subject,
body=body,
login="******",
timestamp=alarm.timestamp,
)
except TemporaryTTError as e:
metrics["escalation_tt_retry"] += 1
log(
"Temporary error detected. Retry after %ss",
RETRY_TIMEOUT)
mo.tt_system.register_failure()
Job.retry_after(get_next_retry(), str(e))
ctx["tt"] = "%s:%s" % (mo.tt_system.name, tt_id)
alarm.escalate(
ctx["tt"],
close_tt=a.close_tt,
wait_tt=ctx["tt"] if a.wait_tt else None,
)
if tts.promote_group_tt and a.promote_group_tt:
# Create group TT
log("Promoting to group tt")
gtt = tts.create_group_tt(
tt_id, alarm.timestamp)
# Append affected objects
for ao in alarm.iter_affected():
if ao.can_escalate(True):
if ao.tt_system == mo.tt_system:
log(
"Appending object %s to group tt %s",
ao.name, gtt)
try:
tts.add_to_group_tt(
gtt, ao.tt_system_id)
except TTError as e:
alarm.set_escalation_error(
"[%s] %s" %
(mo.tt_system.name, e))
else:
log(
"Cannot append object %s to group tt %s: Belongs to other TT system",
ao.name,
gtt,
)
else:
log(
"Cannot append object %s to group tt %s: Escalations are disabled",
ao.name,
gtt,
)
metrics["escalation_tt_create"] += 1
except TTError as e:
log("Failed to create TT: %s", e)
metrics["escalation_tt_fail"] += 1
alarm.log_message("Failed to escalate: %s" % e,
to_save=True)
alarm.set_escalation_error("[%s] %s" %
(mo.tt_system.name, e))
else:
log("Cannot find pre reason")
metrics["escalation_tt_fail"] += 1
if tt_id and cons_escalated:
# Notify consequences
for ca in cons_escalated:
c_tt_name, c_tt_id = ca.escalation_tt.split(":")
cts = TTSystem.get_by_name(c_tt_name)
if cts:
tts = cts.get_system()
try:
log("Appending comment to TT %s", tt_id)
tts.add_comment(c_tt_id,
body="Covered by TT %s" %
tt_id,
login="******")
metrics["escalation_tt_comment"] += 1
except NotImplementedError:
log(
"Cannot add comment to %s: Feature not implemented",
ca.escalation_tt,
)
metrics["escalation_tt_comment_fail"] += 1
except TTError as e:
log("Failed to add comment to %s: %s",
ca.escalation_tt, e)
metrics["escalation_tt_comment_fail"] += 1
else:
log(
"Failed to add comment to %s: Invalid TT system",
ca.escalation_tt)
metrics["escalation_tt_comment_fail"] += 1
# Send notification
if a.notification_group and mo.can_notify():
subject = a.template.render_subject(**ctx)
body = a.template.render_body(**ctx)
logger.debug("[%s] Notification message:\nSubject: %s\n%s",
alarm_id, subject, body)
log("Sending notification to group %s",
a.notification_group.name)
a.notification_group.notify(subject, body)
alarm.set_clear_notification(a.notification_group,
a.clear_template)
metrics["escalation_notify"] += 1
#
if a.stop_processing:
logger.debug("Stopping processing")
break
nalarm = get_alarm(alarm_id)
if nalarm and nalarm.status == "C":
nalarm.log_message(
"Alarm has been closed during escalation. Try to deescalate")
logger.info(
"[%s] Alarm has been closed during escalation. Try to deescalate",
alarm.id)
metrics["escalation_closed_while_escalated"] += 1
if tt_id and not nalarm.escalation_tt:
nalarm.escalation_ts = datetime.datetime.now()
nalarm.escalation_tt = "%s:%s" % (mo.tt_system.name, tt_id)
nalarm.save()
if not nalarm.escalation_close_ts and not nalarm.escalation_close_error:
notify_close(
alarm_id=alarm_id,
tt_id=nalarm.escalation_tt,
subject="Closing",
body="Closing",
notification_group_id=alarm.clear_notification_group.id
if alarm.clear_notification_group else None,
close_tt=alarm.close_tt,
)
elif nalarm == "A" and not nalarm.escalation_tt and tt_id:
logger.error("[%s] Alarm without escalation TT: %s", alarm.id,
tt_id)
logger.info("[%s] Escalations loop end", alarm_id)
def run_alarm(self, alarm):
AlarmEscalation.watch_escalations(alarm)
def raise_alarm(self, r, e):
managed_object = self.eval_expression(r.managed_object, event=e)
if not managed_object:
self.logger.info("Empty managed object, ignoring")
return
# @todo: Make configurable
if not managed_object.is_managed:
self.logger.info(
"Managed object is not managed. Do not raise alarm")
return
if e.managed_object.id != managed_object.id:
metrics["alarm_change_mo"] += 1
self.logger.info("Changing managed object to %s",
managed_object.name)
discriminator, vars = r.get_vars(e)
if r.unique:
assert discriminator is not None
a = ActiveAlarm.objects.filter(
managed_object=managed_object.id,
discriminator=discriminator).first()
if not a:
# Try to reopen alarm
a = ArchivedAlarm.objects.filter(
managed_object=managed_object.id,
discriminator=discriminator,
control_time__gte=e.timestamp).first()
if a:
# Reopen alarm
self.logger.info("[%s|%s|%s] %s reopens alarm %s(%s)",
e.id, managed_object.name,
managed_object.address,
e.event_class.name, a.alarm_class.name,
a.id)
a = a.reopen("Reopened by disposition rule '%s'" %
r.u_name)
metrics["alarm_reopen"] += 1
if a:
# Active alarm found, refresh
self.logger.info(
"[%s|%s|%s] Contributing event %s to active alarm %s(%s)",
e.id, managed_object.name, managed_object.address,
e.event_class.name, a.alarm_class.name, a.id)
# Contribute event to alarm
e.contribute_to_alarm(a)
if e.timestamp < a.timestamp:
# Set to earlier date
a.timestamp = e.timestamp
a.save()
elif e.timestamp > a.last_update:
# Refresh last update
a.last_update = e.timestamp
a.save()
metrics["alarm_contribute"] += 1
return
# Calculate alarm coverage
summary = ServiceSummary.get_object_summary(managed_object)
summary["object"] = {managed_object.object_profile.id: 1}
#
severity = max(ServiceSummary.get_severity(summary), 1)
self.logger.info("[%s|%s|%s] %s: Calculated alarm severity is: %s",
e.id, managed_object.name, managed_object.address,
r.u_name, severity)
# Create new alarm
a = ActiveAlarm(
timestamp=e.timestamp,
last_update=e.timestamp,
managed_object=managed_object.id,
alarm_class=r.alarm_class,
severity=severity,
vars=vars,
discriminator=discriminator,
direct_services=SummaryItem.dict_to_items(summary["service"]),
direct_subscribers=SummaryItem.dict_to_items(
summary["subscriber"]),
total_objects=ObjectSummaryItem.dict_to_items(summary["object"]),
total_services=SummaryItem.dict_to_items(summary["service"]),
total_subscribers=SummaryItem.dict_to_items(summary["subscriber"]),
log=[
AlarmLog(timestamp=datetime.datetime.now(),
from_status="A",
to_status="A",
message="Alarm risen from event %s(%s) by rule '%s'" %
(str(e.id), str(e.event_class.name), r.u_name))
],
opening_event=e.id)
a.save()
e.contribute_to_alarm(a)
self.logger.info("[%s|%s|%s] %s raises alarm %s(%s): %r", e.id,
managed_object.name, managed_object.address,
e.event_class.name, a.alarm_class.name, a.id, a.vars)
metrics["alarm_raise"] += 1
self.correlate(r, a)
# Notify about new alarm
if not a.root:
a.managed_object.event(
a.managed_object.EV_ALARM_RISEN, {
"alarm": a,
"subject": a.subject,
"body": a.body,
"symptoms": a.alarm_class.symptoms,
"recommended_actions": a.alarm_class.recommended_actions,
"probable_causes": a.alarm_class.probable_causes
},
delay=a.alarm_class.get_notification_delay())
# Gather diagnostics when necessary
AlarmDiagnosticConfig.on_raise(a)
# Watch for escalations, when necessary
if config.correlator.auto_escalation and not a.root:
AlarmEscalation.watch_escalations(a)