Page MenuHomePhabricator

check_icinga alerts lack metadata in VictorOps
Open, Needs TriagePublic

Description

In VictorOps, alerts from Icinga itself have 'service' and 'entityDisplayName' fields present in the VictorOps API, which gives us a short alert summary to display on Klaxon -- for example:

{
  "alertCount": 1,
  "currentPhase": "RESOLVED",
  "entityDisplayName": "Host ncredir-lb.esams.wikimedia.org_ipv6 - PING  - Packet loss = 100%",
  "entityId": "Host ncredir-lb.esams.wikimedia.org_ipv6 - PING  - Packet loss = 100%",
  "service": "Host ncredir-lb.esams.wikimedia.org_ipv6 - PING  - Packet loss = 100%",

However, alerts from check_icinga have none of these fields:

{
    "entityDisplayName": "",
    "entityId": "",
    // service field omitted entirely

An example alert produced (as seen by VO) is at https://portal.victorops.com/ui/wikimedia/incident/796/details

Ideally they would have these fields populated (but I'm not sure what this entails in the email parsing).

Event Timeline

Change 664056 had a related patch set uploaded (by CDanis; owner: CDanis):
[operations/software/klaxon@master] Fix cases of incidents missing the 'service' field

https://gerrit.wikimedia.org/r/664056

Change 664056 merged by jenkins-bot:
[operations/software/klaxon@master] Fix cases of incidents missing the 'service' field

https://gerrit.wikimedia.org/r/664056