When we restart Burrow instances there is a brief amount of time in which the metrics go awol causing alarms to fire, especially the Mirror Maker ones. Ideally Burrow should be able to restart without creating any impact to metrics.
Description
Description
Details
Details
Subject | Repo | Branch | Lines +/- | |
---|---|---|---|---|
profile::kafka::mirror::alerts: tune retries and its interval | operations/puppet | production | +2 -0 |
Event Timeline
Comment Actions
Change 438243 had a related patch set uploaded (by Elukey; owner: Elukey):
[operations/puppet@production] profile::kafka::mirror::alerts: tune retries and its interval
Comment Actions
Change 438243 merged by Elukey:
[operations/puppet@production] profile::kafka::mirror::alerts: tune retries and its interval
Comment Actions
Tested one burrow restart manually, alarms showed up in icinga but cleared up before firing an alarm. This is a good compromise for the moment, we'll revise the alarm when a more powerful and flexible prometheus check metrics script will be available.