You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The smallest non-zero aggregation boundary for request duration metrics (such as gatekeeper_validation_request_duration_seconds_bucket) is 5 seconds (up from 1 millisecond). We are alerting if Gatekeeper has a high 99th percentile request latency, so this course aggregation is causing our alerts to fire.
Curling the Gatekeeper metrics endpoint gives the following:
I tried 7349e22 (current tip of release-3.15) and the gatekeeper_validation_request_duration histogram is totally missing from the metrics (also gatekeeper_validation_request_count_total went away). I think I'm getting better results with c68a029 (v3.16.0-beta.1).
Actually I think I used the wrong image when I tried to test 7349e22 so I think the fix is probably going to work for me. Thanks!
Yeah, 7349e22 is working well for me. Thanks again!
What steps did you take and what happened:
The smallest non-zero aggregation boundary for request duration metrics (such as
gatekeeper_validation_request_duration_seconds_bucket
) is 5 seconds (up from 1 millisecond). We are alerting if Gatekeeper has a high 99th percentile request latency, so this course aggregation is causing our alerts to fire.Curling the Gatekeeper metrics endpoint gives the following:
What did you expect to happen:
Request duration metric aggregation boundaries to be as specified here:
gatekeeper/pkg/webhook/stats_reporter.go
Line 83 in 6e2dba5
Anything else you would like to add:
I guess this issue was introduced in this PR: #3011
Environment:
kubectl version
):The text was updated successfully, but these errors were encountered: