PG: Add wait events counts from pg_stat_activity #20588
Conversation
bonnefoa
force-pushed
the
bonnefoa/pg-wait-events
branch
3 times, most recently
from
4cb43e3 to
4fd8fc5
Compare
| @@ -557,6 +557,23 @@ def trim_leading_set_stmts(sql): | |||
| """.strip(), | |||
| } | |||
|
|
|||
| QUERY_PG_WAIT_EVENT_METRICS = { | |||
There was a problem hiding this comment.
We can do the same aggregations (and more) with samples in the UI. The issue is that we’re not collecting samples for background processes. I assume the reason for adding these metrics is to get visibility into them. Would it be better to include background processes in sampling instead?
There was a problem hiding this comment.
Yeah, the system processes are not visible and having visibility on the recovery process would have helped a lot during recent incidents. Also, wait events from sampling is not a metric and the 15 days retention make it hard to use.
If that can be done with samples, that would be great but AFAIK, the last attempt failed years ago? And that wouldn't solve the 15 days retention issues.
There was a problem hiding this comment.
If that can be done with samples, that would be great but AFAIK, the last attempt failed years ago?
That's feasible with not too much effort (we have to do some additional filtering), but we haven't prioritized it yet. Do you have a strong need for bg samples?
And that wouldn't solve the 15 days retention issues.
Understood - let's implement this regardless of the bg samples initiative.
Report wait_event in a postgresql.activity.wait_event metric. This metric is aggregated by user, db, app and backend_type. This metric will provide more visibility on system processes: Number of active autovacuum workers or parallel workers, the state of recovery process...
bonnefoa
force-pushed
the
bonnefoa/pg-wait-events
branch
from
4fd8fc5 to
f63ce79
Compare
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files
Flags with carried forward coverage won't be shown. Click here to find out more. 🚀 New features to boost your workflow:
|
What does this PR do?
Report wait_event in a
postgresql.activity.wait_eventmetric.Motivation
This metric is aggregated by user, db, app and backend_type. This metric will provide more visibility on system processes: Number of active autovacuum workers or parallel workers, the state of recovery process...
Review checklist (to be filled by reviewers)
qa/skip-qalabel if the PR doesn't need to be tested during QA.backport/<branch-name>label to the PR and it will automatically open a backport PR once this one is merged