ajwerner
ajwerner
>So do we know if this issue is another cause of https://github.com/cockroachdb/cockroach/issues/82034 and https://github.com/cockroachdb/cockroach/issues/80764? Yes, I believe we do know that this is a cause of #80764 (https://github.com/cockroachdb/cockroach/issues/80764#issuecomment-1206720146).
An approach we've taken elsewhere is to classify the errors which can be swallowed safely. We don't have a unified way to do this presently. In the cases I'm thinking...
> Is it ok/desirable that this is a cluster setting (aka, not permitting decisionmaking on a per session basis)? I'll go one further: this needs to be a session variable....
I don't think we need such a setting now that we have database level session defaults
TFTR! bors r+
My theory on why the stats job gets slower over time is that for each iteration of the loop, it scans over all not-yet-GC'd data from all previous iterations of...
This is on my list to look at tonight or tomorrow morning.
Yeah, I'm on board with deleting the test.
I think what's going on is that the job is failing due to https://github.com/cockroachdb/cockroach/issues/84073 and then it backs off. You can `PAUSE JOB 785222573203750976` / `RESUME JOB 785222573203750976` to reset...