damccorm commented on PR #41205: URL: https://github.com/apache/beam/pull/41205#issuecomment-6805376684
> I see the bot starts to bump timeout. This may not be a good pattern in general, at least the bot should justify the distribution of current latency (medium and 95% percentile latency) As a rule I agree (and I've had it rewrite a couple PRs that were, for example, bumping test timeouts). In cases like this, though, I think it is actually warranted. Specific criteria that I think justifies it: 1) It is infra off of the Beam path (e.g. we're not releasing something that got slower) 2) It is not actually something we can feasibly control without meaningful and potentially unhelpful change (this isn't our tooling letting us down) 3) We've seen it multiple times within this week (for infra flakes, I require this). In this case, it is the same failure across multiple jobs which is a little more tenuous, but these also don't run that often. So the failure rate is at least meaningful. 4) The blast radius of a higher timeout is small (worst case, postcommit CI jobs fail after 5 minutes instead of 1, and the issue is more clearly not a timeout) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
