| 12 Oct 2023 |
 delroth | I agree that we should at the very least try to measure how often these problems happen before making any decision, but I don't think a low rate of false positives necessarily needs to be a blocker - it would still be a massive improvement | 16:06:01 |
| delroth | (imo) | 16:06:03 |
 Lily Foster | Yeah I definitely want to get measurements for how often those transient failures happen, before changing anything. Stuff like build timeout, OOM, and nix daemon/builder error should be technically distinguishable from derivation-builder-error, though, so ideally we'd have the ability to do both "neutral" and "failure" conditions depending on how it failed
I think there would be value in having at least some scoped conditions surrounding builds get a red X though, because I see a lot of PRs get merged with failing builds or tests because failures of, e.g. build timeouts due to rebuilding llvm or something, are not clearly distinguishable from failures due to the build just not actually working, without diving into the logs on the ofborg website
I'd have to see where the decision was originally made, but I feel like, too, if there's enough of the community really wanting the red X now, we could decide to try it temporarily and roll it back if it does in fact turn out to be worse than status quo (even though that is admittedly hard to measure) | 17:53:51 |
 @adam:robins.wtf | I've definitely had PRs merged that were failing a build, and nobody noticed. It seems like some of the concerns (timeouts, OOM, hardware failure) are common across other CI systems, yet it is also quite common for them to mark builds as failing through the UI. | 18:14:03 |
| @adam:robins.wtf | Is the difference just a lack of ability to retry? | 18:15:55 |
| Lily Foster | In reply to @adam:robins.wtf
Is the difference just a lack of ability to retry? You can actually have ofborg retry-ish now by requeueing the same attrs (@ofborg build attr1 attr1.tests attr2 attr2.tests ...) | 18:18:57 |
| Lily Foster | It would be nice if there was a better way to request a retry of one build specifically though | 18:19:12 |
| Lily Foster | In reply to @adam:robins.wtf
Is the difference just a lack of ability to retry? * You can actually have ofborg retry-ish now by requeueing the same attrs (commenting @ofborg build attr1 attr1.tests attr2 attr2.tests ...) | 18:20:12 |
| Lily Foster | * You can actually have ofborg retry-ish now by requeueing the same attrs (commenting @ofborg build attr1 attr1.tests attr2 attr2.tests ... on the PR) | 18:20:20 |
|  Andreas Schrägle left the room. | 22:24:38 |
