| tkajinam | AttributeError: 'StreamFollower' object has no attribute 'zuul_log_id' | 02:11 |
|---|---|---|
| tkajinam | wonder if this is known | 02:11 |
| tkajinam | seen in multiple recent runs (eg. https://zuul.opendev.org/t/openstack/build/b701a4d1ecca40049d9c483949076aad ) | 02:11 |
| Clark[m] | tkajinam: I think it is a bug in a recent zuul change. I've made note of it in the zuul matrix room but can't debug or fix further at the moment. Looks like it occurs in cleanup code for threads that don't close quickly enough so it won't be a consistent failure | 04:37 |
| Clark[m] | tkajinam: can we cleanup your held node for heat functional testing? | 04:37 |
| *** jroll04 is now known as jroll0 | 07:28 | |
| frickler | https://review.opendev.org/c/zuul/zuul/+/939579 should be the fix. also the issue would only be triggered by changes that exceed the 1G log limit, so those would need looking at anyway? | 07:28 |
| tkajinam | Clark[m], yeah though I've not yet made real progress about debugging. I might request new one later. | 13:06 |
| corvus | Clark: frickler i'll restart zuul-web now with the fix. i think that does not only affect >1g output; but it does affect cases where the command is still running after the end (which could be related to a lot of output, or not). so, *something* is going wrong, but the severity of that something isn't clear. | 15:27 |
| corvus | sorry, not zuul-web, zuul-executor. | 15:27 |
| corvus | given the current load characteristics, i will do a hard-stop on all executors and rely on retries. | 15:28 |
| corvus | #status log restarted zuul-executors to pick up fix log stream message fix | 15:32 |
| opendevstatus | corvus: finished logging | 15:33 |
| Clark[m] | Thanks and ya I missed that the fix had already been proposed and landed | 15:39 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!