| opendevreview | Michal Arbet proposed openstack/pbr master: Add support for local version as defined in PEP 440 https://review.opendev.org/c/openstack/pbr/+/942594 | 12:05 |
|---|---|---|
| acoles | swift-tox-py37 zuul job seems to have stopped working across the board https://zuul.opendev.org/t/openstack/builds?job_name=swift-tox-py37&project=openstack/swift | 14:35 |
| acoles | did we miss a change? | 14:35 |
| frickler | acoles: yes, bionic is gone for good | 14:39 |
| fungi | i don't think we have any maintained platforms that still ship cpython 3.7 by default | 14:40 |
| acoles | frickler: ack. we're so behind the curve :'( | 14:41 |
| frickler | the last passing change shows it tries to run on ubuntu-bionic https://zuul.opendev.org/t/openstack/build/99e69b3163b24b509afd00db1e0fa55d , though I surprised that this has been working for so long | 14:41 |
| acoles | thanks | 14:44 |
| fungi | frickler: acoles: i'm checking now to confirm, i think we may have only dropped ubuntu-bionic-arm64 images so far, the amd64 ubuntu-bionic label is still configured | 14:46 |
| frickler | oh, I somehow assumed that we dropped bionic with the move to zuul-launcher, but seems I mixed this up with xenial. so possible some other issue at play here | 14:53 |
| fungi | yeah, i've rechecked a change of mine that still runs a py37 job just for confirmation | 14:53 |
| clarkb | fungi: frickler acoles right bionic is still around for now but doesn't work with ansible 11 which is our new default | 14:54 |
| clarkb | only arm64 bionic was dropped from zuul-launcher due to lack of use | 14:55 |
| frickler | oh, that, right | 14:55 |
| clarkb | you can pin to ansible 9 for the short term. But once zuul drops ansible 9 support bionic and anything older will effectively become unsupported by the ci system | 14:55 |
| acoles | ok! thanks for digging in to that | 14:56 |
| fungi | yeah, going to need to make some hard choices about py37 support in a few of our tools at that point too | 14:56 |
| frickler | it's just weird that there is no log at all e.g. in https://zuul.opendev.org/t/openstack/build/9c8d1295e69d44adbd943bc65c8b994c , shouldn't at least stuff from the executor get logged anyway? | 14:57 |
| fungi | granted, i have a pile of packaging updates for bindep that are on hold until we can drop some older python support | 14:57 |
| clarkb | frickler: it crashes in fact gathering so you get logs on the executor but not within the job | 14:57 |
| clarkb | (the job logs require that we run a remote payload aiui) | 14:57 |
| *** darmach47 is now known as darmach4 | 14:58 | |
| fungi | that could be solved by splitting the log collection tasks so that you still gather executor-side localhost logs even when the remote logs can't be collected. we've discussed that possibility in the past | 14:58 |
| clarkb | I don't think there are any logs | 14:59 |
| clarkb | there are the logs generated from the run ansible-playbook side collected by the service itself. But no job side logs which are collectible by jobs | 14:59 |
| fungi | oh, so the base playbook output from the ansible callback is never even generated? | 15:00 |
| clarkb | right I think that ansible attempts to collect facts implicitly when it starts running and this occurs before any tasks you define run. This setup phase for fact gathering crashes so ansible exits with rc 2 and the job fails | 15:02 |
| *** sfinucan is now known as stephenfin | 17:02 | |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!