| *** rlandy is now known as rlandy|out | 00:52 | |
| *** dasm|off is now known as Guest149 | 03:32 | |
| *** yadnesh|away is now known as yadnesh | 05:05 | |
| *** yadnesh is now known as yadnesh|afk | 08:02 | |
| *** yadnesh|afk is now known as yadnesh | 08:38 | |
| *** ykarel_ is now known as ykarel|lunch | 08:56 | |
| *** ykarel|lunch is now known as ykarel | 10:20 | |
| opendevreview | Slawek Kaplonski proposed openstack/project-config master: Update Grafana's dasboard for Neutron https://review.opendev.org/c/openstack/project-config/+/868068 | 11:18 |
|---|---|---|
| *** rlandy|out is now known as rlandy | 11:39 | |
| *** kopecmartin|sick is now known as kopecmartin | 11:55 | |
| *** Guest149 is now known as dasm | 12:30 | |
| *** JasonF is now known as JayF | 15:04 | |
| *** yadnesh is now known as yadnesh|away | 15:12 | |
| *** dviroel is now known as dviroel|lunch | 15:14 | |
| *** JayF is now known as JasonF | 15:23 | |
| *** JasonF is now known as JayF | 15:23 | |
| *** dviroel|lunch is now known as dviroel | 16:25 | |
| xek | fungi: hey, can you provide some pointers as to why the depends-on doesn't work in this case: https://zuul.opendev.org/t/openstack/build/4cae7eaba1254e73838e54b782897ef3 ? | 19:54 |
| xek | fungi: the dependent change from keystoneauth is pulled into the git repository in /opt/git/opendev.org/openstack/keystoneauth | 19:55 |
| xek | but I'm not sure how it is or should be utilized inside the py38 virtualenv | 19:56 |
| xek | the tests are failing on a missing argument which is added by the keystoneauth patch | 19:58 |
| fungi | xek: i'll look in a moment, but keep in mind that jobs need to be designed to install projects from source on disk to take advantage of that | 20:00 |
| fungi | so for example it works with devstack jobs because devstack installs most openstack projects from source | 20:00 |
| xek | fungi: this is a standard unit test job from the openstack-python3-jobs template | 20:01 |
| fungi | xek: and you normally expect that job to install releases of dependencies like keystoneauth from packaged releases right? | 20:02 |
| fungi | not from unreleased git source trees like an integration testing job would? | 20:03 |
| xek | but yeah, from the logs it looks like it's being installed from pypi, but I looked at other jobs from other repos and the output looks the same | 20:03 |
| fungi | typically you'd have a separate keystoneauth integration job which expects ksa from git source instead of from a release. if zuul allowed you to merge a change to barbican which depended on the unreleased state of ksa that would be bad, right? | 20:04 |
| xek | fungi: (I was just looking at this job result: https://zuul.opendev.org/t/openstack/build/7129a79d8bca443886e4bd8c78178e36/log/job-output.txt) | 20:04 |
| fungi | especially if a new version of barbican got tagged and then wouldn't work for anyone because it still depended on unreleased ksa | 20:04 |
| xek | fungi: what you're saying makes sense, I just expected that there might be a task that would uninstall the pip package and install package from the git repository in the test environment | 20:07 |
| fungi | and yes, the osc unit test job you linked is similar. it depends on packaged releases of keystoneauth not an unreleased git source tree | 20:07 |
| xek | fungi: but it wouldn't merge though with the depends-on that did not merge? | 20:08 |
| fungi | xek: you can do that, there's a magic "siblings" behavior in the tox parent job which acts like you describe for any project declared as a required-project in the job definition | 20:08 |
| xek | fungi: yeah, I saw that, do you have examples where this is used? | 20:09 |
| fungi | xek: even if it waited for the ksa change to merge before merging (which yes it would) that doesn't mean there's anything preventing barbican from releasing with code that only works with keystoneauth unreleased | 20:09 |
| xek | fungi: nevermind, I see an example in keystone | 20:09 |
| xek | fungi: right | 20:10 |
| xek | fungi: ok, thanks for clarifying this :) | 20:10 |
| fungi | so it makes sense to keep the current unit test job working the way it does now in order to prevent you from merging your change until there's a release of ksa which works with it, but you can also add a very similar job which uses ksa from source if you like so that you can see it passing with the depends-on even if the normal job is still failing and blocking you from merging the | 20:11 |
| fungi | change | 20:11 |
| fungi | xek: however, you might consider making that ksa-source job non-voting so that if an incompatible change merges to ksa it doesn't block development on barbican (but you'd still want to get up with the ksa devs right away to get it fixed so they don't release a version which breaks you, of course) | 20:13 |
| clarkb | the big remaining todo on the nox work is to add siblings support to it | 20:20 |
| clarkb | I'm sorting out lunch, but one of the things I need to do after lunch is delte the iweb mirror srver and send an email to them saying we're off of it | 20:20 |
| clarkb | please chime in now if this should be delayed for some reason :) | 20:21 |
| fungi | i know of no reason the iweb deletion should ever be forgot | 20:26 |
| fungi | or something like that | 20:26 |
| fungi | (remember remember the nineteenth of december...) | 20:27 |
| opendevreview | Merged openstack/project-config master: Update Grafana's dasboard for Neutron https://review.opendev.org/c/openstack/project-config/+/868068 | 20:58 |
| *** dviroel is now known as dviroel|out | 21:00 | |
| *** rlandy is now known as rlandy|bbl | 21:42 | |
| *** dasm is now known as dasm|off | 23:40 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!