| frickler | JayF: yes, I think with USE_VENV you're mostly on your own. what's the exact reason you want to use that? it isn't too clear to me from your commit | 07:12 |
|---|---|---|
| frickler | kopecmartin: gmann: I'm getting this failure https://paste.opendev.org/show/bEDQ0paQ1vib4Ktu3sqD/ when running tempest in an environment without glance. seems this except clause should be extended to include this error? https://opendev.org/openstack/tempest/src/branch/master/tempest/scenario/test_unified_limits.py#L59 | 10:57 |
| kopecmartin | frickler: this error seems different than those we want to catch by the except block, contextually different, ... does your tempest.conf have glance = True or False? | 12:48 |
| kopecmartin | if it's False, then the test should be skipped, that would mean a bug but not in that except block | 12:48 |
| JayF | frickler: tbh I started it because I mixed up use_venv and global_venv and now that my change is already almost done, I was going to carry it over the finish line. | 13:39 |
| JayF | Earlier versions of that patch were oriented towards allowing us to install our QA helpers, like vpdu or vvmc in a separate virtual environment, but that was complicated for other reasons | 13:40 |
| frickler | kopecmartin: glance=false, yes, seems the tests themselves would get skipped, but not the class setup? | 14:00 |
| kopecmartin | maybe not, .... in api tests, we have something like this: https://opendev.org/openstack/tempest/src/commit/2731314d4423b6b21cac95d7c736527b62aa49f8/tempest/api/image/base.py#L33-L38 | 14:03 |
| kopecmartin | if you have an active env, can you try to add that to the class of that scenario test? | 14:03 |
| kopecmartin | maybe there has to be "skip_checks" method present to skip everything from the class ( i don't remember now the execution order ) | 14:04 |
| frickler | not easy to test right away, but I can look into that, thx for the pointers | 14:04 |
| kopecmartin | based on this, it should help: https://opendev.org/openstack/tempest/src/commit/2731314d4423b6b21cac95d7c736527b62aa49f8/tempest/test.py#L64-L67 | 14:05 |
| gmann | kopecmartin: frickler: we should be checking the glance service available or not for that test and skip | 17:03 |
| gmann | skip_checks is very first thing we run to skip the test class's tests early | 17:03 |
| opendevreview | Ilya Popov proposed openstack/tempest master: Wait for VM ACTIVE state is added for fixed network case https://review.opendev.org/c/openstack/tempest/+/934165 | 18:06 |
| opendevreview | Ghanshyam proposed openstack/tempest master: Check for glance service for glance tests https://review.opendev.org/c/openstack/tempest/+/934168 | 18:12 |
| gmann | frickler: kopecmartin ^^ this should fix the glance things | 18:12 |
| opendevreview | Ghanshyam proposed openstack/tempest master: Rename test_unified_limits.py test file https://review.opendev.org/c/openstack/tempest/+/934169 | 18:17 |
| gmann | frickler: kopecmartin ^^ also renaming the file name which confuse about what service tests this is | 18:17 |
| gmann | melwitt: ^^ is it fine to rename it? or you want to extend the same file for other services unified limit tests but i see we have separate for test_compute_unified_limits.py? | 18:19 |
| opendevreview | Ilya Popov proposed openstack/tempest master: For test VM validation resource were added. https://review.opendev.org/c/openstack/tempest/+/933308 | 18:20 |
| melwitt | gmann: I believe dansmith wrote test_unified_limits.py so the question would be for him :) | 18:59 |
| melwitt | and I named mine test_compute_unified_limits.py to differentiate. I figured it would be nicer to have them separate and smaller | 18:59 |
| dansmith | gmann: aren't the scenario tests mostly for cross-service things? I guess I would expect to just add the compute stuff in there alongside it for consistency | 19:00 |
| dansmith | but if not, rename is fine with me | 19:00 |
| gmann | dansmith: it is but thing is it specifically test the image service (need glance to be present during resource_setup). I agree to make a generic scenario test file but that might end up with skip the whole test class if all services are present. | 19:02 |
| gmann | and that way we might skip compute unified limit test if glance is not present | 19:02 |
| dansmith | gmann: right but can't you just add another class into that same file? | 19:02 |
| dansmith | seems like that is best to keep them somewhat consistent style and stuff | 19:02 |
| gmann | dansmith: ++ yeah that is possible. we can do that way too | 19:03 |
| dansmith | I named the class ImageQuotaTest for that reason | 19:03 |
| melwitt | gmann: if you want to collapse the compute tests into the same file in a different class, that is also fine with me. I don't care that much about it | 19:03 |
| dansmith | anyway, that was just my intent anyway.. I'm not super opinionated, I just wanted to explain why I assumed we'd put all unified limits tests in that file and thus the name :) | 19:04 |
| gmann | melwitt: dansmith: if that is the plan what dansmith mentioned, I think merging test_compute_unified_limits.py into existing test_unified_limits.py make sense | 19:04 |
| gmann | let me do this way and it will be clear to have a unified file for unified limit tests | 19:07 |
| gmann | or if we want to test cross services unified limit together | 19:08 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!