| *** diablo_rojo is now known as Guest7830 | 03:11 | |
| *** sshnaidm|afk is now known as sshnaidm | 09:20 | |
| *** redrobot8 is now known as redrobot | 10:20 | |
| frickler | gtema: could you have another look at https://review.opendev.org/c/openstack/python-openstackclient/+/736973 pls? I had messed up pep8. slaweq this may also be interesting for you | 10:22 |
|---|---|---|
| gtema | sure, but first need to find reason for one functest failure we currently have | 10:23 |
| frickler | gtema: I haven't seen that, do you have a pointer? | 10:24 |
| gtema | https://review.opendev.org/c/openstack/python-openstackclient/+/820512 | 10:24 |
| frickler | volumes once again ... sigh ... will take a look | 10:26 |
| gtema | thks a lot | 10:26 |
| frickler | gtema: so I can reproduce locally and if I revert https://review.opendev.org/c/openstack/python-openstackclient/+/817989/8 the issue is gone. maybe try to revert that for now? | 13:15 |
| gtema | we thought yesterday on marking the test as expected to fail | 13:15 |
| gtema | https://review.opendev.org/c/openstack/python-openstackclient/+/819473/1 added the functest itself (was not tested before at all) | 13:16 |
| frickler | gtema: my current suspicion is that the test is good, but 817989 broken | 13:18 |
| frickler | or is might be uncovering some issue in nova | 13:18 |
| gtema | yeah, everything is possible | 13:19 |
| frickler | so 817989 is broken for sure. nova returns a 404 for the delete_volume_attachment(), but the code ignores that instead of raising an error | 13:29 |
| frickler | gtema: stephenfin: ^^ | 13:31 |
| frickler | e.g. 15:08:07.785042 in https://210be1f5cbeabbdaba80-eeccdc6968a6b16fa4ca2e3ee7c1080d.ssl.cf2.rackcdn.com/820512/2/check/osc-functional-devstack-tips/4af9a0d/controller/logs/screen-n-api.txt | 13:34 |
| frickler | delete_volume_attachment() expects the ID of the attachment, while the OSC patch hands it the volume ID | 13:41 |
| frickler | hmm, no, ignore that | 13:42 |
| frickler | the sdk code is broken it seems, it uses the attachment_id for the _delete call instead of the volume id | 13:58 |
| gtema | ouch, bad | 13:59 |
| frickler | this may be somehow related to https://review.opendev.org/c/openstack/openstacksdk/+/817997/1/openstack/compute/v2/volume_attachment.py , but I'm kinda confused about it still. I do remember trying to make sense of the nova api-ref there | 14:02 |
| frickler | I also don't understand how it could pass CI if that were true | 14:03 |
| gtema | and so the loop closes :-) | 14:05 |
| gtema | we started failing with the raising of openstacksdk | 14:05 |
| frickler | oh, wow, this is even weirder. with old microversion, nova returns the volume_id as attachment_id. with 2.89, it returns the real attachment_id. | 14:14 |
| frickler | so that's actually the non-voting failure for 817997 https://zuul.opendev.org/t/openstack/build/1f7a3ab5e8b942c2b7b58b9ec2e27367 | 14:18 |
| frickler | and it has been failing since then https://zuul.opendev.org/t/openstack/builds?job_name=osc-functional-devstack-tips&project=openstack/openstacksdk | 14:18 |
| frickler | gtema: stephenfin: ^^ I have no idea how to actually fix that, so maybe that would be the patch to revert for now | 14:19 |
| gtema | uhm, reverting sdk patch would require release to unblock SDK. Perhaps we should rather really mark functest as failing while we are searching for proper fix | 14:28 |
| frickler | gtema: do you really mean @unittest.expectedFailure or just skipping the test? I haven't seen the former used in openstack yet, though it might indeed be useful | 14:52 |
| gtema | expectedFailure perhaps | 14:53 |
| frickler | o.k., I'll propose a patch | 14:53 |
| gtema | great, thks | 14:54 |
| opendevreview | Dr. Jens Harbott proposed openstack/python-openstackclient master: Mark test_server_add_remove_volume as expected failure https://review.opendev.org/c/openstack/python-openstackclient/+/820895 | 14:58 |
| frickler | meh, that is lacking an import | 15:06 |
| opendevreview | Dr. Jens Harbott proposed openstack/python-openstackclient master: Mark test_server_add_remove_volume as expected failure https://review.opendev.org/c/openstack/python-openstackclient/+/820895 | 15:11 |
| *** artom__ is now known as artom | 16:22 | |
| opendevreview | Dr. Jens Harbott proposed openstack/python-openstackclient master: Fix RemoveServerVolume https://review.opendev.org/c/openstack/python-openstackclient/+/820917 | 17:36 |
| frickler | gtema: stephenfin: ^^ this would actually seem to be the fix for OSC matching the current SDK behavior. not sure if you'd want to use that and just amend the descriptions of the sdk functions accordingly, or whether you have an idea to make this volume attachment thing work both with and without 2.89 | 17:38 |
| gtema | ideally should work with and without. Will have a look tomorrow | 17:39 |
| gtema | thanks a lot | 17:39 |
| opendevreview | Rodolfo Alonso proposed openstack/python-openstackclient master: Add network update quota "limit_check" parameter https://review.opendev.org/c/openstack/python-openstackclient/+/806016 | 17:49 |
| opendevreview | Jan Hartkopf proposed openstack/openstacksdk master: fix creation of protected image https://review.opendev.org/c/openstack/openstacksdk/+/820926 | 17:56 |
| *** diablo_rojo__ is now known as diablo_rojo | 18:06 | |
| opendevreview | melanie witt proposed openstack/python-openstackclient stable/victoria: compute: Reorder building of columns for 'server list' https://review.opendev.org/c/openstack/python-openstackclient/+/820959 | 23:06 |
| opendevreview | melanie witt proposed openstack/python-openstackclient stable/victoria: compute: Show flavor in 'server list' with API >= 2.47 https://review.opendev.org/c/openstack/python-openstackclient/+/820960 | 23:06 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!