| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2024-07-30 | |||
| 13:21:30 | jlejeune | yep :) | |
| 13:21:52 | sean-k-mooney | its not a blocker just an anoyance currently | |
| 13:22:18 | sean-k-mooney | i was hopign to do a new release today/tomorrow | |
| 13:22:33 | sean-k-mooney | we could include it if the patch is ready quickly | |
| 13:22:41 | sean-k-mooney | but i dont really want to hold the release just for this | |
| 13:23:36 | jlejeune | I understand | |
| 13:24:16 | mnaser | question.. has anyone ran into this in nova CI? | |
| 13:24:18 | mnaser | `TypeError: canonicalize_version() got an unexpected keyword argument 'strip_trailing_zero'` | |
| 13:24:24 | mnaser | https://zuul.opendev.org/t/openstack/build/7e95381e525e43e6afb34c1a0381e971 | |
| 13:24:34 | mnaser | I know its unmaintained zed but in case this rings a bell and needs us to back port something in specific : | |
| 13:37:51 | dansmith | it's a new change in packageutils or something I think | |
| 13:38:06 | dansmith | I hit it elsewhere outside of openstack | |
| 13:38:36 | dansmith | part of a larger theme of issues everyone has been facing, because the python community decided they should tell us what our version numbers have to look like | |
| 13:38:51 | sean-k-mooney | that has been fixed kind fo | |
| 13:39:20 | sean-k-mooney | its caused by a new release of setuptools conflciting with the max version of packaging we allowed in 2023.1 | |
| 13:39:40 | sean-k-mooney | we resolved that on stable by bumping the max version to the same as 2023.2 | |
| 13:39:55 | sean-k-mooney | alternitivly you can clamp the max veriosn of setup tools on older release | |
| 13:42:57 | opendevreview | Rajesh Tailor proposed openstack/nova master: Rephrase the help message https://review.opendev.org/c/openstack/nova/+/925207 | |
| 13:43:04 | mnaser | hmmm | |
| 13:43:22 | mnaser | so I'd have to fix this in requirements | |
| 13:43:38 | mnaser | or cap it inside nova? | |
| 13:47:07 | sean-k-mooney | mnaser: yes you would | |
| 13:47:18 | sean-k-mooney | you cant cap settup tool in requiremnts | |
| 13:47:29 | sean-k-mooney | you could only bump the pacakging version | |
| 13:47:42 | sean-k-mooney | if you didnt want to do that you would have to clamp it in the ci job | |
| 13:50:59 | opendevreview | Julien Le Jeune proposed openstack/nova master: Fix test_snapshot_running test when qemu-img binary is missing https://review.opendev.org/c/openstack/nova/+/925208 | |
| 13:56:51 | opendevreview | Julien Le Jeune proposed openstack/nova master: Fix test_snapshot_running test when qemu-img binary is missing https://review.opendev.org/c/openstack/nova/+/925208 | |
| 13:59:32 | opendevreview | Rajesh Tailor proposed openstack/nova master: Fix typos and punctuation issues https://review.opendev.org/c/openstack/nova/+/878779 | |
| 13:59:55 | elodilles | mnaser: i'm not sure that patch is needed. i mean, the gate is fine without that patch. but if you could review these CVE related zed backports, that would be awesome o:) (6 patch :S) https://review.opendev.org/c/openstack/nova/+/925018/2 | |
| 14:05:22 | jlejeune | elodilles: hello, following your comment, I backported that commit in 2024.1: https://review.opendev.org/c/openstack/nova/+/925181 | |
| 14:16:28 | elodilles | jlejeune: thanks! looks good to me! +2'd | |
| 14:25:03 | sean-k-mooney | jlejeune: ya we have that dowsn tream too already i agree iwth backporting that | |
| 14:25:26 | jlejeune | elodilles: sean-k-mooney: thanks, I have another small fix on unit test like previous one I made last week: https://review.opendev.org/c/openstack/nova/+/925208 | |
| 14:25:36 | sean-k-mooney | i tought there were already backport of that proposed but i guess not | |
| 14:25:48 | sean-k-mooney | jlejeune: yep i reviewd thats incorrect | |
| 14:26:15 | sean-k-mooney | previosly it was fine to skip if qemu-img was not installed as the tests in question were validtating the behaivor of calling qemu image | |
| 14:26:24 | sean-k-mooney | this test is just missing a mock | |
| 14:26:28 | jlejeune | ok doing | |
| 14:43:48 | mnaser | elodilles: the thing is nova-multistore job is broken so not letting glance job pass so we can't finish landing the CVE fixes in glance :( | |
| 14:45:48 | elodilles | mnaser: ahh, i see | |
| 14:46:13 | elodilles | mnaser: meanwhile i found the issue you mentioned: https://bugs.launchpad.net/neutron/+bug/2073585 | |
| 14:46:47 | mnaser | elodilles: reviewed the zed stuff, lgtm | |
| 14:47:30 | elodilles | mnaser: and it seems there were some discussion in this patch, that might lead to some kind of a fix: https://review.opendev.org/c/openstack/requirements/+/924494 | |
| 14:47:47 | elodilles | mnaser: though i don't see yet what is the right fix o:) | |
| 14:48:00 | mnaser | elodilles: I think here - https://review.opendev.org/c/openstack/requirements/+/925271 | |
| 14:49:09 | opendevreview | Julien Le Jeune proposed openstack/nova master: Fix test_snapshot_running https://review.opendev.org/c/openstack/nova/+/925208 | |
| 14:51:55 | elodilles | mnaser: hmmm, that doesn't seem right to me. i mean, upper-constraints shouldn't be updated on non-master branches (except openstack libraries) in general. is the nova-multistore job's tempest coming from master branch, or zed-last tag? | |
| 14:53:18 | elodilles | mnaser: and thanks for reviewing the zed patches o/ I'll fast-approve them then now | |
| 14:56:07 | clarkb | elodilles: I had to push back on that when this issue first came up | |
| 14:56:19 | clarkb | the issue is a requirements upper constraints bug as far as our ability to fix things goes | |
| 14:56:42 | clarkb | we need to have flexibility in policy to understand that these cases do occasionally occur and allow for fixing them when they do | |
| 14:58:06 | clarkb | we can say "policy is to never change this thing" but then you have to accept it will break at some point due to bitrot. The bitrot has occurred. The decision is "do we want to make this work again or not" and less so "what is our policy" imo | |
| 15:00:42 | elodilles | clarkb: yeah, i mean, we had issues in the past, too, when we needed to bump upper version for some external dependencies. but at a first glance this seems to me that (for zed) we should not use latest master from tempest. | |
| 15:01:32 | clarkb | its just a bump of the upper constraint value of a package to a working version | |
| 15:02:12 | clarkb | and things are already broken so the risk of breaking things harder is low | |
| 15:04:02 | elodilles | i'm not against it, fwiw, just saying that maybe we could use zed-last (given that the issue comes from tempest and the fact that we use it from master branch), in that way we can avoid some other bitrot issues in the future o:) | |
| 15:04:40 | clarkb | I don't think changing the branch will fix the problem because the issue is in setuptools | |
| 15:05:26 | elodilles | clarkb: i see, then it's an other issue. of course in that case probably this is the best workaround | |
| 15:12:27 | jlejeune | sean-k-mooney: something like that : https://review.opendev.org/c/openstack/nova/+/925208 ? | |
| 15:17:32 | elodilles | melwitt dansmith : sorry for pinging you, but when you have some time, could you please review these two fix for the cherry-pick-validator script? o:) https://review.opendev.org/c/openstack/nova/+/924104 and https://review.opendev.org/c/openstack/nova/+/924103 | |
| 15:17:49 | sean-k-mooney | jlejeune: that might be too low level | |
| 15:36:35 | bauzas | reminder: nova meeting in 24 mins here | |
| 15:39:05 | jlejeune | sean-k-mooney: you're tlaking about the mock over processutils.execute() function ? | |
| 15:40:00 | sean-k-mooney | so that test shoudl not be doign image inspection really | |
| 15:40:47 | jlejeune | ok so I need to mock create_image() and extract_snapshot() methods ? | |
| 15:42:48 | sean-k-mooney | this is ment to test https://github.com/openstack/nova/blob/7a7427691e0bd4818bb7a2c5f5371e0244addbbb/nova/virt/driver.py#L757C9-L757C17 | |
| 15:42:55 | sean-k-mooney | using the fake driver | |
| 15:43:18 | sean-k-mooney | so https://github.com/openstack/nova/blob/master/nova/virt/fake.py#L213-L216 | |
| 15:43:59 | sean-k-mooney | however it seam like we are using the libvirt fixutre in theses tests instead | |
| 15:45:44 | sean-k-mooney | in the libvirt driver its calling https://github.com/openstack/nova/blob/master/nova/virt/libvirt/driver.py#L3122 | |
| 15:46:30 | jlejeune | yes | |
| 15:46:38 | sean-k-mooney | we are now intereally callign the image inspection code in that datapaht | |
| 15:46:42 | sean-k-mooney | we sould not be | |
| 15:46:47 | sean-k-mooney | in this thest | |
| 15:46:51 | sean-k-mooney | *test | |
| 15:47:01 | sean-k-mooney | so that is pointing to a gap in the fixutre | |
| 15:47:23 | sean-k-mooney | do you have a traceback of how this failed? | |
| 15:47:46 | jlejeune | here: https://bugs.launchpad.net/nova/+bug/2075178 | |
| 15:50:55 | sean-k-mooney | ok so libvirt_utils.create_image shoudl be stubbed here and should not be callign qemu-img | |
| 15:53:08 | sean-k-mooney | jlejeune: so we need to fix this in https://github.com/openstack/nova/blob/master/nova/tests/unit/virt/test_virt_drivers.py#L817 | |
| 15:53:38 | sean-k-mooney | that subcalses shoudl be stubing the call to create image | |
| 15:53:41 | jlejeune | by adding a new stub | |
| 15:54:08 | jlejeune | ok because I've mocked that function in my ongoing patchset, but I will update it to stub the call instead | |
| 15:54:22 | sean-k-mooney | you are doing it in the wrong place | |
| 15:54:38 | sean-k-mooney | your doing it in the share test function that is used acrros several virt drivers | |
| 15:54:55 | sean-k-mooney | you need to do it in the libvirt specific class that inherits form it | |
| 15:55:16 | jlejeune | ok, ok, I didn't yet pushed it actually | |
| 15:55:35 | jlejeune | I will do it tomorrow, I must leave right now | |
| 15:55:37 | sean-k-mooney | in the bug we see its nova.tests.unit.virt.test_virt_drivers.LibvirtConnTestCase.test_snapshot_running | |
| 15:55:54 | jlejeune | ok I see | |
| 15:55:54 | sean-k-mooney | that is failing so we need to stub it form LibvirtConnTestCase class | |
| 15:56:03 | jlejeune | ok I get it | |
| 15:56:05 | jlejeune | thanks | |
| 15:56:37 | jlejeune | will do it tomoroww, see you | |
| 15:58:48 | sean-k-mooney | dansmith: you might have some insight into what exactly might have changed to cause https://bugs.launchpad.net/nova/+bug/2075178 | |
| 15:59:11 | sean-k-mooney | you mentioned seeing some NotImplementedError in the tests | |
| 15:59:46 | dansmith | I think that's because we call qemu-img in those paths but didn't before right? | |
| 15:59:46 | sean-k-mooney | i had taught we have a poison fixtreure in our tests that prevented all calls to qemu like the libivrt one but perhaps imagined that. | |
| 15:59:54 | sean-k-mooney | yep | |
| 16:00:03 | sean-k-mooney | well yes and no | |