| opendevreview | Merged openstack/tempest master: Avoid wait for vol detach in live-migrate tests https://review.opendev.org/c/openstack/tempest/+/875500 | 00:20 |
|---|---|---|
| whoami-rajat | hi #openstack-qa , need some reviews on a simple patch, an important cinder patch depends on it https://review.opendev.org/c/openstack/tempest/+/875372 | 01:20 |
| gmann | whoami-rajat: sure, checking | 01:22 |
| whoami-rajat | gmann, thanks! | 01:23 |
| whoami-rajat | we are removing the functionality to create multiattach volumes as tempest does today, so these tests will break with it | 01:24 |
| gmann | whoami-rajat: but that is backward incompatible change right? and should be done with microversion bump | 01:25 |
| gmann | older microversion should continue to request create volume with multiattach=True in request body | 01:26 |
| gmann | otherwise it break users doing so as Tempest tests are breaking | 01:26 |
| whoami-rajat | gmann, we deprecated the functionality in queens and removed the client support in train, we kept the API behavior for backward compatibility but certainly forgot to remove it at a later stage | 01:27 |
| whoami-rajat | gmann, the original behavior is not part of microversion so i don't think we need to keep that support, also we don't want non-admin users to be able to create multiattach volumes | 01:28 |
| whoami-rajat | gmann, the way it should be done is via an admin user creating a multiattach volume type and non-admins using that type to create the multiattach volume | 01:28 |
| whoami-rajat | gmann, you can see we raise a deprecation warning since queens that this way of creating multiattach volumes is deprecated https://github.com/openstack/cinder/blob/master/cinder/api/v3/volumes.py#L394-L398 | 01:29 |
| whoami-rajat | and will be removed in a future release | 01:29 |
| gmann | whoami-rajat: for API, deprecation and removal process is not easy and it can break users when we remove the things (unless we do not have any other way). That is why microversion is for | 01:30 |
| gmann | whoami-rajat: this change is changing the API request format/body which is clear case of doing it via bumping the microversion | 01:31 |
| whoami-rajat | gmann, i understand that but we don't want to support the old behavior anymore so a microversion bump will still allow the old behavior with microversion < the one we use here | 01:32 |
| whoami-rajat | gmann, also the old behavior is from v2 API and there is no microversion tagged to it | 01:33 |
| gmann | whoami-rajat: one good example is nova proxy API which we deprecated long back but will not be able to remove without microversion that is why they are still working for older microversion https://docs.openstack.org/api-ref/compute/#deprecated-apis | 01:33 |
| gmann | whoami-rajat: I think with v3 also user can create multiattach volume with request parameter right? | 01:33 |
| gmann | current tempest test run on v3 API | 01:33 |
| whoami-rajat | yes correct, but there is no 3.X microversion that supports this functionality, 3.0 base is equivalent to 2.0 | 01:34 |
| gmann | yeah so all 3.X support it as it did not got removed in any microverison | 01:34 |
| gmann | whoami-rajat: that is what my point is it is API backward incompatible change and Tempest is to block such change. | 01:35 |
| whoami-rajat | gmann, ok, i can discuss with the cinder team again but we really don't want to keep the old behavior as users end up creating multiattach volumes without a cluster aware filesystem and end up corrupting there data so there is a risk associated with the old way | 01:36 |
| whoami-rajat | s/corrupting/corrupting data | 01:37 |
| gmann | whoami-rajat: ok, if this way is a bug currently (does not work at all) as you mentioned above (I am not expert on that) then it is ok to remove without microversion but I can see (like tempest test) there are some cases it is successful so those users will be broken. | 01:39 |
| gmann | whoami-rajat: +1 please check with cinder team, thanks | 01:39 |
| gmann | in nova, we have lot of such deprecated things which we do want to remove in favor of new things but we have to keep them for older microversion users and CANNOT remove them until we bump the min supported microversion | 01:40 |
| gmann | that is not best way to handle via microversion but we need to maintain backward compatibility with cost of extra maintenance of old code | 01:41 |
| whoami-rajat | gmann, yes, this was raised on ML as an issue with existing deployments https://lists.openstack.org/pipermail/openstack-discuss/2023-February/032389.html | 01:46 |
| whoami-rajat | gmann, thanks I will check with the team | 01:47 |
| whoami-rajat | gmann, i agree with that, it's some what of a special case where we don't want users to use the old way else we've kept consistency groups for several years even though they're deprecated in favor of volume groups | 01:48 |
| gmann | whoami-rajat: yeah, let me check email but I strongly feel if it was working fine then we should not remove it without microversion. and Tempest shows that it was working. | 01:50 |
| whoami-rajat | gmann, it does work but we're in process of removing it https://review.opendev.org/c/openstack/cinder/+/874865 | 01:51 |
| gmann | whoami-rajat: please check/discuss I will reply on email also to have broader pic but as this is release time and you need something urgent, I will be approving the tempest change on priority If i agree to change :) | 01:51 |
| whoami-rajat | gmann, sure, will discuss in today's cinder meeting and update, thank you! | 01:52 |
| gmann | whoami-rajat: thanks, i can comment on cinder change also if you want or email reply is ok ? | 01:54 |
| whoami-rajat | gmann, the email reply should be good for reference | 01:54 |
| gmann | ack | 01:54 |
| gmann | whoami-rajat: replied https://lists.openstack.org/pipermail/openstack-discuss/2023-March/032502.html | 02:03 |
| whoami-rajat | gmann, thanks! | 02:14 |
| *** ralonsoh_ is now known as ralonsoh | 07:36 | |
| *** jpena|off is now known as jpena | 08:45 | |
| *** ralonsoh__ is now known as ralonsoh | 10:18 | |
| opendevreview | Lukas Piwowarski proposed openstack/tempest master: Fix server cleanup in test_multiple_create test https://review.opendev.org/c/openstack/tempest/+/875751 | 10:26 |
| opendevreview | Takemi Asakura proposed openstack/tempest master: Cinder client fails if identity API version specified https://review.opendev.org/c/openstack/tempest/+/874094 | 10:46 |
| opendevreview | Ade Lee proposed openstack/devstack master: Modify devstack-base to allow for fips https://review.opendev.org/c/openstack/devstack/+/871606 | 11:13 |
| opendevreview | Ade Lee proposed openstack/devstack master: Modify devstack-base to allow for fips https://review.opendev.org/c/openstack/devstack/+/871606 | 12:57 |
| opendevreview | Dan Smith proposed openstack/grenade master: Generate SSH key locally for cinder resources https://review.opendev.org/c/openstack/grenade/+/875940 | 15:31 |
| dansmith | gmann: kopecmartin ^ apparently we don't run grenade on jammy at all yet? | 15:40 |
| opendevreview | Dan Smith proposed openstack/grenade master: Generate SSH key locally for cinder resources https://review.opendev.org/c/openstack/grenade/+/875940 | 15:45 |
| gmann | dansmith: yes, in week or two once all devstack projects are branched then we will branch grenade also and then update the job setting so that on master (2023.2) job will run 2023.1->2023.2 on jammy | 16:21 |
| gmann | currently it is zed->2023.1 which is on focal as zed is on focal | 16:21 |
| *** jpena is now known as jpena|off | 17:35 | |
| opendevreview | Dan Smith proposed openstack/grenade master: Generate SSH key locally for cinder resources https://review.opendev.org/c/openstack/grenade/+/875940 | 19:17 |
| gmann | kopecmartin: https://review.opendev.org/c/openstack/tempest/+/874094 | 19:18 |
| kopecmartin | gmann: +w, thanks for bringing that up | 19:31 |
| gmann | thanks | 19:35 |
| kopecmartin | gmann: this is interesting https://review.opendev.org/c/openstack/tempest/+/875751 | 20:20 |
| gmann | kopecmartin: agree with you. replied | 20:29 |
| kopecmartin | thanks | 20:30 |
| opendevreview | Dan Smith proposed openstack/grenade master: Factor out a grenade-skip-level-always job https://review.opendev.org/c/openstack/grenade/+/875990 | 22:35 |
| opendevreview | Dan Smith proposed openstack/grenade master: Factor out a grenade-skip-level-always job https://review.opendev.org/c/openstack/grenade/+/875990 | 22:38 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!