| *** dmellado9 is now known as dmellado | 05:04 | |
| opendevreview | Soniya Murlidhar Vyas proposed openstack/tempest master: Enable new servies in tempest.conf https://review.opendev.org/c/openstack/tempest/+/882915 | 06:15 |
|---|---|---|
| opendevreview | Soniya Murlidhar Vyas proposed openstack/tempest master: Enable new servies in tempest.conf https://review.opendev.org/c/openstack/tempest/+/882915 | 06:25 |
| opendevreview | Soniya Murlidhar Vyas proposed openstack/tempest master: Enable new servies in tempest.conf https://review.opendev.org/c/openstack/tempest/+/882915 | 09:16 |
| kopecmartin | gmann: hi, any chance we could make tempest create VMs on specific node? | 12:55 |
| opendevreview | ribaudr proposed openstack/tempest master: [WIP] Test tempest for Nova and Manila https://review.opendev.org/c/openstack/tempest/+/838743 | 13:58 |
| *** spotz is now known as Guest1133 | 14:06 | |
| *** spotz_ is now known as spotz | 14:06 | |
| opendevreview | zoumingzhe proposed openstack/hacking master: handle from-style in hacking_import_alphabetical https://review.opendev.org/c/openstack/hacking/+/882990 | 16:18 |
| opendevreview | zoumingzhe proposed openstack/hacking master: handle from-style in hacking_import_alphabetical https://review.opendev.org/c/openstack/hacking/+/882990 | 16:49 |
| gmann | kopecmartin: you mean on specific compute node? | 16:54 |
| gmann | kopecmartin: hen yes, you can do via scheduler hint and if you want guaranteed booting then use force-host in create server requsest | 16:54 |
| opendevreview | zoumingzhe proposed openstack/hacking master: handle from-style fro H306 https://review.opendev.org/c/openstack/hacking/+/882990 | 17:00 |
| opendevreview | zoumingzhe proposed openstack/hacking master: handle from-style for H306 https://review.opendev.org/c/openstack/hacking/+/882990 | 17:02 |
| opendevreview | zoumingzhe proposed openstack/hacking master: handle from-style for H306 https://review.opendev.org/c/openstack/hacking/+/882990 | 17:12 |
| dansmith | gmann: kopecmartin: would be good to get this in to validate the CVE if you have time: https://review.opendev.org/c/openstack/tempest/+/882876 | 17:27 |
| gmann | dansmith: sure | 17:28 |
| dansmith | gmann: note that this will fail on clouds that don't have the fix | 17:29 |
| dansmith | I dunno how you reconcile that with like stable interfaces and stuff, but it seems more valuable to have the indication than not | 17:29 |
| gmann | dansmith: as long as it is green on all supported branches as fixes are backported and are proposed/merging, tempest is good. we have older version of tempest for EM or old branches | 17:30 |
| dansmith | cool | 17:31 |
| gmann | yeah, as know this case and we have to introduce some back incompatible things which is nothing but better for users than think of breaking them | 17:31 |
| dansmith | yep | 17:31 |
| gmann | I have a quick internal call and after that will review tempest change | 17:32 |
| dansmith | gmann: no problem thanks | 17:32 |
| opendevreview | Ashley Rodriguez proposed openstack/devstack-plugin-ceph master: [WIP][DNM] Remote Ceph with cephadm https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/876747 | 17:36 |
| dansmith | gmann: (or maybe sean-k-mooney) do you recognize the first failure here? https://17bb12c4d334815c4390-11a2e186f8f6505f790122d590a395c2.ssl.cf2.rackcdn.com/882955/2/check/cinder-plugin-ceph-tempest-mn-aa/f9c0ed1/testr_results.html | 18:17 |
| dansmith | kinda looks like the job is configured to try something that assumes non-shared storage? | 18:18 |
| sean-k-mooney | hum | 18:22 |
| sean-k-mooney | well that message implies the test pass block-migration=true with the old microverion right | 18:22 |
| sean-k-mooney | since that parematner is now auto and noone should normally pass it anymore | 18:22 |
| sean-k-mooney | i wonder why that is happening | 18:22 |
| dansmith | right that's why I'm wondering if that should ever work in this job configuration | 18:23 |
| sean-k-mooney | i think there is a tempest config option for this | 18:23 |
| dansmith | it's marked n-v so it's possible it just never worked | 18:23 |
| sean-k-mooney | do you have a link to the zuul job | 18:23 |
| sean-k-mooney | i just eant to look at the jobs cofnig | 18:24 |
| dansmith | https://review.opendev.org/882955 | 18:24 |
| sean-k-mooney | thanks | 18:24 |
| dansmith | it's about to report | 18:24 |
| sean-k-mooney | so there is this https://github.com/openstack/tempest/blob/master/tempest/config.py#L493C36-L496 | 18:25 |
| sean-k-mooney | 'block_migration_for_live_migration' | 18:25 |
| sean-k-mooney | that should be false | 18:25 |
| dansmith | I wonder if this is related: https://github.com/openstack/cinder/blob/master/playbooks/cinder-multibackend-matrix.yaml#L14 | 18:27 |
| dansmith | block_migration_for_live_migration = True | 18:28 |
| dansmith | it's getting set | 18:28 |
| sean-k-mooney | i dont think so | 18:28 |
| sean-k-mooney | ok form where i was tracking up the job parents | 18:28 |
| sean-k-mooney | i dont see it set as a devstack paramter | 18:28 |
| dansmith | so that should be true when, when we're not on ceph? | 18:29 |
| dansmith | (or nfs or other shared storage for root) | 18:30 |
| sean-k-mooney | mother rang brb | 18:31 |
| gmann | but this is set to True in this job https://zuul.opendev.org/t/openstack/build/f9c0ed1331854ae1a57c8be97bd18215/log/controller/logs/tempest_conf.txt#81 | 18:35 |
| dansmith | gmann: right | 18:36 |
| dansmith | and I'm wondering if it should not be | 18:36 |
| dansmith | but I also don't know where it's getting set from | 18:36 |
| gmann | dansmith: yeah checking how it is set but if you see nova live migration ceph job it is explicitly set to false https://opendev.org/openstack/nova/src/branch/master/.zuul.yaml#L218 | 18:37 |
| dansmith | cool | 18:38 |
| dansmith | so I guess I can just set it false at least and see how that goes | 18:38 |
| gmann | dansmith: from here it is set https://github.com/openstack/tempest/blob/master/zuul.d/base.yaml#L81 | 18:38 |
| dansmith | ah okay | 18:39 |
| gmann | we need to make it false explicitly in ceph base multinode job | 18:39 |
| gmann | here https://opendev.org/openstack/devstack-plugin-ceph/src/branch/master/.zuul.yaml#L122 | 18:39 |
| dansmith | okay | 18:39 |
| gmann | dansmith: but was it working in this? https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/882483 | 18:40 |
| dansmith | gmann: they don't even run that job in their config | 18:40 |
| dansmith | gmann: and the only one that I knew about was nova's live migration ceph job | 18:40 |
| gmann | ohk | 18:41 |
| gmann | ah right, just saw | 18:42 |
| dansmith | I'll just add it to that patch and depends-on that | 18:42 |
| opendevreview | Dan Smith proposed openstack/devstack-plugin-ceph master: Make multinode ceph job use cephadm https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/882483 | 18:43 |
| dansmith | oh wait, I can't do that because it doesn't work for multinode yet | 18:44 |
| gmann | may be separate change | 18:46 |
| dansmith | yup | 18:46 |
| dansmith | also need to enable validations on that base job anyway | 18:46 |
| opendevreview | Dan Smith proposed openstack/devstack-plugin-ceph master: WIP Make multinode ceph job use cephadm https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/882483 | 18:47 |
| opendevreview | Dan Smith proposed openstack/devstack-plugin-ceph master: Enable validation and disable block-migration https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/882987 | 18:47 |
| dansmith | gmann: not sure if you saw my note in nova this morning, but this cinder job is the only one I still see throwing volume detach failures (that aren't caused by other things) | 18:48 |
| dansmith | not critical to fix, but it seems to be failing a *lot* and if it has config that means it never passes.. probably not very useful :) | 18:48 |
| dansmith | failing with volume detach a lot I mean | 18:48 |
| sean-k-mooney | ok back | 18:48 |
| sean-k-mooney | block_migration_for_live_migration = True | 18:49 |
| sean-k-mooney | should only really be set if you are trying to test old migrovition i think | 18:49 |
| dansmith | okay it's on by default in the devstack-multinode job | 18:50 |
| sean-k-mooney | and where the job does not have ceph or nfs | 18:50 |
| sean-k-mooney | well the default behavior since like 6+ years i think has been auto | 18:50 |
| gmann | dansmith: ohk | 18:50 |
| sean-k-mooney | https://docs.openstack.org/nova/latest/reference/api-microversion-history.html#id12 | 18:50 |
| sean-k-mooney | 2.14 | 18:51 |
| sean-k-mooney | so form mitaka nova will fiture it out and do the right thing | 18:51 |
| sean-k-mooney | so tempest proably should just remove that config option at this point | 18:51 |
| sean-k-mooney | actully 2.25 was for live migration | 18:52 |
| sean-k-mooney | still mitaka https://docs.openstack.org/nova/latest/reference/api-microversion-history.html#maximum-in-mitaka | 18:52 |
| sean-k-mooney | 2.14 was for evacuation | 18:52 |
| gmann | yeah but it is still supported for older mivroversion so keeping in tempest is ok | 18:52 |
| sean-k-mooney | it should not be on by default | 18:53 |
| sean-k-mooney | only in a job tha tis not usign ceph and does not have nova /var/lib/nova/instance on nfs | 18:53 |
| sean-k-mooney | gmann: im wondiring is thre a way to not pass any value in tempets or pass auto | 18:55 |
| dansmith | sean-k-mooney: that's the job it's default on in, | 18:55 |
| gmann | yeah | 18:55 |
| dansmith | it's just we need to flip it off on the job that implements ceph | 18:55 |
| dansmith | which I'm doing | 18:55 |
| gmann | yes, like nova did for ceph job | 18:55 |
| dansmith | right | 18:55 |
| dansmith | nova's job should also get it from the flag I'm setting in devstack-ceph-plugin eventually | 18:56 |
| sean-k-mooney | sure im just concend that tempest might not be testign the default behavior that we have for new microveion which is what we expect people to actully use | 18:56 |
| gmann | yeah that will be good for all ceph job hoping they inherit from ceph plugin jobs | 18:56 |
| sean-k-mooney | like https://github.com/openstack/tempest/blob/e5da6756b9e7fb8fbedade5d3a428e0a8bff94ff/tempest/api/compute/admin/test_live_migration_negative.py#L36-L37 | 18:56 |
| sean-k-mooney | is always setting true or false | 18:56 |
| gmann | sean-k-mooney: we do https://github.com/openstack/tempest/blob/master/tempest/api/compute/admin/test_live_migration.py#L377 | 18:57 |
| sean-k-mooney | ok cool | 18:57 |
| gmann | microversion specific test are with microversion caps so that it does not fail for older version. | 18:58 |
| gmann | and we keep testing older behavior also with existing tests | 18:58 |
| sean-k-mooney | im not going to hold my breat to get my wish to bumnp the min microversion nova supports anytime soon | 18:59 |
| gmann | :) | 18:59 |
| dansmith | good :) | 18:59 |
| sean-k-mooney | but i would deally like to bump it to 2.60 | 19:00 |
| sean-k-mooney | https://docs.openstack.org/nova/latest/reference/api-microversion-history.html#maximum-in-queens | 19:00 |
| gmann | dansmith: reviewed CVE fix tests. can we move those to scenario test as they are good tests to validate the cross services things ? | 19:41 |
| dansmith | gmann: reading now | 19:41 |
| gmann | some doc string comment, and what you think of this test https://review.opendev.org/c/openstack/tempest/+/882876/2/tempest/api/volume/test_volumes_negative.py#553 | 19:41 |
| gmann | k | 19:41 |
| dansmith | gmann: I don't actually know what the distinction is supposed to be.. scenario are just "full stack" sorts of things? | 19:41 |
| dansmith | in this case, these tests are *only* testing the cinder change, not the nova one and not the brick/glance ones either | 19:42 |
| gmann | dansmith: our negative tests are mainly API interface validation tests and less of cross service operation validation | 19:42 |
| dansmith | so it seems pretty specific to cinder | 19:42 |
| dansmith | this is pretty much just cinder api behavior, so seems probably correct for api negative to me | 19:43 |
| gmann | but it involve the nova service if attachment is there or not righ | 19:43 |
| dansmith | it requires attaching to an instance yes, but it could be nova from zed and it'd still work | 19:43 |
| dansmith | but if that's enough to move to scenario, that's fine | 19:43 |
| gmann | I feel cinder talk to nova for these operation and test the cross service things. so putting these in scneario can be good place | 19:44 |
| dansmith | okay, sure | 19:44 |
| gmann | our negative tests are more like just a very upper level interface/DB validation which is almost done by service side unit tests but we kept them due to interop using them | 19:45 |
| dansmith | well, this is kinda that, but since it checks with nova to determine if it should enforce it, it's a little blurred I agree | 19:45 |
| gmann | I do not want these good test falling in those catagory and someday we just delete all negative tests make these tests also go | 19:45 |
| dansmith | ack | 19:45 |
| dansmith | gmann: okay can I make the changes you describe and do the reno and work on the additional (good) test case in a follow-on patch? | 19:46 |
| gmann | reno and additional test is all good to do in followup but you mean moving them to scneario test too? | 19:47 |
| dansmith | no, I'll do the move to scenario, the reno, and the fixes now | 19:48 |
| dansmith | just save the new case for follow-on | 19:48 |
| gmann | sure, make sense | 19:48 |
| dansmith | so new file in tempest/scenario or is there one in there that could take these? | 19:49 |
| dansmith | volume_boot_scenario is not quite right | 19:49 |
| gmann | yeah, may be new file, I also could not find relevant file | 19:50 |
| dansmith | okay | 19:51 |
| dansmith | gmann: I assume you want these separate test cases folded into a more scenario-like single test that boots one server and tries a bunch of things against it right? | 19:58 |
| opendevreview | Ashley Rodriguez proposed openstack/devstack-plugin-ceph master: [WIP][DNM] Remote Ceph with cephadm https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/876747 | 20:13 |
| gmann | dansmith: yeah that will be much better | 20:19 |
| dansmith | gmann: ack, I gotta restack so it'll probably be monday before I get going on this much | 20:19 |
| gmann | ack | 20:19 |
| dansmith | gmann: I also noticed none of these wait for sshable so they needed work there as well | 20:19 |
| dansmith | sort of showed up in parallel so I wasn't thinking about that | 20:19 |
| dansmith | (and I didn't write them) | 20:20 |
| gmann | yeah | 20:20 |
| dansmith | that fixed the cinder ceph multinode test, except for the second failure which is also what seems like a job misconfiguration related to volume retype | 20:33 |
| dansmith | but, progress at least | 20:33 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!