| opendevreview | Martin Kopec proposed openstack/whitebox-tempest-plugin master: DNM experimenting with enforce scope https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/878793 | 13:00 |
|---|---|---|
| dansmith | gmann: when you're around...c-t-p depends on more in tempest master than just the SSHABLE support and I'm not sure how to resolve it and make sure it gets tested exactly | 16:42 |
| dansmith | with more changes to tempest's scenario manager, we can pass validation_resources from c-t-p through, but tempest master doesn't actually pass validatable= through even though it's an argument to create_server() | 16:42 |
| dansmith | so I guess I'm just confused about what needs to run with what versions and what dance we need to do to get these things playing together | 16:43 |
| dansmith | I'm half thinking we should just not run the c-t-p plugin in the nova job so we can proceed with the ceph job stuff and let someone else sort this out | 16:43 |
| dansmith | we're incredibly deep in the dependency cycle here | 16:43 |
| dansmith | gouthamr: ^ | 16:43 |
| opendevreview | Dan Smith proposed openstack/tempest master: WIP: Also require resources for PINGABLE https://review.opendev.org/c/openstack/tempest/+/882102 | 16:54 |
| opendevreview | Dan Smith proposed openstack/tempest master: Allow passing validation_resources in scenario https://review.opendev.org/c/openstack/tempest/+/882342 | 16:54 |
| dansmith | I think this ^ might less us write c-t-p in a way that it will work on both old and new tempest | 17:03 |
| gouthamr | dansmith: thanks; that should work. sorry this is frustrating and not related to the problem you're solving | 17:08 |
| gouthamr | maybe kopecmartin won't mind a tempest release too; but ctp maintainers will figure out the requirements bump once that release comes | 17:09 |
| dansmith | I need to make a couple more changes to c-t-p... does it normally run against the released tempest or does it always run against tempest master? | 17:09 |
| gouthamr | dansmith: runs against master, unless it is pinned (which it is stable/wallaby and earlier) | 17:19 |
| gouthamr | this requirement is just an indication to distros that need to pull specific versions.. | 17:19 |
| dansmith | gouthamr: okay but there's no job that runs against wallaby, right? so how do you know that it works against what is in requirements? | 17:20 |
| dansmith | ah, I see.. :) | 17:20 |
| opendevreview | Julia Kreger proposed openstack/grenade master: DNM: Try and get more details on grenade failure https://review.opendev.org/c/openstack/grenade/+/882346 | 17:39 |
| artom | Is https://zuul.opendev.org/t/openstack/build/e8e103293f134a7ba59a64335dcdd897/log/controller/logs/grenade.sh_log.txt#3156-3159 a known thing currently with the gate? | 18:42 |
| artom | Seems deterministic... I suppose that this being the grenade job for stable/2023.1 maybe I'm actually the first one to pay attention? | 18:43 |
| artom | *facepalm* ignore me, it's non-voting | 18:44 |
| opendevreview | Merged openstack/devstack master: git: support git checkout for a commit hash https://review.opendev.org/c/openstack/devstack/+/880892 | 18:46 |
| dansmith | artom: should be voting on nova, but it's not required for everyone | 18:57 |
| dansmith | I would expect that to break nova (unless we didn't mark it as voting) | 18:58 |
| gmann | dansmith: checking | 18:59 |
| artom | dansmith, it's not voting on the stable branch | 19:10 |
| artom | dansmith, https://zuul.opendev.org/t/openstack/build/e8e103293f134a7ba59a64335dcdd897 specifically | 19:11 |
| dansmith | oh skip level on 2023.1 meaning from yoga? | 19:12 |
| dansmith | that makes sense I was thinking master | 19:12 |
| dansmith | I'm surprised that broke though because it was working | 19:13 |
| dansmith | and broke on the source release even | 19:14 |
| gmann | dansmith: tosky: gouthamr: I replied on tempest new tag requirement, https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/881764/4..7/cinder_tempest_plugin/scenario/test_volume_multiattach.py#b54 | 19:15 |
| gmann | it will work as it is also but for safer side yes we can release tempest tag and bump requirement in cinder-tempest-plugin | 19:15 |
| dansmith | gmann: okay I guess I don't really understand what it supposed to work with what then | 19:16 |
| dansmith | gmann: however, if you see my most recent change to that patch, I think I have it working without the changes in master tempest (but there's no test for using tempest 30 it seems) | 19:16 |
| gmann | well our initial and main idea of tempest and plugins was master version of it work and thats it but with EM branch concpet we are trying multiple compatible version of plugin with tempest | 19:17 |
| dansmith | but it requires another change to tempest (which we probably want anyway) to allow passing your own validation_resources on the scenario tests | 19:17 |
| dansmith | gmann: okay I had assumed it was all "master works with master" on this stuff, so I was surprised by the requirements.txt having tempest in it | 19:17 |
| gmann | dansmith: it is all for making EM branch testing compatible versions otherwise for our supported branch + master testing we do not need to worry about these compatible versions thing | 19:17 |
| dansmith | gmann: okay but yesterday you said we *should* bump tempest to 30 there, so I figured that the c-t-p stuff had to work with tempest 30 :) | 19:18 |
| gmann | dansmith: agree, that is not needed actually but cinder-tempest-plugin want to run its master with older tempest compatible version too so they had it | 19:18 |
| dansmith | hmm | 19:19 |
| gmann | I think it is easy to release tempest and bump cinder-temepst-plugin to that version. which is ONLY needed for EM and very old branch testing otherwise that is NOT AT ALL used for supported branch + master testing | 19:19 |
| dansmith | gmann: I just replied to your keypair thing, tell me if I'm wrong | 19:20 |
| dansmith | anyway, we can do whatever you want. I'm happy to roll these all back to what was working as of this morning, or continue with what I have here | 19:20 |
| gmann | EM branch testing making brannchless concept more confusing not just version compatible but when I do pin tempest there I need to take care a lot | 19:20 |
| dansmith | I just really want to get the ceph job bumped so I can go back to my regular job :) | 19:20 |
| gmann | dansmith: all good the changes you did and it work well on all supported branch and master. we will just release new tempest, bump that in cinder-tempest-plugin and that's it | 19:21 |
| gmann | dansmith: checking keypair reply | 19:22 |
| dansmith | keypair is the new arg I added which is not supported in <master tempest, and was the point of tosky's comment | 19:22 |
| dansmith | without key_name the following sshing won't work on older tempest | 19:22 |
| gmann | dansmith: replied. with key_name from newly created keypair and no keypair in validation_resource make SSHABLE fail | 19:27 |
| dansmith | gmann: the only reason we will have generated a keypair is if run_validation is off | 19:28 |
| dansmith | if it's on, it's the same keypair | 19:28 |
| dansmith | I tested with on and off locally | 19:28 |
| gmann | key_name of keypairA and having keypairB in validation_resource still won't work for SSHABLE right? that is why you overridded keypair in validation_resopurces in manager.py | 19:28 |
| dansmith | I can test again after my next meeting | 19:29 |
| dansmith | gmann: they're the same keypair because of setUp() | 19:29 |
| gmann | dansmith: but this else part? https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/881764/6/cinder_tempest_plugin/scenario/test_snapshots.py#33 | 19:29 |
| dansmith | and I modified manager.py to take valdiation_resources | 19:29 |
| dansmith | gmann: that only happens if run_validation is off | 19:29 |
| dansmith | in which case we will not actually wait for SSHABLE | 19:29 |
| dansmith | but we still pass the assert because validation_resources={} not None | 19:30 |
| gmann | ohk so that is only just injected in server and we actually does not use it | 19:30 |
| dansmith | ? | 19:30 |
| dansmith | it's used by the test further down, without validation on | 19:30 |
| dansmith | here https://review.opendev.org/c/openstack/cinder-tempest-plugin/+/881764/6/cinder_tempest_plugin/scenario/test_snapshots.py#70 | 19:30 |
| dansmith | it needs a keypair regardless of if run_validation is on or off, but if it's on it uses the same one as validation_resources | 19:31 |
| dansmith | if off, it generates, like it did before I added SSHABLE | 19:31 |
| gmann | dansmith: ohhk, I did not that test validation and thinking just from tempest create_server() validation part. | 19:32 |
| gmann | got it now | 19:32 |
| gmann | all good there | 19:32 |
| gmann | dansmith: last one now. validatable, we need to change this also to take arg validatable onw https://review.opendev.org/c/openstack/tempest/+/882342/1/tempest/scenario/manager.py#338 | 19:33 |
| gmann | dansmith: if any test pass validatable=False, wait_until=SSHABLE, validation_resource=None then anyways we override validatable=True here https://review.opendev.org/c/openstack/tempest/+/882342/1/tempest/scenario/manager.py#338 | 19:35 |
| gmann | if we are taking it from kwargs the I will say pass it as it is to compute.py and let them fail if all these three combination are wrong | 19:36 |
| gmann | otherwise test will thing I passed validatable=False and why temepst doing validation | 19:37 |
| opendevreview | Ghanshyam proposed openstack/tempest master: Add release notes for 2023.2 cycle intermediate release. https://review.opendev.org/c/openstack/tempest/+/882353 | 19:55 |
| opendevreview | Ghanshyam proposed openstack/tempest master: Use stable constraint in tox to release Tempest new tag https://review.opendev.org/c/openstack/tempest/+/882354 | 20:01 |
| opendevreview | Ghanshyam proposed openstack/tempest master: Switch to the master constraint in tox.ini https://review.opendev.org/c/openstack/tempest/+/882355 | 20:03 |
| opendevreview | Ihar Hrachyshka proposed openstack/devstack stable/2023.1: git: support git checkout for a commit hash https://review.opendev.org/c/openstack/devstack/+/882297 | 20:06 |
| opendevreview | Ihar Hrachyshka proposed openstack/devstack stable/zed: git: support git checkout for a commit hash https://review.opendev.org/c/openstack/devstack/+/882298 | 20:07 |
| opendevreview | Ihar Hrachyshka proposed openstack/devstack stable/yoga: git: support git checkout for a commit hash https://review.opendev.org/c/openstack/devstack/+/882299 | 20:08 |
| opendevreview | Ihar Hrachyshka proposed openstack/devstack stable/xena: git: support git checkout for a commit hash https://review.opendev.org/c/openstack/devstack/+/882300 | 20:08 |
| gmann | dansmith: gouthamr kopecmartin tosky I started the tempest new release https://review.opendev.org/q/topic:tempest-34-2-0-release | 20:08 |
| tosky | gmann: thanks! | 20:10 |
| gmann | dansmith: and we can discuss if we need to include https://review.opendev.org/c/openstack/tempest/+/882342 | 20:10 |
| dansmith | gmann: okay I didn't really think the hard override was a problem but I can change it | 20:29 |
| gmann | dansmith: ok, thanks | 20:32 |
| gmann | dansmith: if cinder-tempest-plugin is not merging without new tempest release. I will try to do release it by tomorrow and you can bump new version in cinder-tempest-plugin | 20:35 |
| dansmith | gmann: does tempest not do the coordinated release? | 20:37 |
| gmann | dansmith: it is with cycle-with-intermediary release so we do at least one release per cycle but there can be more | 20:38 |
| gmann | so this will be intermediate release in 2023.2 | 20:38 |
| dansmith | okay | 20:38 |
| gmann | and at the end of 2023.2 we will do the final one | 20:38 |
| dansmith | ack | 20:38 |
| dansmith | gmann: clarification question for you here before I change it: https://review.opendev.org/c/openstack/tempest/+/882342/1/tempest/scenario/manager.py | 21:46 |
| gmann | checking | 21:47 |
| gmann | dansmith: replied, yeah we can make it None which make sense | 21:50 |
| dansmith | ack | 21:50 |
| tosky | gmann: about that change in cinder-tempest-plugin - I'm confused, if the last version of tempest which supports xena is already tagged, can't the last tagged version of cinder-tempest-plugin be considered the last one for xena as well? | 21:53 |
| tosky | I mean, isn't the tagging of <version>-last happening at the same time for tempest _and_ the plugins? | 21:54 |
| gmann | tosky: yes | 21:54 |
| gmann | all *-last are compatible as they are cut when tempest and plugins was working fine together | 21:54 |
| opendevreview | Dan Smith proposed openstack/tempest master: Allow passing validation_resources in scenario https://review.opendev.org/c/openstack/tempest/+/882342 | 21:55 |
| opendevreview | Dan Smith proposed openstack/tempest master: WIP: Also require resources for PINGABLE https://review.opendev.org/c/openstack/tempest/+/882102 | 21:55 |
| tosky | ok, so there is really no issue, apart from the dependency in requirements that may need to be tuned once the version of tempest which contains the tempest incompatible change is tagged | 21:56 |
| tosky | but just for downstream users, to make it clear that version is needed | 21:56 |
| dansmith | tosky: don't say there is really no issue because I spent my day retooling this :) | 21:56 |
| dansmith | *just* to make tosky happy :) | 21:56 |
| tosky | I was waiting for a confirmation from the-people-who-know-better-than-me :/ | 21:57 |
| gmann | tosky: yes | 21:57 |
| gmann | we just started thoser *-last and compatible version for old EM and downstream testing if anyone really testing in very old branches | 21:58 |
| gmann | otherwise main goal of tempest and plugin being branchless is TEST EVERYTHING SUPPORTED with master tests to catch the interop issues | 21:59 |
| gmann | means testing all with same set of tests to capture the backward incompatible and interop changes on service side | 21:59 |
| dansmith | I just pushed tempest and c-t-p patches to hopefully fix the last few issues introduced by this change, but if ya'll decide to I can just revert to what was working this morning before that diversion :) | 22:00 |
| opendevreview | Ghanshyam proposed openstack/devstack stable/yoga: [stable-only]: Set system_scope_supported https://review.opendev.org/c/openstack/devstack/+/882365 | 22:54 |
| opendevreview | Ghanshyam proposed openstack/devstack stable/yoga: [stable-only]: Set system_scope_supported https://review.opendev.org/c/openstack/devstack/+/882365 | 22:56 |
| opendevreview | Ghanshyam proposed openstack/devstack stable/xena: [stable-only]: Set system_scope_supported https://review.opendev.org/c/openstack/devstack/+/882366 | 22:57 |
| dansmith | gmann: the base tempest patch is going to fail because of some live migration test: test_server_connectivity_live_migration | 23:31 |
| dansmith | looks like it tests uptime to see if it was migrated or something | 23:31 |
| dansmith | so I'll recheck that in a bit, but otherwise It has passed everything else unless another tempest-slow test fails | 23:32 |
| dansmith | would really love to get this, c-t-p and devstack-ceph merged tomorrow (or at least closer) | 23:32 |
| dansmith | you know, to avoid stretching into yet another week :P | 23:33 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!