| opendevreview | OpenStack Proposal Bot proposed openstack/devstack master: Updated from generate-devstack-plugins-list https://review.opendev.org/c/openstack/devstack/+/989473 | 02:27 |
|---|---|---|
| opendevreview | Brian Haley proposed openstack/tempest master: Changed some more Neutron code to use project_id https://review.opendev.org/c/openstack/tempest/+/991754 | 03:33 |
| opendevreview | Merged openstack/devstack master: Add cinder config for multi-ceph setup https://review.opendev.org/c/openstack/devstack/+/952088 | 11:30 |
| opendevreview | Merged openstack/devstack master: Add support for Cinder replication https://review.opendev.org/c/openstack/devstack/+/953045 | 11:48 |
| -opendevstatus- NOTICE: Gerrit on review.opendev.org is being upgraded to version 3.13 and will be offline starting at 1600 UTC. We have allocated an hour for the outage window lasting until 1700 UTC. https://lists.opendev.org/archives/list/service-announce@lists.opendev.org/thread/UIFRQR34WYMF3KDMNQFMO222DIK3HTUD/ | 15:02 | |
| -opendevstatus- NOTICE: Gerrit on review.opendev.org is being upgraded to version 3.13 and will be offline momentarily. We have allocated an hour for the outage window lasting until 1700 UTC. https://lists.opendev.org/archives/list/service-announce@lists.opendev.org/thread/UIFRQR34WYMF3KDMNQFMO222DIK3HTUD/ | 16:01 | |
| dansmith | gmaan: I am completely flailing trying to run tempest with just a small set of tests | 16:56 |
| dansmith | gmaan: feeling like an idiot but I used to do `tox -eall -- someregexstuff` but all I get is "that is not a tempest command" | 16:56 |
| dansmith | help my friday brain? | 16:56 |
| gmaan | dansmith: it should work, you can try dierctly tempest run command tempest run --regex <someregexstuff> | 16:58 |
| gmaan | you are running locally right or want to configure in job | 16:58 |
| gmaan | tox -eall also run run command internally so not sure why tox ia giving error | 16:59 |
| dansmith | yeah I tried running manually too | 17:01 |
| dansmith | .tox/tempest/bin/tempest run --regex ServersUpdatePinnedAZ2104Test | 17:01 |
| dansmith | tempest: 'run --regex ServersUpdatePinnedAZ2104Test' is not a tempest command. See 'tempest --help'. | 17:01 |
| dansmith | glad I'm not crazy at least | 17:01 |
| gmaan | are you running from inside tempest dir? | 17:03 |
| dansmith | yes, working dir is /opt/stack/tempest | 17:03 |
| gmaan | humm | 17:03 |
| gmaan | and what tempest --help gives? I am wondering it is correctly picking installed tempest or not | 17:04 |
| dansmith | yeah doesn't show run as a command | 17:04 |
| dansmith | https://pastebin.com/NEE76zQG | 17:05 |
| gmaan | it seems it is not installed amd pointing to other CLI or something | 17:06 |
| gmaan | try to install from inside tempest dir "python3 setup.py install' | 17:08 |
| gmaan | or via virtual env | 17:10 |
| dansmith | I don't get why I have a tempest cli if not installed | 17:14 |
| dansmith | this is just from devstack | 17:14 |
| dansmith | .tox/tempest/bin/python setup.py install | 17:14 |
| dansmith | did stuff, help still shows no run | 17:14 |
| gmaan | it is again pointing to the globally installed/broken maybe? | 17:17 |
| dansmith | I dunno what you mean, I'm trying to run it direct from the tox venv ^ | 17:18 |
| dansmith | same error from `/opt/stack/data/venv/bin/tempest run` if that's what you mean | 17:18 |
| gmaan | install from tempest workspace not from venv | 17:18 |
| dansmith | I'm confused, shouldn't I be running tempest via tox? | 17:19 |
| dansmith | this is in a clean noble VM that just had devstack run in it | 17:19 |
| gmaan | it should work but it seems installation is broken there. either recreate the new venv | 17:19 |
| gmaan | not sure why it is happening, i never faced it in clean devstack but faced when i reinstated it after some tempest cli change | 17:20 |
| melwitt | fwiw I have never seen the bin/tempest command before, is that a relatively new thing? | 17:20 |
| dansmith | melwitt: tox itself runs tempest via the tempest command | 17:21 |
| melwitt | I used to do just like source /opt/stack/venv/bin/activate followed by just the tox -eall -- stuff | 17:21 |
| dansmith | gmaan: okay I'll destroy the VM and start over | 17:21 |
| melwitt | oh ok | 17:21 |
| dansmith | melwitt: eh? why would you activate the venv before running tox? tox runs everything in a venv | 17:21 |
| melwitt | after changing into /opt/stack/tempest | 17:22 |
| gmaan | dansmith: not VM but try python setup.py install from /opt/stack/tempest | 17:22 |
| dansmith | gmaan: with the devstack venv activated? | 17:22 |
| gmaan | no venv activated | 17:22 |
| melwitt | is tox not only installed in the venv? that's why I would, it's been awhile though | 17:22 |
| dansmith | gmaan: installed into the system python? | 17:22 |
| dansmith | melwitt: okay I always use system tox | 17:23 |
| gmaan | that is one I want to see if it works there | 17:23 |
| gmaan | if not then I will just recreate the venv and activate again and try from there | 17:23 |
| melwitt | in my devstack generally I have to activate /opt/stack/venv before I do anything bc that's where everything is installed | 17:23 |
| dansmith | gmaan: okay I can't do that into the system python lib without root.. is that really what you expect? | 17:23 |
| gmaan | with root, I mean if you are going to destroy the VM anyways I want to try if system wide installation fix it at system wide | 17:24 |
| gmaan | and is not then something really broken somewhere | 17:24 |
| dansmith | gmaan: sure, I just don't understand why the system python environment has anything to do with this | 17:24 |
| dansmith | gmaan: when you say "recreate the venv" are you talking about the devstack global venv (/opt/stack/data/venv) or the tox one (.tox/{full,tempest]) ? | 17:25 |
| gmaan | not sure, i hope you tried to delete the venv and ran tox again? | 17:25 |
| gmaan | tox one only .tox. i usually delete .tox and run fresh tox | 17:26 |
| dansmith | I didn't because this is a completely fresh install and .tox was just created this morning, but.. deleting.. | 17:26 |
| gmaan | k | 17:26 |
| dansmith | rm -Rf .tox | 17:26 |
| dansmith | tox -eall | 17:27 |
| dansmith | tempest: 'run --regex ' is not a tempest command. See 'tempest --help'. | 17:27 |
| gmaan | i have not tried fresh devsatack this week so nt sure if something broken there | 17:27 |
| gmaan | humm | 17:27 |
| dansmith | I will nuke my vm and start over | 17:27 |
| gmaan | k | 17:27 |
| dansmith | hopefully it will be monday before it finishes :D | 17:27 |
| gmaan | :) i thought you mentioned it takes 8-10 min for you to bring devstack up | 17:28 |
| dansmith | me? no :D | 17:28 |
| melwitt | I dunno, in my experience after fresh install I just activate the global venv (/opt/stack/data/venv) then cd /opt/stack/tempest then tox -eall -- stuff. if that doesn't work anymore then I will be sad haha | 17:28 |
| dansmith | and you're activating to get tox? | 17:29 |
| gmaan | melwitt: it should work, I am using it same way and working but it is my old setup not the fresh | 17:29 |
| melwitt | maybe not but it does get literally everything that got installed by devstack. everything is in there | 17:29 |
| dansmith | it seems so weird to activate a venv just to have a tool that will activate a vevn | 17:29 |
| melwitt | a lot of things don't work if I don't do that | 17:29 |
| melwitt | and gmaan I know what you are referring I think, for awhile after the devstack parallel and some async improvements some were saying devstack was installing quite fast | 17:30 |
| gmaan | if not dansmith then maybe it was sean-k-mooney mentioned somewher to me that it takes 8-10 min | 17:31 |
| gmaan | but do not remember exactly who and when | 17:31 |
| sean-k-mooney | on baremental yes | 17:31 |
| melwitt | I suspect it might be a mix, like somethings are inside the tempest venv but other things in the global. so I have a feeling you need to be in the global to begin with to maximize success | 17:31 |
| sean-k-mooney | you can deploy in under 10 mins | 17:32 |
| sean-k-mooney | locally it take me about 27 mins to do 2 node devsack in libvirt vms | 17:32 |
| dansmith | melwitt: but the point of tox is to literally isolate the environment the tests run in.. so the venv that tox creates (in .tox/$target) should be whole | 17:32 |
| sean-k-mooney | dansmith: i normaly jsut cd to /opt/stack/tempest and source .tox/tempest and run `tempest run ...` | 17:33 |
| dansmith | sean-k-mooney: yeah also tried that but no go | 17:33 |
| melwitt | I know :) just trying to help unblock you. maybe I'm totally wrong but just saying at least in the past it was not hard at all to run tempest. it used to be very easy | 17:33 |
| dansmith | melwitt: I know, which is why I'm asking .. never had a problem before :) | 17:34 |
| melwitt | could be a bad coincidence and maybe something is currently broken | 17:34 |
| melwitt | though we would expect to see broken in CI then I think | 17:34 |
| gmaan | dansmith: I tried on my env (last week installation), deleted .tox and then tox -eall did everything fine- creating fresh venv, running tempest run command successfully | 17:34 |
| dansmith | gmaan: ack, well, I'm re-stacking in a fresh VM now | 17:35 |
| gmaan | k | 17:35 |
| sean-k-mooney | the relevent config for tempet is stored in /opt/stack/tempest/ect/tempst.conf by the way | 17:35 |
| sean-k-mooney | so nuking the tox envs wotn remove the generate config that devstack created for you | 17:35 |
| sean-k-mooney | that the default tempest "workspace" location | 17:35 |
| dansmith | I don't seem to even be getting to the configs phase | 17:38 |
| dansmith | okie doke | 18:34 |
| dansmith | turns out I didn't commit my schema file, so when I pulled the branch to my test VM, it was missing that file, import of the tempest module would fail (silently) and give that behavior | 18:35 |
| sean-k-mooney | huh as in the api repsoncde schemas? i woudl have expected python or tempst to be very mad and load about that kind of missing import error but glad you fixed it | 18:42 |
| sean-k-mooney | i have seen things like that mased by pyc files before but not often | 18:42 |
| melwitt | oof | 18:49 |
| dansmith | yeah, kinda feel stupid but also kinda feel like that could have a better failure mode :D | 18:57 |
| gmaan | ohh, it should be import or module error when tempest try to register service client but i think its cliff hiding those and end up no command error | 19:04 |
| gmaan | or maybe other underlying module | 19:04 |
| dansmith | idk.. I should have tried master I guess, but there was zero chance in my head that that failure mode was related to the branch I have checked out with just some test files edited :) | 19:06 |
| melwitt | yeah I mean that's painful to have a simple mistake make a silent problem like that | 19:21 |
| melwitt | not sure if there is a way we could improve something to make it not hidden | 19:22 |
| gmaan | its seems its stevedore hide the actual error, | 19:22 |
| gmaan | 1. stevedore does not load the extension if file missing and it end up cliff not to load that cli: https://github.com/openstack/cliff/blob/master/cliff/commandmanager.py#L131 | 19:22 |
| gmaan | 2. when we run tempest command, cliff does not find it and raise no command error https://github.com/openstack/cliff/blob/master/cliff/commandmanager.py#L227 https://github.com/openstack/cliff/blob/master/cliff/app.py#L410 | 19:22 |
| gmaan | not sure if something stevedore can raise if module is not importable. I am not sure where to improve though | 19:23 |
| gmaan | as cliff cannot make assumption that command is missing due to file missing or stevedore actually did not find it bcz it is not installed | 19:23 |
| gmaan | we can try to pre-import the tempest.cmd in tempest/cmd/main before it actuall ship to cliff and if not importable then raise error with right msg | 19:27 |
| dansmith | well, maybe not worth all that | 19:43 |
| dansmith | just seems like a really obscure thing to debug | 19:43 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!