| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2021-10-11 | |||
| 15:25:19 | bauzas | melwitt: ack, opening a tab and looking at them once we discuss about other issues ;) | |
| 15:25:28 | melwitt | thanks :) | |
| 15:35:15 | opendevreview | Elod Illes proposed openstack/nova stable/stein: [stable-only] Pin virtualenv and setuptools https://review.opendev.org/c/openstack/nova/+/813451 | |
| 17:31:32 | opendevreview | Dmitrii Shcherbakov proposed openstack/nova master: [yoga] Add PCI VPD Capability Handling https://review.opendev.org/c/openstack/nova/+/808199 | |
| 17:31:33 | opendevreview | Dmitrii Shcherbakov proposed openstack/nova master: [yoga] Support remote-managed SmartNIC DPU ports https://review.opendev.org/c/openstack/nova/+/812111 | |
| 19:39:54 | mdbooth | Not really an OpenStack issue, but as I'm hitting it running devstack on Ubuntu Focal I'm hoping somebody here has seen it before 🤞 c-vol fails to come up because lvcreate fails to create a thin volume due to: "modprobe: FATAL: Module dm-thin-pool not found in directory /lib/modules/5.4.0-1047-kvm". Anybody seen that? | |
| 19:41:02 | mdbooth | My google-fu is unfortunately not up to the task. | |
| 19:57:42 | clarkb | mdbooth: you might have to install something like linux-image-extra-virtual but for the kvm kernel | |
| 19:58:34 | mdbooth | clarkb: Thanks. I was just coming to the realisation that this image might have a stripped down kernel for kvm in it. I think I'm going to toss it and try a different cloud image. | |
| 19:59:23 | clarkb | linux-image-virtual doesn't have that module either fwiw, but they do ship linux-image-extra-virtual which I expect includes it | |
| 20:00:08 | mdbooth | Thanks! | |
| 20:00:11 | clarkb | I don't see a virtual package for extra kvm image kernel modules but there are package for specific kernel versions like linux-modules-5.4.0-1009-kvm | |
| 20:01:54 | clarkb | I did check the kernel config for the linux-image-virtual kernel I've got and it lists DM_THIN_PROVISIONING=m so there should be a module for it in a package somewhere | |
| 20:21:19 | mdbooth | Looking good: # find /usr/lib/modules/5.4.0-88-generic/ -name "*thin*" | |
| 20:21:32 | mdbooth | /usr/lib/modules/5.4.0-88-generic/kernel/drivers/md/dm-thin-pool.ko | |
| 20:21:38 | mdbooth | clarkb: Thanks again :) | |
| #openstack-nova - 2021-10-12 | |||
| 07:45:30 | bauzas | good morning Nova | |
| 07:57:36 | pslestang | Hey all, I'm facing an issue on a fresh installed devstack when running tox in nova directory which seems related to a conflict between oslo-vmware and suds-jurko | |
| 07:57:45 | pslestang | ERROR: Cannot install -r /opt/stack/nova/test-requirements.txt (line 28) because these package versions have conflicting dependencies. | |
| 07:57:47 | pslestang | The conflict is caused by: | |
| 07:57:49 | pslestang | oslo-vmware 3.9.1 depends on suds-jurko>=0.6 | |
| 07:57:51 | pslestang | The user requested (constraint) suds-jurko===0.6 | |
| 07:58:16 | pslestang | Is there a recommended fix or a solution to handle that? | |
| 07:58:34 | bauzas | pslestang: weirdo | |
| 07:58:59 | bauzas | which tox target ? | |
| 07:59:34 | bauzas | I usually don't run my unittest on devstack but rather on my local machine | |
| 08:00:06 | bauzas | at least because sometimes my whole local nova repo can disappear if I clean up devstack :) | |
| 08:00:40 | bauzas | I rather prefer to have a local working repo and a git remote on my devstack | |
| 08:00:43 | pslestang | I did not specify any target | |
| 08:01:02 | bauzas | so, just 'tox' ? | |
| 08:01:05 | pslestang | yep | |
| 08:01:10 | bauzas | ok, so unittests | |
| 08:01:33 | bauzas | you should tox -r | |
| 08:01:41 | bauzas | which will recreate the tox venv | |
| 08:01:56 | pslestang | ok trying | |
| 08:02:18 | bauzas | and you should also specify a target :) | |
| 08:02:28 | bauzas | -e py38 per say :) | |
| 08:03:15 | bauzas | also, running the whole testsuite is nice but... needs a lot of coffee to get answers :) | |
| 08:03:39 | bauzas | you should rather point tox to check only a few tests you know | |
| 08:03:40 | pslestang | running 'tox -r' it does not change anything | |
| 08:04:02 | bauzas | pslestang: paste the outputs please | |
| 08:04:17 | bauzas | so I can try to reproduce the exact command | |
| 08:06:27 | bauzas | mmm, maybe we have a regression in our CI, checking zuul | |
| 08:07:22 | bauzas | actually, no, all looks good https://zuul.openstack.org/builds?project=openstack/nova | |
| 08:09:41 | pslestang | which paste tool are you using, paste.openstack.org tells me that I send spam | |
| 08:13:27 | pslestang | bauzas: https://privatebin.net/?18300286850a2784#BxNwm4jkzNKC4gYbPiTS11uXJvLUuq6UVHxRAv7yYXyK | |
| 08:14:07 | bauzas | pslestang: you can force paste.o.o to accept your paste AFAICR | |
| 08:14:48 | bauzas | hah | |
| 08:14:49 | bauzas | error in suds-jurko setup command: use_2to3 is invalid. | |
| 08:15:12 | bauzas | that rings a bell to me :) | |
| 08:18:05 | bauzas | testing locally | |
| 08:18:40 | pslestang | ok thank you | |
| 08:19:15 | bauzas | pslestang: one quick thought could be that your nova local repo is a bit old and not getting the latest releases | |
| 08:19:58 | bauzas | pslestang: yeah, confirmed, I can run tox locally on py38 target without any problem | |
| 08:20:05 | pslestang | The repo is fresh from this morining | |
| 08:20:18 | bauzas | ? | |
| 08:20:31 | bauzas | I just did a git pull before I recreated the venv :) | |
| 08:21:03 | bauzas | pslestang: last commit on your nova git repo ? | |
| 08:21:21 | pslestang | fdfdba265833d237e22676f9a223ab8ca0fe1e03 | |
| 08:21:21 | bauzas | mine is : | |
| 08:21:22 | bauzas | Date: Fri Oct 8 02:36:41 2021 +0000 | |
| 08:21:22 | bauzas | Author: Zuul <zuul@review.opendev.org> | |
| 08:21:22 | bauzas | Merge: a8d3ab2513 ad227d7085 | |
| 08:21:22 | bauzas | commit fdfdba265833d237e22676f9a223ab8ca0fe1e03 (HEAD -> master, origin/master, origin/HEAD) | |
| 08:21:23 | bauzas | Merge "Update min supported service version for Yoga" | |
| 08:21:35 | pslestang | same here | |
| 08:22:07 | bauzas | pslestang: can you try to run tox on a local repo which is *not* devstack ? | |
| 08:22:20 | bauzas | git clone nova and shoot the tox runs | |
| 08:22:24 | pslestang | let me give a try | |
| 08:22:50 | bauzas | I'd suspect some pip cache | |
| 08:25:40 | pslestang | same result, I'm gonna try by cleaning pip cache it sounds like a good point | |
| 08:44:01 | frickler | pslestang: bauzas: https://bugs.launchpad.net/devstack/+bug/1946340 . we should mark that as FAQ somewhere | |
| 08:44:48 | frickler | the issue isn't seen in our CI because we use pre-built wheels | |
| 08:45:18 | frickler | if you do a local build without old pip cache, it will fail the same way | |
| 08:46:08 | pslestang | frickler: you're completely right | |
| 08:47:50 | bauzas | dependency* even | |
| 08:47:50 | bauzas | pslestang: yeah, eventually spotted the depency issue with devstack by reinstalling my testbed | |
| 08:51:17 | pslestang | I will try by pining the setuptools to 58.0.0 as proposed https://bugs.launchpad.net/devstack/+bug/1946340/comments/5 | |
| 08:57:19 | bauzas | ergh, network glitch here | |
| 09:06:42 | bauzas | frickler: fwiw, cinder installation on devstack failed due to this | |
| 09:07:00 | bauzas | I can't see any proper workaround | |
| 09:07:33 | sean-k-mooney[m] | for now you just remove oslo.vmware from cinders requirements file | |
| 09:07:47 | bauzas | frickler: pslestang: about documenting it, we already discussed about the 2_to_3 issues we got and the pinned setuptools and we said "well, this isn't a nova issue, so why should we create some relnotes for this ?" | |
| 09:07:59 | sean-k-mooney[m] | ill try and submit a fix to oslo.vmware to use suds-community shortly | |
| 09:08:15 | bauzas | sean-k-mooney: locally, you mean ? ok | |
| 09:08:19 | sean-k-mooney[m] | this is not a devstack bug | |
| 09:08:23 | sean-k-mooney[m] | yes locally | |
| 09:08:29 | sean-k-mooney[m] | i just commented it out | |
| 09:09:01 | sean-k-mooney[m] | but the issue is in oslo.vmware it depens on suds-jaroko or something like that | |
| 09:09:11 | sean-k-mooney[m] | suds-community is python 3 compatible | |
| 09:09:31 | sean-k-mooney[m] | so once we fix there requirements file you can skip the local workaround | |
| 09:09:42 | bauzas | yes | |
| 09:15:29 | frickler | bauzas: yes, local workaround is to remove oslo.vmware from cinder requirements | |
| 09:15:37 | bauzas | kk | |
| 09:16:18 | frickler | sean-k-mooney[m]: there is already one, it needs to be added to reqs first, though: https://review.opendev.org/c/openstack/oslo.vmware/+/813377 | |
| 09:17:14 | frickler | sean-k-mooney[m]: also, the most recent version of suds-community still hasn't fixed the 2to3 issue. but they provide wheels on pypi, which masks that issue | |
| 09:17:25 | sean-k-mooney[m] | ah ok good | |
| 09:17:48 | frickler | they need a new tag with this one https://github.com/suds-community/suds/pull/58 | |
| 09:21:10 | frickler | ah, that'll be a major version bump, so the test release is suds-community-1.0.0b1 | |
| 09:21:50 | frickler | which I can install locally from source just fine, so that looks like progress | |