| noonedeadpunk | looks like LXC jobs are also not happy | 08:21 |
|---|---|---|
| noonedeadpunk | actually with same claim of absent glance.... | 08:22 |
| noonedeadpunk | huh... I wonder if `groups['glance_all'] | length > 0` would be evaluated as "true" here: https://zuul.opendev.org/t/openstack/build/e556fe6df5ad4234867ed3032a69c324/log/logs/etc/host/openstack_deploy/openstack_inventory.json.txt#974-983 | 08:25 |
| noonedeadpunk | I doubt so.... | 08:25 |
| noonedeadpunk | we probably should store tempest conffig as well... | 08:26 |
| jrosser | that packagecloud.io repo removal thing in rabbitmq role is odd | 08:42 |
| jrosser | if we dont add it, then we remove it (?), things break | 08:42 |
| jrosser | oh wait i totally misunderstand /o\ | 08:43 |
| jrosser | noonedeadpunk: is it the same problem with a conditional here https://opendev.org/openstack/openstack-ansible-os_tempest/src/branch/master/tasks/tempest_resources.yml#L216 | 09:00 |
| jrosser | needs also `tempest_service_available_glance` in the when: | 09:01 |
| jrosser | that is the task that fails | 09:01 |
| jrosser | and then there is contradiction here https://opendev.org/openstack/openstack-ansible-os_tempest/src/branch/master/tasks/tempest_resources.yml#L241-L242 | 09:05 |
| jrosser | `we set an empty string if it is not set above to ensure the template will parse correctly.` - except there are some mixture of conditionals on the set_fact tasks | 09:06 |
| noonedeadpunk | ahhhhh | 09:14 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_tempest master: Add missed openstack_resources_image condition https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/907886 | 09:35 |
| opendevreview | Andrew Bonney proposed openstack/openstack-ansible-os_nova master: Fix nova device_spec to support multiple values https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/907969 | 09:36 |
| noonedeadpunk | ugh, how I missed that all /o\ | 09:41 |
| noonedeadpunk | fwiw, I'm also about to look at ovn drivers for vpnaas/bgp/octavia | 09:42 |
| jrosser | cool - there is the patch from jamesdenton fixing up ovn+regular octavia too | 09:44 |
| noonedeadpunk | oh, I need to see it | 11:44 |
| nixbuilder | While following this guide (https://docs.openstack.org/openstack-ansible/latest/admin/scale-environment.html) I am attempting to add three compute hosts to the cloud. I did edit the /etc/openstack_deploy/openstack_user_config.yml per instructions... however when I ran the 'openstack-ansible setup-hosts.yml --limit localhost,compute035,compute036,compute037' II get errors saying that "Could not mat | 12:56 |
| nixbuilder | noring: compute035"... etc. Is there another step I need to run? | 12:56 |
| noonedeadpunk | nixbuilder: feels that error was trimmed | 13:03 |
| noonedeadpunk | Can you paste please what is specifically failing through https://paste.openstack.org/ | 13:03 |
| nixbuilder | noonedeadpunk: Ooops... my bad... while pasting the details I noticed what I did wrong :-) | 13:08 |
| noonedeadpunk | haha | 13:10 |
| noonedeadpunk | ok, cool then :D | 13:10 |
| noonedeadpunk | always glad to help hehe | 13:11 |
| dmsimard[m] | noonedeadpunk: btw it turns out that the ruamel.yaml issue was more problematic than I thought it was so I'm releasing a 1.7.1 of ara including a fix sometime soon | 13:27 |
| noonedeadpunk | Yeah, I saw your reply today in gitlab issue | 13:38 |
| noonedeadpunk | though I was quite sure that ruamel bugfix release reverted some behaviour back | 13:39 |
| noonedeadpunk | but potentially - they could decide to implement it again or smth... | 13:39 |
| noonedeadpunk | thanks for the heads-up though | 13:39 |
| frickler | noonedeadpunk: may I remind you of https://review.opendev.org/c/openstack/releases/+/906577 (yoga-unmaintained)? | 13:47 |
| mgariepy | is it only me or ubuntu are not publishing patches for some pkg in LTS release in the main repos ? | 13:51 |
| mgariepy | ex: https://ubuntu.com/security/notices/USN-5175-1 | 13:52 |
| jrosser | mgariepy: looks like it is missing for focal | 14:18 |
| jrosser | but ok for jammy https://ubuntu.com/security/CVE-2020-15025 | 14:18 |
| mgariepy | it's published in ESM | 14:19 |
| jrosser | oh well i mean focal without ubuntu-pro | 14:20 |
| mgariepy | i just wonder why that is. | 14:20 |
| mgariepy | focal is still supported .. | 14:20 |
| mgariepy | i think i'll move to debian | 14:21 |
| noonedeadpunk | frickler: thanks, I totally missed it | 14:26 |
| noonedeadpunk | #startmeeting openstack_ansible_meeting | 15:00 |
| opendevmeet | Meeting started Tue Feb 6 15:00:22 2024 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
| opendevmeet | The meeting name has been set to 'openstack_ansible_meeting' | 15:00 |
| noonedeadpunk | #topic rollcall | 15:00 |
| noonedeadpunk | o/ | 15:00 |
| jrosser | o/ hello | 15:03 |
| noonedeadpunk | #topic office hours | 15:04 |
| noonedeadpunk | So, I've started reviewing some patches from https://etherpad.opendev.org/p/osa-capi | 15:04 |
| noonedeadpunk | and most things looks jsut nice except maybe some small nits | 15:04 |
| jrosser | yeah, thankyou for taking a look | 15:05 |
| jrosser | this needs some attention really https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/900527 | 15:05 |
| jrosser | i am not happy with that tbh | 15:05 |
| noonedeadpunk | So... Maybe with just moving all playbooks to the plugins repo will eliminate a need for such role, as then we can continue using tasks include.... | 15:07 |
| noonedeadpunk | ie local ones | 15:07 |
| noonedeadpunk | as they will be local to playbooks | 15:07 |
| noonedeadpunk | But ofc depends on what exactly you don't like ofc | 15:07 |
| jrosser | perhaps though the pointt of that patch was i need to do the same thing in the ops repo collection | 15:07 |
| noonedeadpunk | Also potentially I didn't get the problem you was solving good enough | 15:07 |
| jrosser | so adjacent to playbook is unhelpful | 15:07 |
| NeilHanlon | o/ hi folks | 15:08 |
| noonedeadpunk | ok, yeah. And the reason you don't like this patch is...? | 15:08 |
| noonedeadpunk | \o/ | 15:08 |
| jrosser | line 172 here https://review.opendev.org/c/openstack/openstack-ansible-ops/+/901450/12 | 15:09 |
| jrosser | sorry here https://review.opendev.org/c/openstack/openstack-ansible-ops/+/901450/12/mcapi_vexxhost/playbooks/k8s_install.yml#172 | 15:09 |
| jrosser | what i don't like is that the only way to make it "work" is to run it under roles: | 15:10 |
| noonedeadpunk | and doing that with import_role? | 15:10 |
| noonedeadpunk | ugh | 15:10 |
| noonedeadpunk | ok, I don't like that either then | 15:10 |
| jrosser | then the vars defined in defaults/main.yml just are not visible | 15:10 |
| jrosser | right | 15:10 |
| noonedeadpunk | how does ceph_ansible handles it? | 15:10 |
| jrosser | so we need some other way to do this | 15:10 |
| jrosser | probably with set_fact as a guess | 15:11 |
| noonedeadpunk | https://github.com/ceph/ceph-ansible/tree/main/roles/ceph-defaults | 15:11 |
| noonedeadpunk | no, not really I guess | 15:11 |
| jrosser | oh i was thinking of https://github.com/ceph/ceph-ansible/tree/main/roles/ceph-facts/tasks | 15:11 |
| noonedeadpunk | But I saw ceph-defaults being included/respected as well actually | 15:12 |
| jrosser | interesting they are using import_role | 15:12 |
| jrosser | the empty tasks file is maybe worth a try | 15:12 |
| noonedeadpunk | we have them in roles :( https://opendev.org/openstack/openstack-ansible/src/branch/master/playbooks/ceph-install.yml#L93-L99 | 15:12 |
| jrosser | anyway - i really had some difficulty to make that work | 15:13 |
| jrosser | but tbh i think that i am missing some obvious thing | 15:13 |
| noonedeadpunk | Hm, I really don't understand how it works actually.... | 15:14 |
| noonedeadpunk | as - should included vars in role work the same way then? | 15:15 |
| noonedeadpunk | ie - being available outside of the role scope? | 15:15 |
| noonedeadpunk | yeah, that worth to have a closer look obviously | 15:16 |
| jrosser | right its a very good question | 15:16 |
| jrosser | and one that obviously is important for ceph-ansible too | 15:16 |
| noonedeadpunk | and potentially it depends on import vs include | 15:18 |
| jrosser | i will write a test after the meeting | 15:18 |
| noonedeadpunk | btw tempest fix https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/907886 seems to be working now - https://review.opendev.org/c/openstack/openstack-ansible-rabbitmq_server/+/907758 is happy | 15:19 |
| noonedeadpunk | there're bunch of improvments that should be done for openstack_resources role, first of which to move service_setup role under it | 15:20 |
| noonedeadpunk | I wonder if it's possible to do some kind of collection mapping magic to avoid patching all roles... | 15:20 |
| jrosser | why do we need to do that btw? | 15:20 |
| noonedeadpunk | well..... | 15:21 |
| noonedeadpunk | amount of skipped tasks potentially have smth to do with that: https://opendev.org/openstack/openstack-ansible-plugins/src/branch/master/roles/openstack_resources/tasks/identity.yml#L16-L37 | 15:21 |
| noonedeadpunk | as role is included 2 times when we'd need to iterate properly over list of domains, since in service_setup context there's only 1 domain is possible | 15:23 |
| noonedeadpunk | so for each domain, role is fully included for no good reason more or less | 15:23 |
| noonedeadpunk | just to keep compatibility of formats.... | 15:23 |
| noonedeadpunk | But dunno | 15:23 |
| noonedeadpunk | maybe we don't need that at all | 15:23 |
| noonedeadpunk | I was having some thought after looking through bunch of bugs lately. | 15:28 |
| noonedeadpunk | If there's any will among you folks to do a bug triage/sorting day and spend it on closing/reviving bugs? | 15:29 |
| noonedeadpunk | like it feels that 90% of them are not valid, but there're 10% that really worth paying attention to | 15:29 |
| noonedeadpunk | and could be missed/still valid today (like https://bugs.launchpad.net/openstack-ansible/+bug/1896589) | 15:30 |
| noonedeadpunk | as it's pretty embarassing to have that bug for 4 years.... | 15:30 |
| noonedeadpunk | I know that it's not best time investment and might be hard to justify... | 15:31 |
| noonedeadpunk | So no is also a good answer | 15:31 |
| noonedeadpunk | Like if I'd sort bugs by update date - it's getting to 2020 after ~30-40 of them | 15:33 |
| * noonedeadpunk just found another valid one | 15:37 | |
| jrosser | would be good to have a bug squash at some point | 15:37 |
| jrosser | maybe if it's just sorting out the ancient/invalid and leaving the actual ones | 15:38 |
| noonedeadpunk | damiandabrowski: mgariepy wanna take part in that ^ :) | 15:38 |
| noonedeadpunk | *it was a question | 15:38 |
| noonedeadpunk | Like take couple of hours (or half of the day) for bug fighting day | 15:39 |
| noonedeadpunk | and close what's not relevant / triage set priority for what is valid | 15:39 |
| jrosser | that would be good - otherwise we just fight all the rubbish ones and waste time | 15:46 |
| mgariepy | yes sure when ? | 15:55 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_nova master: Always distribute qemu config file https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/908160 | 15:59 |
| noonedeadpunk | let's agree on some date/time:) | 15:59 |
| noonedeadpunk | the closer to end of week - the calmer days in terms of meetings/distracting factors. | 16:00 |
| noonedeadpunk | that's for me at least :D | 16:00 |
| noonedeadpunk | I can send out a poll after the meeting actually | 16:00 |
| mgariepy | ok | 16:01 |
| noonedeadpunk | awesome | 16:01 |
| noonedeadpunk | #endmeeting | 16:01 |
| opendevmeet | Meeting ended Tue Feb 6 16:01:31 2024 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 16:01 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2024/openstack_ansible_meeting.2024-02-06-15.00.html | 16:01 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/openstack_ansible_meeting/2024/openstack_ansible_meeting.2024-02-06-15.00.txt | 16:01 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2024/openstack_ansible_meeting.2024-02-06-15.00.log.html | 16:01 |
| nixbuilder | Can someone explain what the repo_all group is for? I am attempting to add three compute hosts and get the following errors https://paste.openstack.org/show/by3QYUERRsRQBPaGNh1J/ | 16:30 |
| nixbuilder | This is when running 'openstack-ansible setup-openstack.yml --limit localhost,compute35,compute36,compute37' | 16:32 |
| nixbuilder | Here is how pkg_repo is defined https://paste.openstack.org/show/be5poKiwk3yuonOAH6Y9/ | 16:35 |
| jrosser | nixbuilder: the compiling of python wheels is done on one of the repo_all group | 16:35 |
| jrosser | and you have rocky 9.3 new computes but your infra nodes are 9.2 | 16:35 |
| jrosser | so it doesnt match | 16:35 |
| jrosser | that might be a good question we ask NeilHanlon if it's valid to make a distinction between minor releases of Rocky for this | 16:36 |
| nixbuilder | Ahhh... ok. I guess I can rebuild those new computes on Rocky 9.2 then. | 16:39 |
| jrosser | is that possible? i'm not particualty familiar with the RH derivatives | 16:39 |
| nixbuilder | jrosser: Should be possible... I hope. | 16:40 |
| jrosser | it might be that the minor version distinction that we make there is not valid | 16:41 |
| jrosser | but i'm not sure about that really | 16:41 |
| nixbuilder | Hmmm... perhaps when I did the 'dnf upgrade' it went from 9.2 to 9.3 by itself.... yuch. | 16:42 |
| noonedeadpunk | We actually have fixed this bug | 16:43 |
| nixbuilder | noonedeadpunk: How??? Where??? | 16:43 |
| noonedeadpunk | It's ... zed? Or what you said?:) | 16:43 |
| noonedeadpunk | But I believe it's patrch fixing the issue: https://opendev.org/openstack/ansible-role-python_venv_build/commit/7ca65b91379f2b6e1086e03e33e623968093bac0 | 16:44 |
| noonedeadpunk | nixbuilder: but other way around would be doing `dnf update` on one of repo containers | 16:44 |
| noonedeadpunk | I'd expect it be updated to rocky 9.3 this way and satisfying the requirement | 16:45 |
| noonedeadpunk | just important to clean facts for the container | 16:45 |
| NeilHanlon | nixbuilder: you.. *can* run 9.2, but you shouldn't. We don't test against it, and Rocky does not provide updates anymore for anything except 9.3 and 8.9 | 16:49 |
| nixbuilder | NeilHanlon: So I really should update all infra and compute nodes to 9.3 then.... right? | 16:52 |
| nixbuilder | NeilHanlon: Also all infra and compute nodes are on bare metal | 16:53 |
| NeilHanlon | yep, 9.3 would be the best to test against -- we run dnf upgrade on the components (e.g., if you're using LXC), so the containers we create will be upgraded as well | 16:54 |
| nixbuilder | NeilHanlon: So after I upgrade the OS, how do I update the repo's on the infra nodes? | 16:56 |
| jrosser | nixbuilder: the contents of the "repo server" is python wheels | 17:02 |
| jrosser | they are built on-demand as you run throught the other playbooks | 17:02 |
| opendevreview | Merged openstack/openstack-ansible-os_tempest master: Add missed openstack_resources_image condition https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/907886 | 17:02 |
| jrosser | so theres really nothing to do, just get your repo hosts to 9.3, and try again to deploy your new computes | 17:02 |
| jrosser | noonedeadpunk: https://paste.opendev.org/show/bpBRwAAUELSS5YYxE7l9/ | 17:02 |
| jrosser | this is what i struggled to make work | 17:03 |
| noonedeadpunk | nixbuilder: you should update only 1 repo container in fact | 17:04 |
| noonedeadpunk | or as bare minimum | 17:04 |
| nixbuilder | noonedeadpunk: OK... I will only update one infra node and then retry deploying my new compute nodes. | 17:06 |
| noonedeadpunk | repo container != infra node fwiw | 17:06 |
| noonedeadpunk | updating infra node won't bring container version up | 17:06 |
| jrosser | bare metal deployment i think | 17:07 |
| jrosser | also TIL DEFAULT_GATHER_SUBSET is deprecated and the alternative, module_defaults is pretty unhelpful | 17:08 |
| noonedeadpunk | nixbuilder: basically - `ansible -m command -a "dnf update" 'repo_all[0]'` | 17:08 |
| noonedeadpunk | ah, ok, for bare metal then yes | 17:08 |
| noonedeadpunk | yeah, I'm looking at this deprecation warning for a while | 17:08 |
| noonedeadpunk | clueless on how to replace | 17:08 |
| noonedeadpunk | leaving trouble for tomorrow me :D | 17:09 |
| noonedeadpunk | jrosser: frankly speaking, that role does work looks like undocumented behaviour or even a bug... | 17:14 |
| jrosser | so i am not sure what i am missing compared to ceph_ansible? | 17:14 |
| jrosser | they have a role to set defaults and it does seem to work | 17:15 |
| jrosser | and to make this proper it needs to be conditional too, as we have source/distro vars to choose from | 17:16 |
| noonedeadpunk | I'm actually thinking more towards set_facts and then do include some vars/tasks based on source/distro | 17:17 |
| noonedeadpunk | but dunno | 17:17 |
| noonedeadpunk | ok, I have smth I guess | 17:18 |
| noonedeadpunk | that should do the trickyt | 17:19 |
| noonedeadpunk | I can hardly explain, but let be propose some DNM I guess | 17:20 |
| jrosser | i'm not sure if being in a collection changes the behaviour too | 17:20 |
| noonedeadpunk | Dunno, but I think this approach will work nicely... | 17:20 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-plugins master: Add role for provisioning default variables for install methods https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/908179 | 17:26 |
| noonedeadpunk | jrosser: ^ | 17:26 |
| jrosser | oh of course :) | 17:27 |
| jrosser | nice | 17:27 |
| noonedeadpunk | I dunno if it should be install_method or service_install_method though.... | 17:28 |
| noonedeadpunk | like currently it's install_method... But then we have service_install_method in all roles? | 17:29 |
| jrosser | i think i was too hard trying to port the include_vars directly and completely forgot i could take the task with it too | 17:29 |
| noonedeadpunk | for some reason I thought this should not work either | 17:29 |
| noonedeadpunk | Also it's 1 task more... | 17:29 |
| noonedeadpunk | but well. | 17:29 |
| jrosser | `service_install_method` is odd indeed | 17:31 |
| jrosser | as if you wanted to override then surely <foo>_install_method would be the default to set | 17:32 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_octavia master: Implement support for octavia-ovn-provider driver https://review.opendev.org/c/openstack/openstack-ansible-os_octavia/+/868462 | 17:54 |
| noonedeadpunk | My guess is that idea was to replace install_method with service_install | 17:54 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_octavia master: Implement support for octavia-ovn-provider driver https://review.opendev.org/c/openstack/openstack-ansible-os_octavia/+/868462 | 18:10 |
| damiandabrowski | noonedeadpunk: regarding bug triage/sorting day - sure, I'm happy to help | 18:54 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!