| opendevreview | David proposed openstack/watcher-dashboard master: Use ID-based selector for login button https://review.opendev.org/c/openstack/watcher-dashboard/+/996215 | 07:47 |
|---|---|---|
| opendevreview | David proposed openstack/watcher-dashboard master: Add Playwright tests to validate goals and strategies https://review.opendev.org/c/openstack/watcher-dashboard/+/996067 | 07:47 |
| opendevreview | David proposed openstack/watcher-dashboard master: Add Playwright tests to validate goals and strategies https://review.opendev.org/c/openstack/watcher-dashboard/+/996067 | 09:16 |
| opendevreview | David proposed openstack/watcher-dashboard master: Add Playwright tests to validate goals and strategies https://review.opendev.org/c/openstack/watcher-dashboard/+/996067 | 09:59 |
| opendevreview | David proposed openstack/watcher-dashboard master: Add Playwright tests to validate goals and strategies https://review.opendev.org/c/openstack/watcher-dashboard/+/996067 | 10:00 |
| jgilaber | Hi! The watcher irc meeting will start in ~30 minutes, add your topics to the agenda https://etherpad.opendev.org/p/openstack-watcher-irc-meeting#L31 | 12:30 |
| jgilaber | #startmeeting watcher | 13:00 |
| opendevmeet | Meeting started Thu Jul 9 13:00:46 2026 UTC and is due to finish in 60 minutes. The chair is jgilaber. Information about MeetBot at http://wiki.debian.org/MeetBot. | 13:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 13:00 |
| opendevmeet | The meeting name has been set to 'watcher' | 13:00 |
| jgilaber | Hi! Who is around today? | 13:01 |
| sean-k-mooney | acto/ | 13:01 |
| sean-k-mooney | o/ | 13:01 |
| morenod | o/ | 13:01 |
| chandankumar | o/ | 13:01 |
| jgilaber | courtesy ping: dviroel amoralej rlandy | 13:01 |
| rlandy | o/ | 13:01 |
| jgilaber | let's give folks a minute and we can start | 13:01 |
| jgilaber | feel free to add topics to the agenda meanwhile https://etherpad.opendev.org/p/openstack-watcher-irc-meeting#L31 | 13:02 |
| morenod | dviroel and amoralej are out today | 13:02 |
| jgilaber | right, so everyone that usually attends is here | 13:02 |
| jgilaber | let's get started | 13:02 |
| jgilaber | #topic Review Priority label proposal | 13:03 |
| jgilaber | this one I guess it's from dviroel | 13:03 |
| jgilaber | he proposed a patch to add the review prioiry label we discussed last week | 13:03 |
| jgilaber | #link https://review.opendev.org/c/openstack/project-config/+/995886 | 13:03 |
| jgilaber | let's try to review it to agree on the label and meaning for each vote | 13:04 |
| jgilaber | personally I think the current state captures well what we discussed in the last meeting | 13:05 |
| sean-k-mooney | ya i think its mergeable as is | 13:06 |
| jgilaber | any comment/question on this topic? | 13:06 |
| sean-k-mooney | tehre is the convertion inline baout the exact wording but i dont think we need ot bikeshed too much on this | 13:06 |
| jgilaber | agreed. We can discuss further in the patch if necessary, moving to the next topic | 13:08 |
| jgilaber | #topic Spec-related changes | 13:08 |
| jgilaber | we have not merged the spec for preemptible instances | 13:08 |
| jgilaber | #link https://review.opendev.org/c/openstack/watcher-specs/+/987171 | 13:08 |
| jgilaber | we're past the spec freeze, do we want to merge it for this cycle | 13:09 |
| jgilaber | I think the spec is in a good state to be merged | 13:09 |
| sean-k-mooney | im currently reading it | 13:11 |
| sean-k-mooney | i started just before the meeting | 13:11 |
| sean-k-mooney | i think tis proably close enough that we can proceed with it | 13:11 |
| sean-k-mooney | the fact the comment form the previosu iteration were not resolved makes it a little harder to review | 13:11 |
| jgilaber | yep, I had to do a bit of scrolling up and down to check some previous conversations | 13:12 |
| jgilaber | it looks like we can discuss further in the review and decide what to do there | 13:13 |
| sean-k-mooney | well i normally change the base to the version where the comment was left | 13:13 |
| sean-k-mooney | but ya i will +1 or +2 ocne im doen and we can check with doug later/tomrorow if we want to merge i think | 13:14 |
| jgilaber | sounds good to me | 13:14 |
| jgilaber | second patch | 13:14 |
| jgilaber | #link https://review.opendev.org/c/openstack/watcher-specs/+/996114 | 13:14 |
| jgilaber | I pushed a patch adding the dir and files for 2027.1 specs | 13:14 |
| jgilaber | we have a couple of open specs that we'll want to push there | 13:15 |
| jgilaber | it's just the usual change, just mentioning it here for visibility | 13:15 |
| jgilaber | I don't think we need to discuss much about it, if you see anything wrong with the patch, please flag it in the review | 13:16 |
| jgilaber | moving to the next topic | 13:16 |
| jgilaber | #topic Watcher Dashboard reviews | 13:16 |
| jgilaber | I think this one is from morenod? | 13:16 |
| morenod | yes | 13:16 |
| morenod | all three | 13:16 |
| morenod | first 2 are dependencies for running the tests on different environments | 13:16 |
| morenod | the login one is about make tests working both upstream and downstream, login button is different, so with that change, it work on both | 13:17 |
| morenod | and the https is for envs installed with autosigned https certs | 13:17 |
| morenod | the real one is the third, for checking that the name and the description of the goals and strategies is correct on the dashboard. I get them from the API and compare | 13:18 |
| sean-k-mooney | the login button shoudl not be diffent as in it should have the same id | 13:19 |
| sean-k-mooney | it can have diffent css | 13:19 |
| sean-k-mooney | but we shoudl not be depenidng on the css | 13:19 |
| morenod | yep, that is what change does, use the ID instead of the text of the button | 13:19 |
| sean-k-mooney | ah cool | 13:19 |
| sean-k-mooney | do you have links to those for the meeting logs | 13:20 |
| chandankumar | we also thought of going like this self.page.get_by_role("button", name=re.compile(r"Sign In|connect", re.IGNORECASE)).click(), but current fix is much easier | 13:20 |
| sean-k-mooney | we shoudl avoid using names as it woudl break on translation | 13:21 |
| jgilaber | links for the patches | 13:21 |
| jgilaber | #link https://review.opendev.org/c/openstack/watcher-dashboard/+/996069 | 13:21 |
| jgilaber | #link https://review.opendev.org/c/openstack/watcher-dashboard/+/996215 | 13:21 |
| sean-k-mooney | as a general patther we shoudl alwasy use the ids to select items where possibel | 13:21 |
| jgilaber | #link https://review.opendev.org/c/openstack/watcher-dashboard/+/996067 | 13:21 |
| sean-k-mooney | *pattern | 13:21 |
| morenod | agree | 13:22 |
| sean-k-mooney | its odd your gettign https erros by the way | 13:22 |
| sean-k-mooney | i think the ignore is fine | 13:22 |
| sean-k-mooney | its just odd becaue the ca that generates them in devstack should be in the cert store | 13:23 |
| morenod | it is disabled by default btw, you can enable it based on a parameter | 13:23 |
| sean-k-mooney | and the same shoudl be ture for the certs generated by certmanager in our downstream | 13:23 |
| sean-k-mooney | yep just noting that when you enable tls in devstack | 13:24 |
| sean-k-mooney | then tools installed on the host shoudl trust the cert out of the box | 13:24 |
| sean-k-mooney | im fine with the opt out | 13:24 |
| sean-k-mooney | but in both our downstream nd upstream ci's it shoudl not be needed nor shoudl it really be needed for local dev but its ok if you dont wnat to enable tls ectra locally | 13:25 |
| sean-k-mooney | i think we can move on | 13:27 |
| morenod | yep | 13:27 |
| jgilaber | ok, moving on, thanks for working on this morenod! | 13:27 |
| jgilaber | #topic Bugs | 13:28 |
| jgilaber | we have a new bug to triage | 13:28 |
| jgilaber | #link https://bugs.launchpad.net/watcher/+bug/2160195 | 13:28 |
| jgilaber | this is valid and known, but it'll probably be too big in scope to be treated as a bug | 13:29 |
| morenod | I created this morning after the errors on the multicell job, but after talking to sean-k-mooney, we probably can close it, as it is not expected that watcher will work with multicell envs | 13:29 |
| sean-k-mooney | watcher can but only if you use scopes | 13:29 |
| sean-k-mooney | which is why i was instient that scopes are part of the core api an must work for all stragies | 13:29 |
| sean-k-mooney | cross cell live migraiton is not a thing at least not today | 13:30 |
| sean-k-mooney | and cross cell move oeprations are disbaled by default | 13:30 |
| sean-k-mooney | so the only way to use watcher in a multi cell env | 13:30 |
| sean-k-mooney | is to have all autid use scops and scope to host within a cell | 13:30 |
| jgilaber | could we add cell info to the model? | 13:31 |
| morenod | the problem at this moment, is that nova do not provides cell info on the api | 13:31 |
| sean-k-mooney | cells are an internal nova sclaing mechanisum and are intentionl not expsoed via the api today | 13:31 |
| morenod | so we cannot detect cell config when executing the tests | 13:31 |
| sean-k-mooney | addign it would be non trivial | 13:31 |
| sean-k-mooney | it would be possibel but likely take nova changes | 13:32 |
| sean-k-mooney | the best way to do this would be to have nova's conductors enrole each compute node into a placement aggreate based on the cell they manage | 13:32 |
| sean-k-mooney | comptue nodes are not ment ot know what cell they are in | 13:33 |
| sean-k-mooney | so they cannot do that themselves | 13:33 |
| sean-k-mooney | the info is in the cell mapping tablle in nova api db but its purly used inrenally | 13:33 |
| jgilaber | ack, I see, then I think we can close this bug report with a comment summarizing this dicussion | 13:33 |
| sean-k-mooney | its perhaps a pain point ot raise with the nova team in a future ptg | 13:33 |
| sean-k-mooney | its not unfixable its just intenrally not expsoed today | 13:34 |
| morenod | we can ask to create the same host aggregates than cells on multicell jobs, and assume that they are correct, and manage host aggregates as cells in the tests | 13:34 |
| sean-k-mooney | morenod: that is a workaround yes | 13:34 |
| sean-k-mooney | that is efectilly what i was suggestign automating in nova eventually | 13:35 |
| sean-k-mooney | melwitt: just an fyi ^ | 13:35 |
| sean-k-mooney | morenod: jgilaber for nwo we woudl ned to modify the testign and job config | 13:35 |
| sean-k-mooney | to be able to supprot this | 13:35 |
| sean-k-mooney | so it feel like it more hten a bug | 13:35 |
| sean-k-mooney | but this is a limitation we do need to document upstream and downstream | 13:36 |
| jgilaber | agreed, we could keep it as a wishlist bug if we want to have some tracker until we work on it | 13:36 |
| sean-k-mooney | but we have many other undocumented limaitons so its not unqiue in that respect | 13:36 |
| sean-k-mooney | yes that sound good | 13:37 |
| sean-k-mooney | lets keep it wishlist for now | 13:37 |
| jgilaber | ok, so I'll do that after the meeting if there is no objection | 13:37 |
| jgilaber | last thing | 13:38 |
| jgilaber | #topic Volunteers to chair next week | 13:38 |
| jgilaber | any volunteer? | 13:38 |
| rlandy | I can't - have a meeting overlap - will check with dviroel if he can | 13:39 |
| jgilaber | I can do it next week again | 13:39 |
| jgilaber | ok, so any last minute topic before closing? | 13:39 |
| sean-k-mooney | am one but its not urgent | 13:40 |
| jgilaber | we have the time if you want to raise it | 13:40 |
| sean-k-mooney | i would like to add a minmial AGENT.md and weekly ubutnu-resolute job in the next few weeks | 13:40 |
| sean-k-mooney | i havnt really had time to follow up on either but directionally are folks ok with that | 13:41 |
| sean-k-mooney | we can obviously dicuss this more hwen more folks are around | 13:41 |
| sean-k-mooney | but its something i was thinking of doing before the end fo the month | 13:41 |
| jgilaber | did we discuss this for watcher in the PTG or was it a general discussion? | 13:42 |
| jgilaber | I'm +1 on the AGENT.md, the ubuntu job is to test the new version? | 13:42 |
| sean-k-mooney | we discuss the ai tooling a little but not in depth | 13:42 |
| sean-k-mooney | jgilaber: yep | 13:42 |
| sean-k-mooney | jgilaber: its to test 26.04 before it becomes the default next cycle | 13:43 |
| sean-k-mooney | so weekly is enough for that | 13:43 |
| sean-k-mooney | and python 3.14 as a sideffect | 13:43 |
| jgilaber | ack, +1 to that as well, good to catch issues early | 13:43 |
| jgilaber | do we test with 3.14 already? | 13:43 |
| sean-k-mooney | in unit tests | 13:44 |
| sean-k-mooney | i added 26.04 supprot to devstack about 2-3 weeks ago | 13:44 |
| jgilaber | ok, that is good to hear, thanks for doing this work | 13:44 |
| sean-k-mooney | so the basic job runs there but im hopign to add weekly/experimtal jobs to more repos | 13:44 |
| sean-k-mooney | https://review.opendev.org/c/openstack/nova/+/984540/5 | 13:45 |
| sean-k-mooney | that is what the nova minimal ai scafold looks like | 13:45 |
| sean-k-mooney | i and cinder https://review.opendev.org/c/openstack/cinder/+/994007 | 13:45 |
| sean-k-mooney | so im hoping to do the same form cybrog and watcher when i find the time | 13:46 |
| sean-k-mooney | the core idea is it stays very minimal wiht the actual guidence in the contibotr docs in human readable prose | 13:46 |
| sean-k-mooney | this is work that someone else could also do | 13:47 |
| jgilaber | from a quick glance looks good, I have a similar setup locally, would be good to standarize | 13:47 |
| sean-k-mooney | i just wanted to bring this up as a topic i would like to progress | 13:47 |
| sean-k-mooney | for now im still a littel reluctant to go mroe then that in the repo i.e. i dont think builidn skils in the repos is somethign we shoudl do yet | 13:49 |
| sean-k-mooney | but ok i think we can wrap up | 13:49 |
| jgilaber | sounds good, thanks! | 13:49 |
| jgilaber | if there is nothing else we cap wrap early and have 10 minutes back | 13:49 |
| jgilaber | thanks all! | 13:50 |
| jgilaber | #endmeeting | 13:50 |
| opendevmeet | Meeting ended Thu Jul 9 13:50:56 2026 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 13:50 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/watcher/2026/watcher.2026-07-09-13.00.html | 13:50 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/watcher/2026/watcher.2026-07-09-13.00.txt | 13:50 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/watcher/2026/watcher.2026-07-09-13.00.log.html | 13:50 |
| morenod | thanks jgilaber!! | 13:51 |
| melwitt | sean-k-mooney: ack on the nova stuff. I could see there being some "standard" things nova could do internally around cells since they can almost be assumed the hosts in a cell would be an their own aggregate (and additional smaller aggregates could be created around subsets of them by the cloud operator if they desired) | 16:20 |
| sean-k-mooney | ya i was thinkign of somehting along the lines of "auto_report_cell_aggreates=true" | 16:21 |
| sean-k-mooney | and if we have a standard way to generate the uuid or hostaggate name | 16:22 |
| sean-k-mooney | then watcher coudl discover either | 16:22 |
| sean-k-mooney | assuming we didnt just exposee the cell mapping via the api to services/admins | 16:22 |
| melwitt | yeah. that is also something to potentially consider | 16:23 |
| sean-k-mooney | we do already schduler on this internally https://github.com/openstack/nova/blob/master/nova/scheduler/weights/cross_cell.py | 16:23 |
| melwitt | I wonder why that didn't "save" the situation? is it not a default weigher or | 16:24 |
| sean-k-mooney | all weigher are enabled by default | 16:25 |
| sean-k-mooney | but the issue is that watcher seclects the host 90% of the time | 16:26 |
| sean-k-mooney | rather then allowign nova too | 16:26 |
| sean-k-mooney | and sicne watcher does not have this info it cant knwo if it going to succeed or even why it got a no valid host | 16:26 |
| melwitt | oh yeah duh lol. watcher not using the nova scheduler | 16:27 |
| sean-k-mooney | so the info is in the nova api db but there is no rest api for it, the condocor know implictly because it can look at its local cell db for host relathionp and nova scheuelr knows | 16:28 |
| melwitt | right | 16:28 |
| sean-k-mooney | but i dont think there exist a way today for anything esle to know | 16:28 |
| melwitt | yeah I was just being dumb | 16:28 |
| sean-k-mooney | no not at all you did what i do most of the time and assume nova is making the choice | 16:28 |
| sean-k-mooney | this is porably a good dicsusion to start on the mailing list before the next ptg | 16:30 |
| sean-k-mooney | like there a a number of ways to impove this but they all need nova changes | 16:30 |
| sean-k-mooney | and there are pros and cons to each of the approches | 16:30 |
| melwitt | yeah, always good to get it out there earlier so people can have a think on it | 16:30 |
| opendevreview | Merged openstack/watcher master: Generate sample policy file in local directory https://review.opendev.org/c/openstack/watcher/+/961486 | 17:16 |
| -Guest12850- NOTICE: The Gerrit service on review.opendev.org will be offline briefly one hour from now, at 21:00 UTC, while we rename a project. | 20:02 | |
| opendevreview | Merged openstack/watcher master: Remove redundant policy_file override https://review.opendev.org/c/openstack/watcher/+/994178 | 20:12 |
| -Guest12850- NOTICE: The Gerrit service on review.opendev.org is going offline momentarily at 21:00 UTC while we rename a project, but will return within a few minutes. | 21:00 | |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!