12:00:29 <marios> #startmeeting watcher 12:00:29 <opendevmeet> Meeting started Thu Mar 20 12:00:29 2025 UTC and is due to finish in 60 minutes. The chair is marios. Information about MeetBot at http://wiki.debian.org/MeetBot. 12:00:29 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 12:00:29 <opendevmeet> The meeting name has been set to 'watcher' 12:00:52 <marios> hello o/ as usual agenda items are there https://etherpad.opendev.org/p/openstack-watcher-irc-meeting 12:01:04 <marios> we'll get going in a moment while people join 12:01:11 <marios> who is around today? 12:01:16 <chandankumar> o/ 12:01:27 <amoralej> o/ 12:01:41 <rlandy> o/ 12:01:41 <marios> #info weekly meeting agenda at https://etherpad.opendev.org/p/openstack-watcher-irc-meeting 12:02:22 <marios> not a lot there so should be a brief one today. we'll take them as listed. feel free to add any topics that come to mind while we proceed. 12:02:56 <jgilaber_> o/ 12:03:04 <marios> any comment before we begin? 12:03:19 <mtembo> o/ 12:04:01 <marios> #topic (rlandy): Anything additional for PTG? Joined seesion with telemetry requested for Wednesday 12:04:28 <marios> rlandy: do you want to introduce this? or something further to add besides ^^ 12:04:28 <rlandy> thanks marios ... 12:04:41 <rlandy> so just what it says :) 12:05:02 <rlandy> #link https://etherpad.opendev.org/p/apr2025-ptg-watcher has our topics 12:05:24 <marios> thanks i was just looking for that to add a link 12:05:30 <rlandy> they all seem to have owners/chairs (thank you) 12:05:44 <rlandy> and then there are some requested joined sessions 12:05:48 <rlandy> one with nova 12:06:16 <rlandy> #link https://etherpad.opendev.org/p/nova-2025.2-ptg#L62 12:06:48 <rlandy> and one with horizon/telemetry (requested on wednesday - since that day was open for this group) 12:07:18 <rlandy> thanks for dviroel for organizing the watcher sessions 12:07:32 <rlandy> marios: that is all from my side - thank you 12:07:46 <marios> thank you rlandy 12:08:04 <marios> any other comments thoughts concerns or anything else on this topic from anyone? 12:09:43 <marios> ok moving on 12:09:48 <marios> #topic Bug Triage 12:09:59 <marios> #link https://bugs.launchpad.net/watcher/+bug/2103458 12:10:29 <marios> not sure who added this to the agenda there is no name. mtembo is that yours? 12:10:43 <marios> but is by amoralej 12:11:05 <marios> #info workload_stabilization strategy does not report standard_deviation properly https://bugs.launchpad.net/watcher/+bug/2103458 12:11:08 <amoralej> i created the ticket, right 12:11:14 <amoralej> i realized on that while testing 12:11:36 <amoralej> it's just related to the strategy reporting the efficiency 12:11:41 <amoralej> it sets values to 0.0 12:11:58 <amoralej> although internally seems to be calculating it fine 12:12:45 <marios> this sounds like an obvious yes valid/'triaged' bug... is it something you are working on currently amoralej ? 12:12:57 <amoralej> not really 12:13:00 <amoralej> just reported it 12:13:14 <amoralej> can be taken by anyone 12:13:42 <marios> so, triaged and high sound right? 12:13:48 <marios> remaining unassigned 12:14:22 <amoralej> could be high or medium, i'm not sure 12:14:39 <amoralej> and i didn't dig into the code itself yet 12:14:56 <rlandy> there are folks who are up for looking at the strategies in the upcoming weeks - so that can fall into the same bucket 12:15:21 <marios> any other thoughts or comment from anyone on this? 12:15:33 <marios> based on discussion thus far i'll set triaged and high. any objection? 12:16:12 <jgilaber_> +1 to that, if no one is working on it I'll probably take it early next week 12:16:13 <amoralej> wfm 12:17:21 <marios> ok moving on momentarily if there is nothing further. any other bugs we should look at (none in the etherpad) 12:17:35 <amoralej> i also created https://bugs.launchpad.net/watcher-tempest-plugin/+bug/2103640 12:17:54 <amoralej> i logged it as a RFE as it's coverage improvement on the tempest plugin 12:18:13 <marios> #info https://bugs.launchpad.net/watcher-tempest-plugin/+bug/2103640 [RFE] Watcher tempest plugin should provide a way to run tests with real load instead of fake metrics 12:19:07 <marios> as an rfe i am not sure we should/could triage this in the weekly irc meet 12:19:09 <chandankumar> I have moved the importance to wishlist 12:20:01 <marios> well i mean, its hard to argue with the bug subject... so we could set it to triaged. that doesn't speak to when/who would do that 12:20:17 <amoralej> yeah, it was just for awareness 12:20:48 <marios> ok and chandankumar also set triaged 12:21:08 <sean-k-mooney> technially there is already a way 12:21:12 <amoralej> wrt importance, for me it's higher that wishlist 12:21:22 <sean-k-mooney> you set the datasouce in the tempest plugin to none 12:21:33 <sean-k-mooney> but only a small subset of test will work 12:21:50 <sean-k-mooney> im not sure we should provide a generic way for all test to use real vs fake data 12:22:06 <sean-k-mooney> but we proably shoul add some specific test that dont inject data 12:22:20 <amoralej> so you mean, setting datasource to '' and doing the load "from outside" ? 12:22:22 <sean-k-mooney> but do use metrics just to validate that we can wrok without that 12:22:32 <sean-k-mooney> kind of 12:23:00 <sean-k-mooney> what im say is we do not have a way today to generate the load to trigger the rebalnace or whatever 12:23:09 <sean-k-mooney> but i dont think we should provide 2 ways of runing every test 12:23:10 <amoralej> so my proposal was to create a small set of different tests in which we don't create metrics but run load 12:23:18 <amoralej> i don't expect it to work for all scenarios 12:23:19 <sean-k-mooney> i.e with injected data and real data 12:23:28 <sean-k-mooney> ack 12:23:34 <sean-k-mooney> that alings with my thinking too 12:23:46 <sean-k-mooney> although small set could just be 1-2 12:24:10 <amoralej> for me the important is that we chose some strategy that fails if metrics are not found 12:24:26 <sean-k-mooney> yep 12:24:35 <amoralej> some metrics have fallback mechanism, i.e. getting memory size from nova if there is no memory metric 12:24:56 <amoralej> as what i know the workload_stabilization is good candidate 12:25:05 <sean-k-mooney> the problem is unless we have two jobs then we have to reviert disabling the cpu metric collection in node exporter 12:25:18 <amoralej> uh, good point 12:25:24 <sean-k-mooney> i dont think two jobs is a good use of ci resouces 12:25:33 <sean-k-mooney> so i woudl prefer to not do that 12:25:51 <sean-k-mooney> as far as i can tell we were not getting test failure when the cpu metric was collected 12:25:59 <sean-k-mooney> so i dont think reverting that will be a problem 12:26:38 <amoralej> is it possible to do two tempest run in a single job? 12:26:55 <sean-k-mooney> technially yes but its non trivial 12:27:16 <sean-k-mooney> doing service reconfiguration in a job is possibel but not advised 12:27:26 <amoralej> disable cpu metric colletion -> 1st tempest run with only fake metrics -> enable cpu metrics -> 2nd run with only real data one 12:27:36 <sean-k-mooney> it would also be a whitebox test at that point rather then blackbox 12:28:31 <sean-k-mooney> i think this type of testing is a pretty low priority 12:29:02 <sean-k-mooney> im not saying we should not do it just we should not take on a lot of technial debt to do it 12:29:12 <sean-k-mooney> so if we want to do two tempest runs we can 12:29:38 <amoralej> i agree that we should implement it in a proper way 12:29:39 <sean-k-mooney> with serivce reconfituation in between but normally that type of testing is done as a post playbook 12:30:23 <sean-k-mooney> so we would do a normal tempest run with fake data 12:30:30 <amoralej> one option may be to add the feature in the tempest plugin even if we don't use those jobs by default, and make specific job runs periodically or only in experimental pipeline 12:31:00 <sean-k-mooney> ya a weekly perodic would be ok 12:31:07 <sean-k-mooney> we could review its status in this meeting 12:31:10 <amoralej> +1 12:31:32 <sean-k-mooney> that pretty maintainable and low impact 12:31:56 <sean-k-mooney> and ya we can add it to experimental too to be able to trigger on specifc patches as needed 12:32:49 <marios> amoralej: it would be nice if you could capture some of these points into the bug as pointers for when (whoever) picks it up? 12:32:55 <amoralej> yes, i will 12:33:00 <amoralej> thanks 12:33:53 <marios> anything further for now on this one ? 12:34:35 <marios> are there any other bugs someone wanted to hilight today? 12:35:35 <marios> ok moving on momentarily unless there is anything else here 12:36:23 <chandankumar> I need reviews on https://review.opendev.org/q/topic:%22bug/2100741%22 and https://review.opendev.org/q/project:openstack/devstack-plugin-prometheus+status:open 12:37:00 <marios> ok chandankumar lets hilight those on the open floor in a moment ? 12:37:09 <chandankumar> marios: sure 12:37:13 <marios> #topic Volunteer to chair the next meeting 12:37:44 <marios> #info mtembo to chair irc meeting on Thursday 27th March 1200 UTC 12:37:46 <marios> thank you mtembo 12:37:47 <mtembo> I can take the next one 12:38:07 <marios> and that is the end of the agenda for today, so lets do open floor 12:38:11 <marios> #topic open floor 12:38:13 <marios> thank you chandankumar 12:38:30 <marios> #info review requests https://review.opendev.org/q/project:openstack/devstack-plugin-prometheus+status:open https://review.opendev.org/q/topic:%22bug/2100741%22 12:38:46 <marios> did you want to add some more info or anything else here chandankumar ^ 12:39:06 <chandankumar> sure 12:39:31 <chandankumar> https://review.opendev.org/q/topic:%22bug/2100741%22 - contains all the reviews related to movement of functional tests to python-watcherclient 12:40:21 <chandankumar> Below is the review order 12:40:32 <chandankumar> [Replace watcherclient functional job with python-watcherclient-functional](https://review.opendev.org/c/openstack/watcher/+/943136) 12:40:45 <chandankumar> [Add deprecation warning about client_functional tests](https://review.opendev.org/c/openstack/watcher-tempest-plugin/+/943133) 12:41:08 <chandankumar> sean-k-mooney: ^^ I might need help to here on how to highlight deprecation message when we run tests via tempest 12:41:25 <chandankumar> and last one on this [Drop watcherclient-tempest-functional job](https://review.opendev.org/c/openstack/watcher-tempest-plugin/+/945009) 12:42:20 <chandankumar> On prometheus devstack plugin side 12:42:36 <chandankumar> [Use get_extra_file for downloading node_exporter/prometheus tarballs](https://review.opendev.org/c/openstack/devstack-plugin-prometheus/+/944622) 12:42:57 <chandankumar> [Add devstack-prometheus job to validate prometheus plugin](https://review.opendev.org/c/openstack/devstack-plugin-prometheus/+/944095) - to validate the plugin itself 12:43:39 <chandankumar> https://review.opendev.org/c/openstack/devstack-plugin-prometheus/+/944829 - Install podman prometheus exporter via devstack plugin 12:44:03 <chandankumar> that's it from my side, feel free to take a look! 12:44:11 <chandankumar> thank you marios! 12:44:30 <marios> thank you chandankumar holding to see if there are any comments or questions on any of that for now 12:45:48 <marios> any other items anyone wants to raise today? 12:46:29 <marios> i will close out the meeting in a moment if there is nothing further 12:46:29 <rlandy> do we want to skip the week of PTG? 12:47:03 <marios> good point. we could, though we also don't have sessions scheduled for that week. though given the ptg, indeed it is redundant probably? not sure 12:47:39 <marios> sorry, i meant we don't have sessions scheduled for that *day* 12:48:13 <marios> no strong opinions on this one rlandy perhaps revisit closer to (like next week irc meeting) 12:48:23 <marios> if there are no objections then probably canceling is fine 12:48:25 <rlandy> ack 12:49:00 <marios> any other topics today before we close out? 12:50:28 <opendevreview> Merged openstack/watcher master: Query by fqdn_label instead of instance for host metrics https://review.opendev.org/c/openstack/watcher/+/944795 12:51:01 <marios> ok then, thank you all for attending today o/ 12:51:04 <marios> #endmeeting