Thursday, 2026-06-04

« Wednesday, 2026-06-03 Index
jgilaber#startmeeting watcher12:01
opendevmeetMeeting started Thu Jun  4 12:01:36 2026 UTC and is due to finish in 60 minutes.  The chair is jgilaber. Information about MeetBot at http://wiki.debian.org/MeetBot.12:01
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.12:01
opendevmeetThe meeting name has been set to 'watcher'12:01
jgilaberHi all! Sorry I'm late12:01
jgilaberwho is around today!12:02
jgilaber?12:02
morenodo/12:02
jgilaberfeel free to add any topics to the agenda https://etherpad.opendev.org/p/openstack-watcher-irc-meeting#L3112:02
jgilaberlooks like we have a few topics lined up for today!12:02
jgilabercourtesy ping: dviroel amoralej sean-k-mooney chandankumar rlandy12:02
chandankumaro/12:03
jgilaberok let's start with the first topic12:05
jgilaber#topic Add spec for Preemptible Instances feature12:05
sean-k-mooneyo/12:05
jgilaberfrom winiciusallan[m] 12:05
jgilaberwho looks like could not make it today12:06
jgilaber#link https://review.opendev.org/c/openstack/watcher-specs/+/98717112:06
sean-k-mooneywe need to reveiw there spec12:06
jgilaberyes, I have not looked at it yet12:06
jgilaberlooks like Doug did add some comments12:06
sean-k-mooneyit looks liek tehy have startd filling it out more then the last time i looked12:06
sean-k-mooneyits on my list ill see if i can make tiem for it this week12:07
jgilaberI'm not sure if the points on the agenda are questions to discuss12:08
jgilaberor just an update12:08
sean-k-mooneyit could he either but if they cant attend lets loop back next week12:09
jgilaberI'll try to make some time to review it as well12:09
jgilaberyep, I'll leave a note12:09
jgilaberlet's move to the second topic12:09
jgilaber#topic Updates on playwright patches (just announcement)12:09
jgilaberchandankumar, go ahead12:09
chandankumarfinally https://review.opendev.org/c/openstack/requirements/+/983421: Add playwright requirement patch is merged now12:10
chandankumarhttps://specs.openstack.org/openstack/watcher-specs/specs/2026.2/approved/improve-watcher-dashboard-ui-testing.html is already merged last week12:10
chandankumarI have also updated remaining patches based on requirement patch     https://review.opendev.org/c/openstack/watcher-dashboard/+/97035312:10
chandankumar    https://review.opendev.org/c/openstack/watcher-dashboard/+/97659412:10
chandankumarthat's it from my side in this12:10
sean-k-mooneywe bumpt to the latest playright correct12:11
sean-k-mooneywas there any fallout form that12:11
chandankumarnope12:11
sean-k-mooneycool12:11
sean-k-mooneythat what i was hopeing12:11
chandankumarsean-k-mooney: when you get time , can you update your watcher-dahsboard patches12:11
chandankumarone or tow of them have some comments12:12
sean-k-mooneyyes i just rememeber i need to update https://review.opendev.org/c/openstack/watcher-dashboard/+/978304/512:12
sean-k-mooneylet me see if i can do that today12:12
chandankumarthank you!12:12
sean-k-mooneyim condierign defering or abandoning some of the service layer refactor patches12:12
sean-k-mooneyi still thnk that woudl be good to do12:12
sean-k-mooneyim jsut not sure ill have time to get all of those into a state where im oke with merging them12:13
sean-k-mooneyso ill keep those out of the critail path12:13
sean-k-mooneybut on a releated note12:13
sean-k-mooneywe merged the nova version of https://review.opendev.org/c/openstack/watcher-dashboard/+/983336/212:13
sean-k-mooneyi will update that based on how we proceeded with this in nova12:14
sean-k-mooneyand pull them onto master instead of that series12:14
sean-k-mooneyso ill see if i can do that before the next meeting and we can dicuss12:14
chandankumar+1 to keeping it seperate from the series12:14
jgilaberthanks for the update and your work on this chandankumar 12:16
jgilaberany other comment question on this topic?12:16
chandankumarnope, we can move to the next topic12:18
jgilaberack12:18
jgilabernext two are from amoralej 12:18
jgilaber#topic Remove nova api calls from zone_migration12:18
amoralejyep12:18
jgilaber#link https://launchpad.net/bugs/211995712:18
jgilaber#link https://review.opendev.org/c/openstack/watcher/+/99129112:18
jgilaber#link https://review.opendev.org/c/openstack/watcher/+/98828412:19
amoralejwe discussed about this one last week, I decided to split in two reviews, one for the change in the model, where i included hypervisor_hostname host and created12:19
amoralejand second one with the actual change in zone_migration12:19
amoraleji find it better to review, but i'm open to squash them if you prefer12:19
jgilaber+1 to having two patches12:19
jgilaberit will be easier to review 12:20
amoralejalso, i want to mention that nova notifications does not provide host and hypervisor_hostname but host and "node", from code, i'd say node in notifications is the same as hypervisor_hostname12:20
amoralejbut i hope sean-k-mooney can confirm it :)12:20
sean-k-mooneysorry distracted12:21
sean-k-mooneyyes i belive it is12:21
sean-k-mooneyi can tripel check tghat when doing the code review12:21
amoralej https://github.com/openstack/nova/blob/488d490b5d6ca39966b633d19d3eb7b23b9cb13f/nova/compute/manager.py#L10386 https://github.com/openstack/nova/blob/488d490b5d6ca39966b633d19d3eb7b23b9cb13f/nova/compute/manager.py#L1040312:21
amoralejthat's what i found12:21
amoralejso, i think that was it from the zone_migration topic i think, unless there are more questions12:22
sean-k-mooney not form me anyway. if i have any ill ask in the review12:22
amoralejthanks12:23
jgilaberthanks amoralej, moving on12:23
jgilaber#topic investigating in the vm_workload_consolidation 12:23
jgilabernot sure it's the best summary :)12:23
amoraleji started checking12:23
amoralejyes, good enough :) 12:23
amoraleja scalability bug on that strategy, i think i found where the problem is and will send a fix for it12:24
amoralejbut while investigating,  i found that it doesn't check allocations of vms when creating the migrations so it can easily try to migrate more vms to a node that the ones that nova will accept based on resources. 12:24
amoralejonly check usage metrics, not allocations12:24
amoralejwhich is wrong imo12:24
amoralejso, the question is, is that a bug?12:25
jgilaberchecking allocations means talking to placement?12:25
amoralejno need, it can check in the model12:25
amoralejbut based on the info it has from placement, yes12:25
sean-k-mooneyoh12:26
jgilaberoh so it's information we have?12:26
amoralejthe actual memory, cpus, disks12:26
sean-k-mooneyits checkign usage not requested capastiy12:26
amoralejyes, yes, other strategies do it12:26
sean-k-mooneyya the reviers is valid 12:26
jgilaberthen I would say it's a bug in the strategy12:26
sean-k-mooneyi.e. checkign the reuqest but not the usage but jsut using usage is incorrect12:26
amoralejexactly12:26
amoraleji see it more as a bug, tbh12:27
sean-k-mooneyya i think we can codnier that a seperate bug form the first12:27
amoralejbut could be also understood as a missing feature, dunno12:27
sean-k-mooneyliekly with its own commit12:27
jgilaber+112:27
amoraleji can create a new bug and fix in different bug as the scalability issue12:27
sean-k-mooneywell the diffent eb twen the vm adn node version 12:27
sean-k-mooneywas the vm version was ment to check both12:28
amoralejthat was my understanding too12:28
sean-k-mooneywehere as the node version only checked the requested amoutn and didnt use the metrics right?12:28
amoralejnode_consolidation is known to only check allocations12:28
amoralejbut vm one should check both not only usage12:28
amoralejyes, you are right, that was by design12:28
sean-k-mooneyya so i think we can make athat change and make sure we dupdate the docs and call it out in a relase note12:28
sean-k-mooneywe can decied seperatly if we only do that in master or backport it12:29
amoralejack12:29
amoralejI will create the bug report, thanks12:30
jgilaberthanks amoralej, anything else on this topic?12:31
amoralejno, that was it12:31
jgilaberack let's move to reviews12:31
jgilaber#topic Reviews12:31
jgilaber#link https://review.opendev.org/c/openstack/watcher-specs/+/98717112:31
jgilaberthat's the preemptible instances spec we discussed earlier12:31
jgilaberdoes anyone have any other patch to call out?12:31
jgilaberif not we have a few bugs to triage12:32
jgilaber#topic Bugs12:33
jgilaberfirst one 12:33
jgilaber#link https://bugs.launchpad.net/watcher/+bug/215262312:33
jgilaberthis one comes from a ci job, looks valid but low priority12:33
jgilaberok with setting the bug to confirmed, low?12:34
sean-k-mooneytriaged rather then confirmed if we have set importance12:35
sean-k-mooneyand yes i agree it logs valid to me12:35
amoralejyep12:35
sean-k-mooneyif the data souce is not enabeld its code shoudl not be executed12:35
jgilaberright12:35
sean-k-mooneyi also think low is fien since it is just noice in the log12:36
sean-k-mooneybut i personally hate "standard errors"12:36
jgilaberack, I've also added the "datasources" tag12:37
sean-k-mooneythat we are jsut expected to ignore so it would be nice to clean this up12:37
amoralejagree, an ERROR should mean an error :) 12:37
sean-k-mooney+112:37
jgilaberyep, agreed, high signal to noise ratio in the logs is something important12:37
jgilaberlet's move to the next three which have all been opened recently by morenod 12:38
jgilaber#link https://bugs.launchpad.net/watcher/+bug/215480512:38
jgilaber#link https://bugs.launchpad.net/watcher/+bug/215480612:38
morenodyes12:38
jgilaber#link https://bugs.launchpad.net/watcher/+bug/215480712:38
morenodas we agreed last week, this is about the audit scope, one per strategy12:38
amoralejthe actual issue is only with one particular scope parameter, right? instance exclussion, iiuc12:39
jgilaberI think we should add a bit more information on those reports12:40
morenodyes, this is only affecting instance scope, host scope works fine on all strategies12:40
jgilaberand at the very least a link to the previous bug which had more details12:40
amoralejyes, an example of the scope config and the expected vs found behaviour would be great12:40
sean-k-mooneyso all 3 of those im inclided to marek incomplete12:41
sean-k-mooneyi know morenod  filed them12:41
sean-k-mooneyhowever they bugs are not self contaiend12:41
sean-k-mooneythey do not expalin what is broken, howe its broken or how to repoduce/expected outcome12:42
sean-k-mooneyso until they are actully updated with that info they shoudlnot be condier valid or triaged 12:42
morenodok, I will add more information12:42
jgilaber+1 we can revisit them next week12:42
sean-k-mooneycool once that is adde i woudl say medium or low12:42
sean-k-mooneyjust because they are so latent12:42
sean-k-mooneywe shoudl totally fix them12:43
sean-k-mooneybut im not sure operator woudl rush to upgadge just for the fix so its not high or critial12:43
jgilabermedium sounds right to me12:44
sean-k-mooneymorenod: are you plannign to work on the fix for tehse or tests or just flaging thses based on your local testing12:44
amoralej+112:44
morenodsean-k-mooney, I have some tests prepared, that is how I discover the bugs, but I wont push them until the fix will be merged12:45
sean-k-mooneymorenod: ok you coudl jsut push the tempest tests12:45
sean-k-mooneyits oke if they fail we can use them to test the fixes before they merge12:45
sean-k-mooneyeither is fine12:45
sean-k-mooneybut know its oke to push tempetss test that demonstrate bugs12:46
sean-k-mooneywe will just keep the review open until the bug is fixed12:46
morenodok, then I will, so you can use to test the patch against it12:46
jgilaberanything else on this topic?12:47
jgilaberor any last minute topic to raise? we've reached the end of the agenda for today, since dviroel already volunteered to chair next meeting12:48
sean-k-mooneynote really12:48
sean-k-mooneydoug had some patches12:48
sean-k-mooneyform last week reslated to the lookcing12:49
sean-k-mooneythat i thnk we can likely procveed with12:49
sean-k-mooneyi just have not had time to look at them12:49
sean-k-mooneyhas anyone else12:49
jgilaberwe merged the first two I think12:49
sean-k-mooneyamoralej: i think you started12:49
sean-k-mooneyah cool12:49
sean-k-mooneyi skimmed them and didn see anyting obvious wrong with them12:49
sean-k-mooneybut i didn have time to dig in12:49
amoralejyes, we got the first fix merged actually i think12:49
sean-k-mooneythe first 2 were for a prexisitng bug right12:49
jgilaberthere are two more ready for review, but I'm not sure I'll have the time to get to them this week12:50
amoralejand reviewed the one about rlock12:50
sean-k-mooneyand the secodn 2 are the change to per instnace locking12:50
sean-k-mooneyrather then a gloabl lock?12:50
jgilaberyes, we merged the fix for a race condition between the periodic collector and notification updates12:50
sean-k-mooneycool12:50
sean-k-mooneysoudn like good progress12:51
sean-k-mooneywe can likely wrap up so12:51
jgilaberthanks all for the discussions, that's all for today12:51
jgilaber#endmeeting12:51
opendevmeetMeeting ended Thu Jun  4 12:51:39 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)12:51
opendevmeetMinutes:        https://meetings.opendev.org/meetings/watcher/2026/watcher.2026-06-04-12.01.html12:51
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/watcher/2026/watcher.2026-06-04-12.01.txt12:51
opendevmeetLog:            https://meetings.opendev.org/meetings/watcher/2026/watcher.2026-06-04-12.01.log.html12:51
sean-k-mooneyjgilaber: on a diffent topic12:53
sean-k-mooneydid you see my comment on the mdev pathces12:53
sean-k-mooneyabout not overriedign the base flavor and usign the whitebox gpu tests instead12:54
jgilaberI saw the notification this morning but have not really looked into it yet12:54
jgilaberI need to go and check which tests are actually using that12:54
sean-k-mooneyack the third WIP patch is replacign the default flavor use by tempest12:54
sean-k-mooneybut that is causing other test to fail12:54
amoralejthanks jgilaber!12:55
sean-k-mooneyso all test that done use there own flavor were beign made request a vgpu12:55
sean-k-mooneywe can chat about that more on the nova channel but i just wanted to call out that that wasn only a temp hack12:55
sean-k-mooneythe inten was not to replace the flavor12:55
sean-k-mooneyit was to eventully have whitebox or new tempest test for the mdev work12:56
sean-k-mooneyso i didnt want you to spend too much time debuging the test failures12:56
jgilaberoh great thanks for the heads up, I saw the failures yesterday but did not dig into them12:58
« Wednesday, 2026-06-03 Index

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!