12:01:04 <dviroel> #startmeeting watcher 12:01:04 <opendevmeet> Meeting started Thu May 15 12:01:04 2025 UTC and is due to finish in 60 minutes. The chair is dviroel. Information about MeetBot at http://wiki.debian.org/MeetBot. 12:01:04 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 12:01:04 <opendevmeet> The meeting name has been set to 'watcher' 12:01:28 <dviroel> hi all, who's around? 12:01:40 <jgilaber> o/ 12:02:08 <chandankumar> o/ 12:02:20 <rlandy> o/ 12:02:22 <mtembo> o/ 12:02:29 <dviroel> courtesy ping: amoralej 12:02:49 <dviroel> i proposed the courtesy ping list in one of the past meetings 12:03:31 <dviroel> the idea is that, if you put your nick there, you are going to receive a ping at the beggining of the meetings ... unless you are already here, i can skip your nick :) 12:03:55 <dviroel> let's start with today's meeting agenda 12:04:09 <dviroel> #link https://etherpad.opendev.org/p/openstack-watcher-irc-meeting#L36 (Meeting agenda) 12:04:19 <dviroel> feel free to add your own topics to the agenda 12:04:42 <dviroel> there is a topic to place your changes that requires attention from reviewers 12:05:07 <dviroel> as also a bug triage topic, if you want to bring any for discussion 12:05:34 <dviroel> starting with the first one 12:05:44 <dviroel> #topic Eventlet Removal 12:06:09 <dviroel> the plan here is to circle this topic every week 12:06:30 <dviroel> to follow up progress for this goal, get some feedback and call for actions 12:07:02 <dviroel> this is the etherpad that we capture progress and discussions around eventlet removal in watcher 12:07:05 <dviroel> #link https://etherpad.opendev.org/p/watcher-eventlet-removal 12:07:49 <dviroel> at the top, there are some changes that were already proposed, but this is just a start 12:08:00 <dviroel> to prepare for next changes 12:08:10 <dviroel> so i am calling for review of these ones 12:08:20 <dviroel> #link https://review.opendev.org/c/openstack/watcher/+/949641 (Move eventlet command scripts to a different dir) 12:08:25 <dviroel> #link https://review.opendev.org/c/openstack/watcher/+/949658 (Remove deprecated executor in message handling servers) 12:09:02 <dviroel> we can capture discussions in the etherpad or in the code review 12:09:23 <dviroel> or even here in the irc channel 12:09:38 <dviroel> ok, next thing in this topic is 12:10:14 <dviroel> maas (metal as a service) support 12:10:40 <dviroel> i want to highlight that the maas integration which will be affect by the eventlet removal 12:11:05 <dviroel> the asyncio compatible call in the code will require change and testing 12:11:10 <dviroel> see 12:11:11 <dviroel> #link https://opendev.org/openstack/watcher/src/commit/1b12e808827ac64d3972443927fc2d070c227023/watcher/common/utils.py#L173 12:11:47 <dviroel> but the point is that we don't have the knowledge and cycle to work on this feature 12:11:54 <jgilaber> do we have any testing for it currently= 12:11:56 <jgilaber> ? 12:11:58 <dviroel> not even how to test it 12:12:05 <dviroel> jgilaber: no 12:12:26 <dviroel> we miss a CI testing and more documentation too 12:12:44 <dviroel> while chatting with sean-k-mooney these days 12:13:05 <jgilaber> ack, should we like we did for the monasca and grafana datasources? 12:13:24 <jgilaber> mark as deprecated and see if someone is interested in it? 12:13:26 <dviroel> we discussed about marking it as an experimental feature during this cycle 12:13:30 <dviroel> and call for maintainers 12:13:46 <jgilaber> that sounds good to me 12:13:50 <sean-k-mooney> o/ 12:13:51 <dviroel> in case we don't find any, we could mark it as deprecated in the next cycle 12:14:00 <dviroel> and for removal in the following one 12:14:14 * dviroel sean-k-mooney o/ - talking about maas support 12:14:24 <sean-k-mooney> ack 12:14:52 <sean-k-mooney> this kind of ties back into one fo the ptg topics 12:15:04 <sean-k-mooney> to add a supprot level matrics to our docs 12:15:33 <sean-k-mooney> i think the maas supprot is perhaps the first but not nessisarly the last one to be consierd experimental 12:15:44 <dviroel> yes, we miss that in our docs for sure 12:16:14 <sean-k-mooney> we also do not have testing of the ironic integration but we may be able to fill that gap more easilly since there are ironci devstack jobs so we can likely borrow some config form them 12:16:37 <sean-k-mooney> for nwo it would also consider markign ironic as expermintal applying the same logic but that can be done sepreatly 12:17:10 <sean-k-mooney> i dont belive there is a maas devstack plugin or a refence job we can pull from to close the maas gap ourselves 12:17:23 <dviroel> agree 12:18:20 <dviroel> so the proposal is to call for maintainers of the maas in the ML 12:18:30 <dviroel> since it will be affected by the eventlet removal 12:18:50 <dviroel> and we lack testing or knowledge on this to maintain it 12:18:55 <sean-k-mooney> well that adn the fact is has no docs or integration testing 12:19:12 <sean-k-mooney> so we dont really knwo how to deploy it even if there was a bug with it to fix it 12:19:20 <dviroel> exactly 12:19:28 <sean-k-mooney> i am wawre that there was some ubuntu charm installer supprot 12:19:36 <sean-k-mooney> however they have moved on to sunbeam 12:19:43 <sean-k-mooney> and i dont belive its supproted there 12:20:01 <sean-k-mooney> so im not sure if any activly devloped installer currently suprpot deploying in this mode 12:20:23 <sean-k-mooney> anyway i agree with your propsoal 12:20:56 <dviroel> #action dviroel to send an email to mailing list to call for maintainers of the maas support 12:22:20 <dviroel> ok, anything else in this topic? 12:23:24 <dviroel> #topic call for reviews 12:23:37 <dviroel> 1. Set keystone_client default interface to internal 12:23:47 <dviroel> #link https://review.opendev.org/c/openstack/watcher/+/948322 12:23:56 <sean-k-mooney> i think we ended up defaulting to public? 12:24:12 <jgilaber> yes, I changed that in the last patchset 12:24:27 <sean-k-mooney> ya so long term we shoudl replace our current config option 12:24:38 <sean-k-mooney> with valid_interface form oslo 12:24:48 <sean-k-mooney> valid_interfaces is a preferncally orderd list 12:25:04 <sean-k-mooney> so we can eventually default to internal,public 12:25:21 <sean-k-mooney> and it will use the internal endpoint if its aviabel but fall back to public which should always exist 12:25:39 <sean-k-mooney> i think that woudl be a ncie wishlist feature for all the clients 12:25:50 <dviroel> yeah, that's a got rfe 12:25:52 <sean-k-mooney> but we may want to consider that as part of moving to the openstack sdk 12:25:58 <dviroel> s/got/good 12:26:26 <sean-k-mooney> i have approved https://review.opendev.org/c/openstack/watcher/+/948322 12:26:38 <dviroel> thanks sean-k-mooney 12:26:43 <dviroel> moving to the next one 12:26:49 <jgilaber> thanks! 12:26:55 <dviroel> 2. Migrate value column of efficacy indicator on load 12:27:04 <dviroel> #link https://review.opendev.org/c/openstack/watcher/+/949640 12:27:16 <jgilaber> This is a follow up from https://review.opendev.org/c/openstack/watcher/+/945199 12:27:28 <dviroel> it is a new one (at least for me) 12:27:40 <dviroel> ack 12:27:42 <jgilaber> in that patch we modified the db schema for efficacy indicators 12:28:08 <jgilaber> and we could read the old column as fallback, but we're not updating the db back to use the old column 12:28:14 <jgilaber> this patch implements that 12:28:34 <sean-k-mooney> yes its a followup ot "heal the db over time" it will eventually allow us to drop the old column 12:29:07 <dviroel> ack, going to take a look afterwards, tks jgilaber 12:29:12 <jgilaber> thanks! 12:29:24 <sean-k-mooney> the alternitive would have been a "blocker" onlien migration to be ran prior to upgradign to the release where the old colmn is remvoed 12:29:37 <sean-k-mooney> we likely wont do the removal until 2026.2 or later 12:29:57 <sean-k-mooney> but the sooner we have the migration code the lower the impact 12:30:02 <sean-k-mooney> i have added ot my review list 12:30:16 <dviroel> next, also from jgilaber 12:30:20 <dviroel> 3. Add test for missing destination in zone migration 12:30:29 <dviroel> #link https://review.opendev.org/c/openstack/watcher/+/948865 12:30:51 <dviroel> and 4. Handle missing dst parameters in zone_migration 12:30:54 <dviroel> #link https://review.opendev.org/c/openstack/watcher/+/948866 12:30:59 <dviroel> which are related 12:31:33 <sean-k-mooney> right the first one is the repoducer for https://bugs.launchpad.net/watcher/+bug/2108988 it should have Related-Bug: #2108988 12:31:35 <jgilaber> they are a reproducer and fix for https://bugs.launchpad.net/watcher/+bug/2108988, which we discussed last week I think 12:31:36 <sean-k-mooney> in the commit 12:32:25 <jgilaber> yes, in short, the zone_migration strategy is not handling correctly the cases where the destination is not set by the user 12:32:26 <sean-k-mooney> quick glance both look ok 12:32:37 <dviroel> is it a fix that we plan to backport? 12:32:40 <jgilaber> for both instance and storage migration 12:32:56 <jgilaber> I think this one we could backport 12:33:01 <sean-k-mooney> that a good question 12:33:16 <sean-k-mooney> i think it can be backported potentially 12:33:17 <jgilaber> it's only changing the strategy code 12:33:32 <jgilaber> the parameters are the same 12:34:04 <dviroel> i miss a release note for that, but I will review the patch afterwards 12:34:21 <jgilaber> yes, I just realized that, I'll add it after the meeting 12:34:21 <sean-k-mooney> we shoudl have a release note in the second patch yes 12:34:38 <sean-k-mooney> looking at the input parmater schema 12:34:52 <sean-k-mooney> the dst_node was intended to be optional 12:34:59 <sean-k-mooney> so this is really just fixing that 12:35:12 <sean-k-mooney> and reconsiling the implemetion with the schema 12:36:09 <sean-k-mooney> although 12:36:23 <sean-k-mooney> dest_pool is also ment to be optional but we are treating that as an error 12:36:57 <sean-k-mooney> we can continue this in the gerrit review but if we are going to require the dst_pool we may want to udpate the schema but we might decied to do that in a differnt 12:36:59 <sean-k-mooney> patch 12:37:33 <dviroel> sean-k-mooney: ack, thanks for raising this, we can continue in the review yes 12:37:43 <sean-k-mooney> im inclined to split the fix into two patches one for the compute part and one for the storage part 12:37:44 <jgilaber> that one is more complicated and I'm not sure that what I implemented is the best approach, but yes, let's discuss it in the patch 12:38:06 <jgilaber> that would be fair 12:38:11 <dviroel> tks jgilaber for working on these 12:38:28 <jgilaber> np! thanks for the discussion 12:38:45 <dviroel> we have 2 more in the list 12:39:04 <dviroel> 5. add tests for host_maintenance strategy with backup node 12:39:07 <dviroel> #link https://review.opendev.org/c/openstack/watcher-tempest-plugin/+/947944 12:39:32 <dviroel> from chandankumar 12:39:57 <dviroel> ok, adding tests, i will review afterwards.. 12:40:09 <chandankumar> this cr adds tempest tests for host maintenance strategy with backnode. we have recently fixed the host maintenance strategy with backup node, it will extend the coverage. 12:40:46 <dviroel> chandankumar: ack, tks 12:40:48 <dviroel> and 12:40:51 <dviroel> 6. use get_host_for_server to get server host 12:40:53 <dviroel> #link https://review.opendev.org/c/openstack/watcher-tempest-plugin/+/949557 12:41:09 <chandankumar> we had a nice discussion on this cr related to getting host name of compute server. the second cr is a follow of that 12:41:50 <dviroel> right, i remember that 12:41:51 <sean-k-mooney> i think it would be ok to proceed with this 12:42:09 <chandankumar> tempest scenario manager provides get_host_for_server method https://github.com/openstack/tempest/blob/master/tempest/scenario/manager.py#L1347 that I will using it in all scenario 12:42:19 <sean-k-mooney> although i see althoguh we apprently need a promoation for the operator job to pass 12:42:37 <sean-k-mooney> based on https://review.opendev.org/c/openstack/watcher-tempest-plugin/+/947944/comments/cf2d8a7a_45ef8a99 12:42:51 <chandankumar> yes we need a promotion for operator job to pass which is blocked on different reason 12:43:25 <sean-k-mooney> so we may want ot alter how this works by default 12:43:42 <sean-k-mooney> we might want the master job to alwasy use master watcher form opendev 12:44:03 <sean-k-mooney> and have the content provider build those contianer and avoid usign the rdo content 12:44:19 <sean-k-mooney> we can discuss that seperatly 12:44:28 <sean-k-mooney> but that would eb closer to how we test with devstack 12:44:45 <chandankumar> sean-k-mooney: that's a nice suggestion to build watcher from source always. I will add it to my backlog. 12:45:18 <dviroel> great, tks chandankumar - lets continue this discussion afterwards then 12:45:43 <dviroel> thats all for review in the etherpad for today 12:45:52 <opendevreview> Merged openstack/watcher master: Set keystone_client default interface to public https://review.opendev.org/c/openstack/watcher/+/948322 12:45:54 <dviroel> #topic bug triage 12:46:08 <dviroel> https://bugs.launchpad.net/watcher/+bug/2108994 (apscheduler retrieving decimal.Decimal via sqlalchemy) 12:46:41 <dviroel> #undo 12:46:41 <opendevmeet> Removing item from minutes: #link https://bugs.launchpad.net/watcher/+bug/2108994 12:46:46 <dviroel> #link https://bugs.launchpad.net/watcher/+bug/2108994 (apscheduler retrieving decimal.Decimal via sqlalchemy) 12:47:24 <dviroel> this is a new one 12:49:39 <sean-k-mooney> hum 12:50:22 <sean-k-mooney> this feels like we are again not usign the correct types 12:51:23 <sean-k-mooney> so 12:51:33 <sean-k-mooney> its reported again 2023.2 12:51:41 <sean-k-mooney> against 12:51:50 <sean-k-mooney> which is eol 12:52:08 <sean-k-mooney> and its missing some of the squlight backports i thik 12:52:23 <dviroel> hum good point 12:52:49 <sean-k-mooney> if this can be replicated on master we coudl fix it but if not then i think we would close eol 12:53:04 <dviroel> right 12:53:21 <sean-k-mooney> i think this is yet another place where we shoudl not be using decimal 12:53:47 <sean-k-mooney> i wonder if the colume is declard as numeric, int float or a timestampe of some type 12:54:29 <jgilaber> I looked in a deployment I have active and the next_run_time column is created as a double 12:54:30 <dviroel> jgilaber: hey, if you have some time, can you take a quick look in this bug? 12:54:32 <sean-k-mooney> if we are usign decimal anywhere in watcher that is also proably a code smell for a potential bug 12:55:07 <jgilaber> I assume that is where the problem is coming from the next_run_time column in the apscheduler_jobs table 12:55:48 <jgilaber> I can try to look into it a bit later today 12:56:08 <dviroel> ack, tks jgilaber - please add your findings to the bug 12:57:00 <dviroel> i will add a comment on the bug to reference this meeting logs 12:57:07 <dviroel> lets cover one more 12:57:19 <dviroel> #link https://bugs.launchpad.net/watcher/+bug/2110895 - Drop post/patch/delete undocumented rest api call from watcher action 12:57:35 <dviroel> chandankumar: o/ 12:57:41 <chandankumar> I was reviewing watcher apirefs doc and here is my observations: https://etherpad.opendev.org/p/watcher-apiref-audit based on comparison with watcher rest api, python-watcherclient and api refs docs. 12:58:16 <dviroel> on nice 12:59:13 <sean-k-mooney> yes so whewn i did a quick code review i did not check tf this also apples to any other endpoint 12:59:15 <chandankumar> thanks to sean-k-mooney analysis , We have few undocumented rest api which is not implemented 12:59:24 <dviroel> i need some time to go over everything that you have there 12:59:29 <sean-k-mooney> i would suggest checkign if we ahve an other formiding api actions and remvoe them also 13:00:05 <chandankumar> apart from that, we have very little docs around watcher cli 13:00:15 <chandankumar> sean-k-mooney: sure 13:00:30 <dviroel> chandankumar: which is the next bug in the list? 13:00:37 <chandankumar> yes correct 13:00:51 <dviroel> #link https://bugs.launchpad.net/watcher/+bug/2110897 - Add more docs around openstack optimize audit command in doc 13:00:56 <sean-k-mooney> the cli docs are generally auto generated form the help text 13:01:06 <chandankumar> audit command is commonly used, openstack optimize audit provides a way to update/cancel an audit 13:01:21 <chandankumar> but there is no reference in the docs. 13:01:50 <chandankumar> sean-k-mooney: It means Do we need to add examples in Cli docs? or more text there? 13:02:32 <sean-k-mooney> it depend 13:02:45 <sean-k-mooney> we doen generally add exampel of command output 13:03:28 <sean-k-mooney> if we look at server create as an exmaple https://docs.openstack.org/python-openstackclient/latest/cli/command-objects/server.html#server-create 13:03:32 <chandankumar> Just added it in the doc 13:03:52 <chandankumar> Edit: just added command output in the bug 13:03:57 <sean-k-mooney> well im not really sure we need more docs 13:04:38 <sean-k-mooney> so the problem is https://docs.openstack.org/python-watcherclient/latest/cli/openstack_cli.html 13:04:45 <sean-k-mooney> shoudl be autogenerate based on the help text 13:04:52 <sean-k-mooney> but currently its not being auto generated 13:05:12 <sean-k-mooney> so the way to fix that is to make the doc not be static 13:05:51 <sean-k-mooney> so we shoudl be replacing https://github.com/openstack/python-watcherclient/blob/master/doc/source/cli/openstack_cli.rst?plain=1 13:06:10 <chandankumar> ah ok, let open a new bug to generate the cli docs automatically. 13:06:20 <sean-k-mooney> with autogenerate content we can likely look at other plugins for a reference on how to do that 13:06:39 <dviroel> right 13:06:40 <sean-k-mooney> for example barbican or manilla proably have examples of that 13:07:07 <dviroel> lets revisit this topic again next meeting chandankumar? if you could take a look on this debt 13:07:18 <chandankumar> dviroel: sure 13:07:33 <dviroel> chandankumar: tks 13:07:36 <chandankumar> I will bring it in next meeting, thank you dviroel sean-k-mooney ! 13:07:58 <dviroel> #topic next meeting chair 13:08:09 <dviroel> if you want to be the chair next week 13:08:12 <dviroel> please add your name 13:08:16 <chandankumar> I will chair for the next meeting. 13:08:20 <dviroel> tks chandankumar 13:08:31 <dviroel> that's it 13:08:35 <dviroel> let's wrap up for today 13:08:39 <dviroel> we will meet again next week 13:08:42 <dviroel> thank you all for participating 13:08:47 <dviroel> #endmeeting