14:04:47 <rafaelweingartner> #startmeeting cloudkitty
14:04:47 <opendevmeet> Meeting started Mon Nov 14 14:04:47 2022 UTC and is due to finish in 60 minutes.  The chair is rafaelweingartner. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:04:47 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:04:47 <opendevmeet> The meeting name has been set to 'cloudkitty'
14:04:51 <rafaelweingartner> Sorry for the delay!
14:05:01 <rafaelweingartner> Roll count
14:05:09 <mkarpiarz> Hi!
14:06:51 <priteau> Hello
14:09:34 <rafaelweingartner> #topic The move from ElasticSearch and towards OpenSearch
14:10:02 <rafaelweingartner> We discussed last week about this process, and we all agreed that it would be interesting to draft a notice message to send to the community
14:10:13 <rafaelweingartner> Pierre have you been able to prepare something?
14:10:34 <rafaelweingartner> When you read "last week", you should consider "last meeting"
14:11:33 <priteau> Sorry, I was on travel all of last week. I will start a draft now.
14:11:46 <rafaelweingartner> ok, thanks!
14:12:36 <rafaelweingartner> #topic Monasca deprecation
14:12:54 <rafaelweingartner> I prepared and send the e-mail querying the community about the use of Monasca
14:12:59 <rafaelweingartner> I got no response back
14:13:06 <rafaelweingartner> what do you guys think about that?
14:14:18 <rafaelweingartner> Should we consider that as a sign to start removing support for Monasca?
14:14:26 <mkarpiarz> Maybe no one is using this collector?
14:14:39 <rafaelweingartner> that is what I assumed
14:16:11 <priteau> At StackHPC we have migrated all our customer deployments away from it
14:18:13 <rafaelweingartner> Can we agree to move on with the deprecation notice on Antelope?
14:18:22 <rafaelweingartner> And then, removal on B release?
14:18:55 <opendevreview> Pierre Riteau proposed openstack/cloudkitty master: Announce future deprecation of Elasticsearch  https://review.opendev.org/c/openstack/cloudkitty/+/864412
14:19:00 <priteau> Agreed.
14:19:23 <rafaelweingartner> Ok
14:19:29 <priteau> If someone wanted it back it wouldn't be too hard to bring back, unless we made major changes to the collector interface
14:19:33 <rafaelweingartner> I will conduct that process then
14:20:21 <priteau> Thank you. I propose a note for Elasticsearch.
14:20:28 <rafaelweingartner> Thanks!
14:20:41 <rafaelweingartner> Then, movin on with the meeting
14:20:42 <rafaelweingartner> #topic Target review of the meeting
14:20:56 <rafaelweingartner> #link https://review.opendev.org/c/openstack/cloudkitty/+/861908
14:21:06 <rafaelweingartner> The gnocchi support matter
14:21:12 <rafaelweingartner> have you guys read that patch?
14:21:42 <rafaelweingartner> That is a serious issue in Gnocchi that we have known for a log time. However, it was a bit hard for us to get the patch merged upstream in Gnocchi
14:21:58 <rafaelweingartner> and now, it is time for CloudKitty to use Gnocchi correctly
14:22:17 <priteau> Looking now
14:22:26 <rafaelweingartner> The problem is that it is backwards incompatible. As soon as we add that to Cloudkitty, all people using gnocchi have to update it
14:23:30 <mkarpiarz> Yeah, that's my main concern (see my comment under the patch).
14:23:47 <priteau> What happens if you pass the arg in a request to old Gnocchi?
14:23:57 <priteau> Does it ignore it or deny the request?
14:24:58 <priteau> If it doesn't work (i.e. fails the request) with old gnocchi, I think this should be behind a feature flag
14:25:16 <priteau> i.e. an collector configuration option to enable use_history
14:25:33 <priteau> Default of false for now, and we change it to true in a few releases
14:25:39 <mkarpiarz> I agree.
14:25:55 <priteau> This is what other projects would do :)
14:26:43 <rafaelweingartner> it just ignores the parameter
14:27:17 <rafaelweingartner> I mean, Gnocchi will ignore the parameter, and might return incosistent data
14:27:54 <priteau> But will it be the same behaviour as before using use_history?
14:28:09 <priteau> i.e. broken before this patch, and broken in exactly the same way after?
14:28:16 <priteau> (if you are using old gnocchi)
14:28:35 <rafaelweingartner> exactly
14:28:51 <rafaelweingartner> use_history=false will produce the exact same behvior
14:29:01 <rafaelweingartner> it is the same code that is exeuted when the use_history=false
14:29:06 <rafaelweingartner> which is the default, when it is not used
14:30:08 <priteau> In that case I think we can go ahead with this change and just document the Gnocchi dependency in the collector docs when it is released
14:30:18 <priteau> Please add a release note bug fix though.
14:30:58 <rafaelweingartner> ok
14:31:01 <rafaelweingartner> I will do so then
14:31:04 <rafaelweingartner> thanks!
14:31:30 <priteau> Is there a test we can have in tempest for this?
14:32:16 <rafaelweingartner> not that I am aware of
14:32:26 <rafaelweingartner> tempest is using old Gnocchi versions
14:35:49 <priteau> But the test could be marked as known to be failing
14:35:59 <rafaelweingartner> hmm
14:36:03 <priteau> And then when we would run it against new gnocchi we would see that it is working
14:36:09 <rafaelweingartner> but I guess they are not testing this specific use case
14:36:12 <priteau> This is mostly to catch a possible regression
14:36:20 <rafaelweingartner> ah, I see
14:36:22 <priteau> Don't worry if this is too complex
14:36:45 <rafaelweingartner> I never created a tempest test
14:36:57 <rafaelweingartner> I can try to see how to create it, and then create this test case
14:38:28 <priteau> I don't think I've read much of our tempest plugin, I am not sure what we test
14:38:44 <priteau> It would be a good effort to learn more about it and improve if needed
14:39:19 <mkarpiarz> Definitely!
14:39:27 <rafaelweingartner> exactly
14:39:42 <rafaelweingartner> I will try to add it here then as well
14:40:51 <rafaelweingartner> Are we done with this Gnocchi topic? Or, do you guys have something else to add?
14:41:04 <priteau> Nothing else
14:41:12 <rafaelweingartner> So, moving on
14:41:16 <rafaelweingartner> #link https://review.opendev.org/c/openstack/cloudkitty/+/861806
14:42:05 <rafaelweingartner> I alwaready reviewed that patch, and it is in a good shapre. However, it is important for you guys to review as well
14:43:05 <rafaelweingartner> "alwaready" -> "already"
14:43:24 <rafaelweingartner> "shapre" -> "shape"
14:44:29 <mkarpiarz> I'm using this patch in my test environment and so far I see no issues.
14:45:17 <rafaelweingartner> cool
14:45:17 <mkarpiarz> Except for that missing metric I mentioned in a comment.
14:45:22 <rafaelweingartner> we are already using it as well
14:48:57 <priteau> What I don't really understand with this patch is whether it is Gnocchi-specific or if it would be useful for Prometheus too
14:49:06 <priteau> mkarpiarz: you've used both systems, could you advise?
14:49:13 <rafaelweingartner> useful for others
14:49:34 <rafaelweingartner> That is a new concept we are introducing to enable operators to re-use a metric for different cloudkitty ratings
14:50:38 <priteau> Shouldn't we implement it as collector-agnostic then?
14:51:57 <rafaelweingartner> yes, and no. We only had the need for Gnocchi, and each collector needs to access some context objects. Therefore, we did not want to change the others, and right now they will not support it
14:52:40 <rafaelweingartner> but the idea is to introduce that concept, and then if that is interesting for others, we can add to other collectors
14:54:02 <priteau> I would just like to make sure we can provide the same capability for other collectors without changing the interface
14:54:37 <rafaelweingartner> exactly, we did not change the collector interface
14:54:43 <rafaelweingartner> that is why we did this way
14:54:57 <rafaelweingartner> I would say that this is possible with other collectors as well
14:55:13 <rafaelweingartner> At the end of the day, we are just collecting the metric again
14:55:39 <rafaelweingartner> the only thing is that we have a different alt-name a group by and metadata sections
14:57:06 <priteau> Let's go with it then
14:57:21 <priteau> We have the whole release cycle to change the approach if needed
14:58:05 <mkarpiarz> OK
14:58:47 <rafaelweingartner> cool
14:58:55 <rafaelweingartner> Well, we are running out of time
14:59:09 <rafaelweingartner> I will open the floor for general questions now
14:59:14 <rafaelweingartner> #topic Open floor moment
14:59:26 <rafaelweingartner> Do you guys have something else do add to the meeting?
15:00:12 <priteau> None from me
15:00:17 <mkarpiarz> I would like to mention my message about v1 docs:
15:00:20 <mkarpiarz> #link https://lists.openstack.org/pipermail/openstack-discuss/2022-November/031147.html
15:00:44 <rafaelweingartner> Yes, I saw the message, but I have not read it yet
15:01:31 <mkarpiarz> That's fine. We can discuss this one on through the mailing list.
15:01:38 <rafaelweingartner> ok
15:01:41 <rafaelweingartner> deal
15:01:44 <mkarpiarz> I just wanted to make sure you guys are aware of it. :)
15:01:50 <rafaelweingartner> :)
15:01:59 <rafaelweingartner> Thank you guys for participating. Have a nice week.
15:02:07 <rafaelweingartner> #endmeeting