14:00:06 <rafaelweingartne> #startmeeting cloudkitty
14:00:07 <openstack> Meeting started Mon Mar 22 14:00:06 2021 UTC and is due to finish in 60 minutes.  The chair is rafaelweingartne. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:08 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:10 <openstack> The meeting name has been set to 'cloudkitty'
14:00:19 <rafaelweingartne> Hello guys!
14:00:20 <rafaelweingartne> Roll count
14:01:10 <priteau> o/
14:01:11 <mkarpiarz> Hi!
14:01:15 <Akki58> priteau, what to do now ?
14:01:27 <priteau> Akki58: We're starting a meeting, let's continue later
14:01:47 <Akki58> oh, i am really sorry, i just noticed tha
14:02:02 <Akki58> *that
14:03:32 <priteau> Hi rafaelweingartne and mkarpiarz
14:04:24 <rafaelweingartne> #topic Wallaby release
14:04:35 <rafaelweingartne> I guess that we are moving on to release CloudKitty under Wallaby
14:04:58 <rafaelweingartne> most (if not) of the issues that we wanted for Wallaby have already been merged.
14:05:18 <rafaelweingartne> I have also reviewed the patch to cut RC1 last week if I am not mistaken
14:05:36 <priteau> NOTNUMBOOL didn't make it :/
14:05:37 <rafaelweingartne> Do you guys have some patch that we forgot to merge, and that should still make Wallaby?
14:06:13 <priteau> Well, technically we're past feature freeze so it should be bug fixes only now
14:06:24 <priteau> Do we want this? https://review.opendev.org/c/openstack/cloudkitty/+/777379
14:06:44 <mkarpiarz> It's fine. No point rushing non-critical patches.
14:07:32 <rafaelweingartne> priteau, that one I am not sure about
14:07:46 <rafaelweingartne> I mean, it seems that they were using a wrong configuration
14:08:06 <priteau> I've not worked much with Gnocchi recently so I don't know if it's the correct fix or not
14:08:11 <rafaelweingartne> it seems that they have an Gnocchi installation where the aggregation methods are configured with 300, and that is why they needed that config there
14:08:33 <rafaelweingartne> moreover, if they have some different aggregation method, they can use in their "custom" metric files
14:08:41 <priteau> But this isn't a default in Gnocchi?
14:08:43 <rafaelweingartne> there is no need to add that in CloudKitty
14:09:28 <priteau> So should Thomas ship a custom metrics.yml then?
14:10:13 <rafaelweingartne> let me check gnocchi for the default
14:10:30 <rafaelweingartne> we are used to explicitly define these things, that I am not sure about the default
14:11:55 <priteau> It would be interesting to check what configuration is used when deploying CloudKitty and Gnocchi with DevStack.
14:12:26 <mkarpiarz> Yeah, this is about different granurality set on the archive policy for some metrics.
14:12:48 <mkarpiarz> We've had a discussion about this here: https://storyboard.openstack.org/#!/story/2008598
14:14:14 <mkarpiarz> The default granularity is set in Cloudkitty to 3600s.
14:14:46 <rafaelweingartne> by default gnocchi uses all granularities that is configured in the archive policy in the aggregation API (used by cloudkitty)
14:14:59 <rafaelweingartne> and as mkarpiarz said, the default used in CloudKitty is 3600
14:15:11 <mkarpiarz> Which of course works for metrics with granularity set to 1h but not for those with granularity 5min.
14:15:35 <rafaelweingartne> exactly
14:16:01 <priteau> Would a typical ceilometer + gnocchi deployment have granularity of 300 on some metrics?
14:16:39 <rafaelweingartne> https://gnocchi.xyz/operating.html#default-archive-policies
14:16:55 <rafaelweingartne> priteau, it depends on the arquive policies set to resources
14:17:17 <rafaelweingartne> let me check the default in Ceilometer now
14:18:31 <priteau> I think we should ship defaults that work with defaults of other OpenStack services
14:18:31 <rafaelweingartne> https://github.com/openstack/ceilometer/blob/master/ceilometer/publisher/data/gnocchi_resources.yaml
14:19:09 <priteau> So 5 minutes on ceilometer-low?
14:19:19 <rafaelweingartne> yes
14:20:03 <rafaelweingartne> so, if we want to comply with Ceilometer default configs
14:20:16 <rafaelweingartne> then, yes, it might be interesting to merge that patch
14:20:48 <priteau> At the very least, this change should have:
14:20:54 <rafaelweingartne> the force granularity would need to be applied to other resources as well though
14:20:55 <priteau> Story and Task lines to link to storyboard
14:21:05 <priteau> a release note
14:22:50 <mkarpiarz> I'd also ask to add a note about force_granularity and which metrics might be affected to docs.
14:23:40 <mkarpiarz> Or at least point people to the Story where we have dicussion of how to deal with this problem.
14:23:54 <rafaelweingartne> Do we do these things? Or, we ask for the contributor that is proposing the patch?
14:24:13 <priteau> zigo: are you able to update your patch?
14:24:23 <priteau> If not, we would need to do ourselves
14:24:38 * zigo reads the backlog
14:25:04 <zigo> priteau: To include volumes fix too, isn't it?
14:26:41 <zigo> My config for gnocchi and ceilometer is indeed the default one.
14:26:52 <priteau> I don't know what granularity is used for volumes, sorry
14:26:54 <zigo> So, by default, we have a non-working cloudkitty, which is why I wrote the patch.
14:27:36 <zigo> volume.size also needs force_granularity: 300
14:28:28 <rafaelweingartne> from what I see, all of the default metrics in Ceilometer are configured as granularity=300
14:29:07 <zigo> I'm able to add this, but do all the work with release notes, stories, etc, I wont have time.
14:29:37 <priteau> The story is already there
14:29:46 <priteau> It's just adding the lines in the commit message as you update it
14:29:54 <priteau> We can add a release note if that's a problem
14:30:03 <zigo> I can do the release note. :)
14:30:15 <zigo> What's the syntax for pointing to the story ?
14:32:30 <priteau> See example: https://opendev.org/openstack/cloudkitty/commit/c248af3ecb61f7b6d297ca5325436f55f1c5760f
14:34:05 <mkarpiarz> Will this change close the Story?
14:34:11 <zigo> Yes.
14:35:19 <openstackgerrit> Thomas Goirand proposed openstack/cloudkitty master: Set force_granularity: 300 in metrics.yml  https://review.opendev.org/c/openstack/cloudkitty/+/777379
14:35:21 <mkarpiarz> Like I said, it could be useful to document why force_granularity should be set on some metrics and how to find these metrics.
14:35:37 <mkarpiarz> Otherwise users will run into the same issue over and over again.
14:35:38 <zigo> I very much would love to know myself !
14:36:14 <priteau> Is this a new issue? How comes it has been reported only now
14:36:20 <mkarpiarz> zigo: Have you read the discussion under https://storyboard.openstack.org/#!/story/2008598?
14:36:26 <rafaelweingartne> Do we work to merge that patch before Wallaby? I mean, for Wallaby?
14:36:57 <priteau> Depends if it can have negative side effects?
14:37:04 <rafaelweingartne> priteau, I guess that people using CloudKitty before were always customizing Ceilometer gnocchi_resources files, and of course, CloudKitty metrics.yml as well
14:37:31 <mkarpiarz> Specifically https://storyboard.openstack.org/#!/story/2008598#comment-170473 and https://storyboard.openstack.org/#!/story/2008598#comment-170660 - these are explaining why you need different grantularities and how to find granularities on metrics.
14:38:19 <zigo> In there, it's written: "Now, in Cloudkitty's Gnocchi collector granularity is set to 3600 seconds by default. " Can this be configured somewhere?
14:39:25 <zigo> Oh, I got it ...
14:39:32 <mkarpiarz> Well, the value is specified in the source code and can be overriden with force_granularity.
14:39:37 <zigo> [collect]
14:39:37 <zigo> period = 3600 <--- That ?
14:39:45 <zigo> Ah ok.
14:40:05 <priteau> That's something else ^^
14:40:09 <mkarpiarz> https://opendev.org/openstack/cloudkitty/src/branch/master/cloudkitty/collector/gnocchi.py#L115
14:40:26 <priteau> period is the time window used for rating
14:40:33 <mkarpiarz> Required('force_granularity', default=3600): All(int, Range(min=0)),
14:41:14 <zigo> Got it.
14:41:22 <zigo> Well, in such case, my patch is making sense.
14:41:56 <zigo> If someone goes to edit the gnocchi / ceilometer defaults, then that someone is also expected to know what he's doing and edit cloudkitty's metrics.yml.
14:42:06 <rafaelweingartne> yes, it is, but you need to add the force_granularity for other metrics that have the same default of 300 to make it complete
14:42:11 <rafaelweingartne> zigo: I would say so
14:42:39 <zigo> Except that other metrics aren't rated by default by cloudkitty.
14:43:00 <zigo> The point is just to have these that are configured by default in cloudkitty to match what is also by default in ceilometer.
14:43:11 <rafaelweingartne> Normally, before putting to production, I would expect people to plan the aggregation methods they want and the granularity for each metric
14:43:30 <mkarpiarz> Will a note in docs be part of https://review.opendev.org/c/openstack/cloudkitty/+/777379 or should we create a separate Task for it in the Story?
14:43:47 <zigo> We did lots of trials and error, and found out that the default were kind of ok-ish..
14:43:48 <mkarpiarz> (so a different patch can reference the task number)
14:44:06 <priteau> mkarpiarz: That's up to zigo
14:44:28 <priteau> We can always reopen the story with a TODO task if needed
14:44:43 <mkarpiarz> Ah, OK.
14:44:47 <zigo> I'm just discovering stuff, so I'm not 100% clear on how to write such a doc.
14:45:03 <zigo> I'd prefer someone else takes the job for now, as I'm not confident enough.
14:45:11 <rafaelweingartne> guys, we only have more 15min. So, to wrap up this topic. Zigo is going to amend the patch he proposed, and we are going to merge it for Wallaby? Is that it?
14:45:31 <zigo> Good for me.
14:45:45 <zigo> I got more topics to discuss..
14:46:14 <mkarpiarz> zigo: To update docs you simply edit files in https://opendev.org/openstack/cloudkitty/src/branch/master/doc/source
14:46:27 <zigo> Sure ! :)
14:46:32 <zigo> That's not where I lack knowledge.
14:46:47 <rafaelweingartne> So, moving on. Do you guys have some other review priorities?
14:46:48 <rafaelweingartne> #topic Review priorities
14:46:50 <zigo> Where I lack knowledge is all about archive policies, aggregation, etc.
14:47:15 <rafaelweingartne> Things that we might want to see in Wallaby version of CloudKitty
14:47:20 <zigo> Yeah.
14:47:24 <zigo> cloudkitty dataframes get -p <--- Doesn't work for me.
14:47:46 <priteau> zigo: Let's discuss outside of the meeting if possible
14:48:01 <zigo> When I do -p, cloudkitty returns with a policy error, even though my policy file is in order.
14:48:07 <zigo> priteau: I do believe this is a bug...
14:48:23 <zigo> But ok, I'm fine taking this off-meeting.
14:48:25 <priteau> zigo: Topic is review priorities
14:48:33 <zigo> ok
14:48:47 <zigo> Ping me later then.
14:49:05 <priteau> rafaelweingartne: I don't see other bug fix candidates for Wallaby
14:49:15 <priteau> Of course we can start reviewing Xena patches
14:49:33 <priteau> I've given my +2 to NOTNUMBOOL: https://review.opendev.org/c/openstack/cloudkitty/+/774084
14:49:42 <rafaelweingartne> sure, I aggree
14:51:21 <mkarpiarz> Are Xena patches the ones submitted by the OpenStack Release Bot?
14:51:50 <priteau> Two patches for the Wallaby branch and two for master
14:51:58 <priteau> For each repository
14:52:22 <mkarpiarz> Yep
14:52:41 <mkarpiarz> So we have to review and merge them all, right?
14:52:42 <priteau> There's a failure to investigate here: https://review.opendev.org/c/openstack/cloudkitty/+/782093
14:52:49 <priteau> AssertionError: 1.0000000003174137e-05 not less than or equal to 1e-05
14:53:20 <priteau> Yes, they all need to be merged so the wallaby branch can receive patches and release notes be published
14:54:26 <rafaelweingartne> Do we have something else that we need to discuss with respect to review priority?
14:55:19 <priteau> We should use last 5 min to discuss PTG
14:55:36 <rafaelweingartne> yes
14:55:44 <mkarpiarz> Nothing on my side when it comes to review priorities.
14:56:09 <rafaelweingartne> Now, I open for general questions and topics that people might have.  Then, we can use this space to discuss the PTG
14:56:15 <rafaelweingartne> #topic AOB
14:56:51 <rafaelweingartne> I have participate in a few of them last year, should we try to schedule something?
14:56:53 <priteau> PTG is important because signup closes this week
14:57:25 <priteau> I think it would be good to book maybe 2 hours to go over plans for Xena via video
14:57:51 <rafaelweingartne> Do we have a place to schedule these times?
14:57:57 <priteau> #link https://ethercalc.net/oz7q0gds9zfi
14:58:33 <rafaelweingartne> ok, I will send an email later today
14:58:43 <rafaelweingartne> so we can try to select/converge on some dates and times
14:58:49 <priteau> There's space on Monday so we could schedule around the same time as our IRC meeting
14:58:57 <priteau> Would make it easy to remember
14:59:03 <priteau> i.e. April 19
14:59:12 <priteau> 15-17 for example
14:59:17 <priteau> But do send email :)
15:00:02 <rafaelweingartne> We will need to close. Thank you guys for participating. Have a nice week.
15:00:05 <rafaelweingartne> #endmeeting