14:01:10 <peschk_l> #startmeeting cloudkitty
14:01:11 <openstack> Meeting started Mon Dec  2 14:01:10 2019 UTC and is due to finish in 60 minutes.  The chair is peschk_l. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:12 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:14 <openstack> The meeting name has been set to 'cloudkitty'
14:01:36 <peschk_l> Hi everybody, welcoe to this cloudkity meeting
14:01:47 <peschk_l> today's agenda can be found at https://etherpad.openstack.org/p/cloudkitty-meeting-topics
14:02:18 <peschk_l> as usual, feel free to add any topic you'd like to discuss to the end of the agenda
14:02:39 <peschk_l> and for any cloudkitty-related question, we'll have some time for Q&A at the end of the meeting
14:03:01 <peschk_l> All right let's start with ou first topic
14:03:05 <peschk_l> #topic python2
14:04:07 <peschk_l> python2.7 testing has been dropped for all projects except the client (which should happen between the ussuri-1 and ussuri-2 milestones)
14:04:50 <peschk_l> droppping support for python2.7 is the only community goal for now, so we're completely in time  :-)
14:05:24 <peschk_l> Please do not contribute python3-only code yet, as RDO still builds their packages with python2
14:06:02 <peschk_l> FYI, once python2 support has been completely dropped, the minimal supported python version will be 3.6
14:06:29 <peschk_l> and I think that's pretty much it concerning python2. Any questions on this topic ?
14:07:18 <peschk_l> #topic cloudktityclient 3.2.0
14:08:12 <peschk_l> No that the required endpoints have been integrated in the v2 API (/v2/dataframes and /v2/summary), we'll be able to add support for report generation through the client
14:08:18 <peschk_l> *now that
14:08:47 <peschk_l> Once these features have been merged, and support for python2.7 has been dropped, we'll releas python-cloudkittyclient 3.2.0
14:09:04 <peschk_l> This will allow us to delete cloudkitty-writer
14:09:53 <peschk_l> we'll probably use the df-to-csv formatter for the reports
14:10:49 <peschk_l> does somebody have a question about this ?
14:10:58 <jferrieu> I'm good
14:11:40 <peschk_l> okay, the next one is open for debate
14:11:47 <peschk_l> # topic InfluxDB storage driver
14:11:52 <peschk_l> #topic InfluxDB storage driver
14:12:14 <peschk_l> should the influxdb storage driver be considered stable ?
14:13:04 <peschk_l> it's been running in production in several deployments for a while now
14:13:14 <jferrieu> I think it is ok, as some users did deploy CK with it and didn't raise any issues since then
14:13:41 <peschk_l> we still merge some minor fixes, like https://review.opendev.org/#/c/696090/ , but these are not critical bugs
14:13:51 <jferrieu> true
14:14:06 <peschk_l> and v1 storage is pretty much unmaintained, so it be great to get rid of it quickly
14:14:20 <jferrieu> can't agree more
14:14:34 <peschk_l> well "quickly", as we 'll need at least one deprecation cycle, and provide some scripts to migrate v1 data to v2
14:14:47 <jferrieu> of course
14:15:14 <peschk_l> glad we agree on that one jferrieu :-)
14:15:55 <peschk_l> which makes me think about another concern we had: should "metadata" be droppped ?
14:16:09 <peschk_l> it seems to be confusing for end users more than anything else
14:16:37 <peschk_l> and it'd probably be easier for everyone if we had only one type of attribute
14:16:45 <jferrieu> and the same logics are at play in the back regarding `groupby` and `metadata`
14:16:56 <jferrieu> it was just semantics in the first place or maybe I'm wrong
14:17:12 <peschk_l> yup, only difference being that metadata is not indexed in the influx storage driver
14:17:31 <peschk_l> no in the elasticsearch driver, but one's still experimental
14:17:45 <peschk_l> *neither in the ES driver
14:17:58 <jferrieu> would that impact the overall performance ?
14:18:17 <peschk_l> jferrieu: metadata were supposed to be unindexed and only available for rating rules
14:19:09 <peschk_l> jferrieu: there would be a theoric load on the storage backend, as everything would need to be indexed... But in practice i think that people only use "groupby", as they want to be able to group on anything
14:19:45 <jferrieu> maybe the semantics were misleading in a first place then, we could add an option to disable indexing on some fields later maybe
14:20:02 <jferrieu> if that would be critical
14:20:15 <openstackgerrit> Merged openstack/cloudkitty master: Fix 500 errors in the API when request context bears no project_id  https://review.opendev.org/696090
14:20:17 <jferrieu> but for the moment I agree to just merge the two concepts
14:20:56 <peschk_l> true, we could. Anyway we'll have to discuss this with users as it'd probably not be retroactive if we were to implement it
14:21:09 <jferrieu> true
14:21:16 <peschk_l> for example in influxdb it is not possible to convert tag to fields once they've been pushed
14:21:49 <jferrieu> I see
14:21:52 <peschk_l> and in ES if would require to update huge batches of documents, with some indices being potentially frozen depending on how the cluster is administrated
14:22:39 <peschk_l> jferrieu: anything to add about this ?
14:22:59 <jferrieu> I'm good
14:23:28 <peschk_l> all right, last topic before Q&A then
14:23:53 <peschk_l> #topic new aggregation functions for the prometheus collector
14:24:58 <peschk_l> aimbot31 has proposed a spec adding support for new prometheus query functions (like delta() ) to the prometheus collector
14:25:05 <peschk_l> it has been merged this morning
14:25:13 <peschk_l> the pacth can be found at
14:25:18 <peschk_l> #link https://review.opendev.org/#/c/678013/
14:25:44 <peschk_l> and the html version of the spec here:
14:25:47 <peschk_l> #link https://specs.openstack.org/openstack/cloudkitty-specs/specs/ussuri/add_prometheus_query_functions_to_the_collector.html
14:27:12 <peschk_l> I believe we've adressed all of today's topics. Time for Q&A
14:27:16 <peschk_l> #topic Q&A
14:27:41 <peschk_l> if anyone has a cloudkitty-related question or is looking for help, now's the time to ask
14:28:28 <jferrieu> I'm good for the moment
14:28:58 <julien-pinchelim> Same :)
14:31:34 <peschk_l> All right, thanks everybody for attending
14:31:39 <peschk_l> #endmeeting