15:00:23 <witek> #startmeeting monasca
15:00:24 <openstack> Meeting started Wed May  8 15:00:23 2019 UTC and is due to finish in 60 minutes.  The chair is witek. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:27 <openstack> The meeting name has been set to 'monasca'
15:00:35 <witek> hello everyone
15:00:37 <koji_n> hi
15:01:09 <witek> agenda for today:
15:01:12 <witek> https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:01:58 <dougsz> hello all - apologies for not making Saturday at PTG, the honest answer was that I totally forgot :(
15:02:12 <joadavis> :)
15:02:24 <witek> :) no worries, we covered most of the topics on Friday already
15:02:33 <joadavis> good to know you are human
15:02:48 <dougsz> :-D
15:03:29 <witek> #topic Train tasks prioritization
15:04:03 <witek> I think the list should be complete now
15:04:26 <witek> I've created new or linked to existing Stories
15:04:59 <witek> there is also a link to meeting minutes, I haven't managed to write a summary yet
15:06:21 <witek> please also let me know or just add to the list if something is missing or should be differently handled
15:07:17 <witek> I would like to evaluate the results in one week
15:07:29 <dougsz> ok, thanks witek
15:08:33 <witek> any comments right now to list items?
15:08:41 <witek> or anything related?
15:09:28 <joadavis> its a good list
15:09:36 <dougsz> Should merge APIs task be on there, I guess that should make Train?
15:09:54 <witek> oh yes, forgot to comment on that one
15:10:03 <dougsz> (just thinking about the log-query API which will come after)
15:10:18 <joadavis> right, we discussed if tasks in progress should be on the list
15:10:21 <witek> I assumed we will continue the work on merging APIs and Kafka upgrade
15:10:44 <witek> so these will automatically make it to the priorities list
15:10:52 <dougsz> sounds good, thanks
15:12:25 <witek> joadavis, dougsz: I don't remember if we decided how we should handle these, is it OK to skip them from the game or you prefer to have them as well?
15:13:06 <Dobroslaw> I would prefer them in game so we see all "big" tasks
15:13:16 <dougsz> I guess Kafka upgrade is already part of Middleware upgrade
15:13:21 <Dobroslaw> that are worked on/considered
15:13:41 <joadavis> I'd add a second section below the table for voting and list them there for context
15:13:43 <witek> dougsz: no, client upgrade is meant, it includes code changes
15:13:54 <joadavis> or make them a different color so they stand out as in-progress
15:15:57 <witek> OK, I'll add the two to the list and mark as in progress, we have stories for them already
15:16:14 <dougsz> I suppose from a reviewing perspective, it would be nice to have them all there, to make it clear where review time should be targetted.
15:16:18 <witek> will update after the meeting
15:16:47 <witek> any other comments?
15:17:29 <witek> #topic TimescaleDB
15:17:39 <Dobroslaw> that's from me
15:17:47 <Dobroslaw> I found it few days ago
15:18:10 <Dobroslaw> and it looks promising as a alternative for influx/cassandra
15:18:20 <Dobroslaw> I added more links in etherpad
15:18:34 <Dobroslaw> in short: extension to PostgreSQL
15:19:19 <Dobroslaw> a lot faster than PostgreSQL for storing and handling metrics (use something called hypertable on top of Postgres)
15:19:37 <Dobroslaw> apparently also a lot faster than cassandra
15:20:01 <Dobroslaw> slower than influx if we send few metrics from every node
15:20:29 <Dobroslaw> but if number of metrics from one node increase it's starting to be faster than inflax
15:21:01 <Dobroslaw> big point: we can use most tools for Postgres for HA
15:21:27 <Dobroslaw> https://blog.timescale.com/high-availability-timescaledb-postgresql-patroni-a4572264a831/
15:21:48 <Dobroslaw> There are also cons like much higher disk usage than Influx in exchange for less memory and CPU use. They recomend to run TimescaleDB on top of ZFS as users often see 3-9x compression.
15:21:58 <witek> do they have any dedicated clients, or standard mysql should be used?
15:22:09 <dougsz> I suppose for me, it's an interesting thing to keep an eye on for the future.
15:22:24 <Dobroslaw> as far as I understand standard sql
15:23:13 <witek> how does it scale horizontally, any limitations?
15:23:42 <Dobroslaw> ^ link have more info
15:24:04 <Dobroslaw> using etcd or zookeeper or anything similar for choosing master
15:24:37 <Dobroslaw> limitations I would say similar to Postgres
15:25:06 <Dobroslaw> need streaming replication from HA tools
15:25:52 <witek> https://docs.timescale.com/v1.3/faq#clustered
15:26:06 <witek> "Full, scale-out clustering (i.e., for data volumes > 500 billion rows and ingest rates > 300K row inserts / second) is in the works and will leverage the automated partitioning capabilities already available in single-node TimescaleDB today."
15:26:24 <witek> not sure, if that's up to date
15:27:48 <Dobroslaw> well, it's just another tool on Apache 2.0 to check from time to time
15:28:30 <witek> I've heard of TimescaleDB before and it seems they are actively developing it
15:29:05 <dougsz> I suppose the other question is how well does it handle high cardinality
15:29:12 <Dobroslaw> they raised 15M $ in January
15:29:22 <joadavis> it is appealing that it supports "all of SQL"
15:29:24 <Dobroslaw> https://blog.timescale.com/what-is-high-cardinality-how-do-time-series-databases-influxdb-timescaledb-compare/
15:29:37 <dougsz> :)
15:29:41 <Dobroslaw> dougsz: newest post on they blog :)
15:30:12 <dougsz> I wonder what their business model is
15:30:21 <witek> they also have Prometheus remote read/write adapter
15:30:35 <witek> https://github.com/timescale/prometheus-postgresql-adapter
15:30:38 <Dobroslaw> https://www.timescale.com/pricing
15:31:28 <Dobroslaw> Timescale License basically block Amazon and similar from providing it as a SaaS
15:31:35 <Dobroslaw> DBaaS
15:32:04 <Dobroslaw> dougsz: https://blog.timescale.com/how-we-are-building-an-open-source-business-a7701516a480/
15:33:10 <witek> hm, could licensing also be an issue for us?
15:33:33 <Dobroslaw> Community features (free): Will be licensed under the TSL, but will be made available at no charge. There are two exceptions that are prohibited from using these features (without a commercial agreement): (1) cloud and SaaS providers who just offer a hosted “database-as-a-service” version of TimescaleDB, and (2) OEMs that don’t provide any other value on top of the database. This tier enables us to continue
15:33:34 <Dobroslaw> to provide free features to and invest in our community, while protecting from open-source strip mining.
15:33:47 <Dobroslaw> Base is Apache 2.0
15:34:17 <Dobroslaw> Differences in functionality here: https://www.timescale.com/pricing
15:34:27 <witek> I see some functionality is  not included in Apache license
15:35:43 <Dobroslaw> I did not read this license but Monasca IS providing value on top of the database
15:36:05 <witek> I understand that implementing support for TimescaleDB addresses the same problems as `Sharding model for InfluxDB` on our priorities list
15:36:08 <Dobroslaw> but I'm not lawyer
15:36:26 <witek> Dobroslaw: correct, wanted to comment with the same sentence
15:36:42 <witek> I guess we would need one to clarify
15:36:58 <Dobroslaw> yea... I hate this license wars...
15:36:59 <witek> or check if Apache is enough for us
15:37:40 <witek> should we be adding Timescale to our game list?
15:38:01 <Dobroslaw> don't know, up to you all
15:38:27 <witek> have you discussed it in your team?
15:39:00 <Dobroslaw> I just showed it today to Adrian but he did not look much into this
15:39:20 <Dobroslaw> just said it's nice that we can connect to it like to any SQL db
15:40:26 <witek> I'd be interested in Matthias' opinion
15:40:39 <Dobroslaw> he is sick this week
15:41:32 <witek> for now, I would say we can keep an eye on it and can evaluate, but don't include in the list yet
15:41:43 <Dobroslaw> fine by me
15:41:54 <witek> let's move on
15:42:16 <dougsz> thanks Dobroslaw
15:42:21 <Dobroslaw> :)
15:42:23 <witek> #topic new tag for monasca-log-api Pike
15:42:47 <witek> dougsz: thanks for the heads-up
15:43:03 <witek> I've given +1 today
15:43:21 <dougsz> Thanks witek, I've been trying to gets someone to W+1 in reviews
15:43:45 <witek> everyone busy after Summit I assume
15:43:58 <dougsz> yeah
15:44:16 <dougsz> I guess that is it for this
15:44:30 <joadavis> Looks ok to me, should I give it the W+!?
15:44:43 <witek> joadavis: it
15:44:45 <dougsz> Can you do it in releases?
15:45:21 <joadavis> oh, good point. :(
15:45:29 <witek> I can ping in #openstack-release
15:46:00 <witek> #topic Readme formating errors in Pike
15:46:12 <witek> https://review.opendev.org/652854
15:46:17 <witek> reminder from last time
15:46:36 <witek> we cannot tag on Pike because of formatting errors in Readmes
15:47:58 <witek> http://logs.openstack.org/54/652854/3/check/openstack-tox-validate/14a09ef/job-output.txt.gz#_2019-04-18_05_37_47_484135
15:48:11 <witek> it would be great if someone could pick it up
15:48:36 <witek> I can help as well
15:49:21 <witek> #topic Python 3 support in monasca-agent
15:49:45 <joadavis> I posted a WIP patch for py36 environment enabling in mon-agent
15:49:55 <witek> nice, thanks
15:50:15 <witek> still a lot of work to turn it green, right?
15:50:17 <joadavis> There are still a number of errors that result from the patch, and will need to track them down
15:51:29 <joadavis> zuul says 86, which is more than I saw locally. :)
15:51:46 <witek> do you want to work on this?
15:52:11 <joadavis> I did when I thought the list of errors would be shorter. ;)
15:52:21 <witek> haha
15:52:34 <witek> perhaps it could be split
15:52:41 <joadavis> I will take another pass at it today, and will ask for help where I can't find an easy fix
15:53:01 <witek> cool, thanks
15:53:09 <joadavis> I suspect there are many repeated errors, and most can be solved by a quick google for the equivalent call, or importing six
15:53:42 <joadavis> We should check all the other monasca-* projects are properly invoking the py36 unit tests
15:53:43 <dougsz> Will try and take a look sometime
15:54:11 <witek> I think monasca-agent is the only one not executing tests
15:54:18 <joadavis> thanks
15:54:55 <witek> btw., I've added the story to the new board
15:54:55 <joadavis> and I would appreciate anyone with a bit more tox expertise reviewing the tox.ini change to make sure I didn't miss something. :)
15:54:58 <witek> https://storyboard.openstack.org/#!/board/141
15:55:41 <witek> #topic Telemetry meeting
15:56:00 <witek> Telemetry team has scheduled a bi-weekly meeting
15:56:15 <witek> on Thursdays 2am UTC
15:56:40 <witek> http://eavesdrop.openstack.org/#Telemetry_Team_Meeting
15:57:11 <witek> middle of the night for me
15:57:28 <joadavis> 7pm here, but I have a conflict tonight
15:58:13 <witek> we've seen a lot of interest during the Summit from people asking how to migrate from Ceilometer/Aodh to Monasca
15:58:16 <joadavis> I may have to just review the logs.  I do want to get a conversation with the telemetry team about publishers
15:58:45 <dougsz> +1 People have been showing up in Kolla-Ansible asking for help with Monasca for this purpose
15:59:10 <joadavis> yes, it might be good to write a quick article about how to migrate and what are some differences
15:59:27 <witek> volunteers? :)
15:59:40 <dougsz> What is the reference design for Monasca?
15:59:44 <joadavis> hah, are you looking at me?
16:00:15 <joadavis> dougsz: good question. should we be steering users to kolla or docker or a distribution?
16:00:37 <joadavis> or just give a simple devstack based example
16:00:57 <witek> dougsz: what do you mean by reference?
16:01:18 <dougsz> Ceilometer replacement
16:01:51 <witek> don't understadn
16:02:10 <dougsz> I suppose these people want to do billing?
16:02:33 <joadavis> some will, some just want monitoring
16:04:22 <witek> I think Monasca publisher is important, we should also describe installation methods (or link) in the central place (d.o.o)
16:04:24 <joadavis> I do wish we had a recent user who had moved so we could reference their experience.
16:04:53 <witek> article is a great idea
16:05:19 <witek> we could also reach out to Telemetry team and ask if they are interested in migrating
16:05:24 <joadavis> It might be easiest to start hacking out the article as a page on the wiki, then when ready we can try submitting it to the Superuser magazine
16:05:32 <witek> and provide use cases which have to be supported
16:06:37 <witek> OK, I think we can wrap up now, we're over time anyway
16:06:49 <witek> but we should definitely follow up on this
16:06:59 <witek> thanks for joining
16:07:04 <joadavis> yes.  I'll start an outline and will post the link to this channel
16:07:09 <dougsz> Thanks all
16:07:09 <witek> please remember about priorities
16:07:17 <witek> thanks joadavis
16:07:17 <koji_n> thanks
16:07:20 <witek> thanks everyone
16:07:27 <witek> bye
16:07:31 <witek> #endmeeting