13:00:21 <witek> #startmeeting monasca
13:00:21 <openstack> Meeting started Tue Feb  4 13:00:21 2020 UTC and is due to finish in 60 minutes.  The chair is witek. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:22 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:25 <openstack> The meeting name has been set to 'monasca'
13:00:33 <witek> hello everyone
13:00:40 <chaconpiza> Hello
13:00:41 <piotrowskim> hi
13:00:48 <adriancz> Hello
13:00:51 <Dobroslaw> hi
13:00:53 <arseni-lipinski> hi
13:01:03 <witek> hi, strong Fujitsu today :)
13:01:08 <adriancz> :)
13:01:16 <witek> nice
13:01:27 <witek> the agenda for today
13:01:32 <witek> https://etherpad.openstack.org/p/monasca-team-meeting-agenda
13:01:58 <witek> please add items if applicable
13:02:17 <witek> #topic ujson not maintained
13:02:59 <witek> it was raised in openstack-discuss that ujson library is causing problems when deploying Ceilometer in Kolla
13:03:23 <witek> the library hasn't been maintained for a long time already
13:04:04 <witek> and we should consider replacing it with some alternative
13:04:35 <witek> I've included related links into etherpad
13:04:43 <witek> https://github.com/esnme/ultrajson/issues/343
13:04:51 <chaconpiza> I imagine that ujson was used in order to improve performance, instead of regular json or simplejson ?
13:05:33 <witek> yes, we're using it for serialization and wanted to improve performance
13:06:57 <witek> from what I recall, deserializing of Kafka messages was one of the items identified in persister to consume most of the time
13:07:27 <witek> Dobroslaw: was that you doing the profiling?
13:07:39 <Dobroslaw> yes
13:08:14 <Dobroslaw> if I remember correctly then main problem with speed was kafka lib
13:08:42 <witek> right, deserialization second?
13:09:40 <Dobroslaw> I think yes
13:10:06 <Dobroslaw> but my memory is hazy about that
13:11:22 <witek> could you try to dig out the results?
13:12:10 <witek> so what alternatives do we have?
13:12:27 <piotrowskim> orjson?
13:12:39 <witek> we could go with json from stdlib, simplejson
13:12:50 <witek> yes, orjson
13:13:01 <witek> there is also rapidjson
13:13:03 <adriancz> https://artem.krylysov.com/blog/2015/09/29/benchmark-python-json-libraries/
13:13:13 <adriancz> pleas cehck this benchmark
13:13:50 <adriancz> it looks like simplejson is faster  then ujson in python3
13:14:06 <witek> what is actually relevant for us is to know, how fast our messages are getting deserialized
13:14:11 <adriancz> but this benchmark is quite old ( 2016/08/13)
13:14:33 <witek> it's all about this line of code I guess:
13:14:38 <witek> https://opendev.org/openstack/monasca-persister/src/branch/master/monasca_persister/repositories/utils.py#L21
13:15:39 <witek> so running a quick comparison check should be easy
13:17:16 <witek> there is also another aspect: orjson and rapidjson are not included in global requirements
13:19:46 <witek> would someone like to do a comparison check of alternatives for our use case?
13:21:03 <chaconpiza> Do you think results using Devstack can be used?
13:21:12 <witek> yes
13:21:44 <witek> I mean, messages from DevStack
13:22:05 <chaconpiza> I will try
13:22:31 <witek> but measuring only deserialization as done in persister
13:24:05 <witek> that would be cool, thanks Martin
13:24:17 <chaconpiza> I will contact in case of more details
13:24:50 <witek> thanks, any other comments on that topic?
13:25:47 <witek> #topic OpenDev + PTG
13:26:03 <witek> continuing the topic from last week
13:26:45 <witek> TC encourages devs and ops to shape the content of the event
13:26:50 <witek> http://lists.openstack.org/pipermail/openstack-discuss/2020-January/012239.html
13:27:14 <witek> is anyone interested in traveling and contributing?
13:27:47 <witek> monitoring session could be interesting to many
13:30:37 <chaconpiza> I am interested. So, I will ask my manager.
13:30:42 <witek> +1
13:30:47 <adriancz> +2
13:31:53 <witek> #AOB
13:31:58 <witek> #topic AOB
13:32:25 <witek> I've attended FOSDEM this weekend
13:32:54 <witek> on Sunday there was a complete track on Monitoring and Observability
13:32:54 <witek> https://fosdem.org/2020/schedule/track/monitoring_and_observability/
13:33:40 <witek> many Prometheus and Grafana related talks
13:34:18 <witek> I think we should better integrate with Prometheus in future
13:34:32 <witek> not necessarily as the fixed model for Monasca
13:34:45 <witek> but to enable integrating it
13:35:08 <witek> so that operators can build own alerting rules and existing community dashboards
13:35:09 <chaconpiza> SOCM 10 ? :D
13:35:21 <witek> there is no SOC 10
13:35:26 <witek> and won't be
13:35:30 <chaconpiza> just kidding
13:36:35 <witek> if we want to get any traction in the project we have to integrate with what people are using
13:37:23 <witek> and the simplest integration we could do is to use the Prometheus remote read
13:38:31 <witek> it doesn't change the general architecture and enables ops to use both Monasca and Prometheus
13:38:59 <witek> what is required is a minor change in time series message format
13:40:20 <Dobroslaw> would be good with backward compatibility so that old messages are working fine
13:41:37 <witek> we could have a configuration option to control that, or implement a migration script
13:42:07 <witek> one presentation I wanted to point to was from Rob Skillington
13:42:17 <witek> about M3DB
13:43:05 <witek> it's a scalable TSDB querying much faster then Prometheus
13:43:18 <Dobroslaw> looks nice
13:44:04 <witek> dougsz was mentioning this DB in the past as one of potential options for us
13:44:48 <witek> also here, they integrate seamlessly with Prometheus with remote read/write mechanism
13:45:36 <witek> one barrier for us, they only have a Go client
13:46:23 <witek> but I chatted shortly with Rob and they seem to have an undocumented support for InfluxDB line protocol
13:46:34 <Dobroslaw> oh, that's interesting
13:46:53 <witek> meaning that on the persister side we could probably just replace it with InfluxDB
13:47:20 <witek> still, we'd have to implement the query API
13:47:38 <witek> I want to follow up with Rob and ask more details
13:48:17 <witek> here the link to presentation:
13:48:24 <witek> https://fosdem.org/2020/schedule/event/m3db/
13:49:16 <witek> do we have any other topics?
13:51:01 <witek> if not, thanks for joining
13:51:12 <chaconpiza> thanks
13:51:22 <witek> thanks Martin for volunteering on benchmarking
13:51:28 <chaconpiza> np
13:51:39 <witek> see you next week, bye
13:51:47 <Dobroslaw> see you
13:51:49 <witek> #endmeeting