15:00:44 <rhochmuth> #startmeeting monasca
15:00:45 <openstack> Meeting started Wed Oct 12 15:00:44 2016 UTC and is due to finish in 60 minutes.  The chair is rhochmuth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:47 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:49 <rhochmuth> o/
15:00:49 <openstack> The meeting name has been set to 'monasca'
15:00:56 <rbak> o/
15:00:56 <koji> o/
15:01:00 <pratid> o/
15:01:01 <witek> hi
15:01:03 <hosanai> o/
15:01:03 <ddieterly> o/
15:01:07 <Kamil_> o/
15:01:11 <tomasztrebski> o/
15:01:13 <rhochmuth> https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:01:22 <rhochmuth> Agenda for Wednesday October 12, 2016 (15:00 UTC)
15:01:22 <rhochmuth> 1.	Reviews
15:01:24 <rhochmuth> 1.	https://bugs.launchpad.net/monasca/+bug/1625573
15:01:24 <openstack> Launchpad bug 1625573 in Monasca "monasca-setup does not support service_type & endpoint_type for Keystone Catalog" [Undecided,Fix committed] - Assigned to Pablo Rodriguez (pratid)
15:01:24 <rhochmuth> 2.	HA improvements for the openstack-ansible-os_monasca role and development of the openstack-ansible-os_monasca-agent role
15:01:24 <rhochmuth> 1.	https://blueprints.launchpad.net/openstack-ansible/+spec/monasca-ha
15:01:24 <rhochmuth> 2.	https://blueprints.launchpad.net/openstack-ansible/+spec/monasca-agent
15:01:33 <Fdaisuke> o/
15:01:38 <pratid> this is me
15:01:56 <rhochmuth> so, let's get started
15:02:19 <rhochmuth> i have a few topics i would like to cover that i didn't put up on the agenda to, mainly summit prep
15:02:24 <shinya_kwbt> o/
15:02:27 <rhochmuth> bet, we can cover the agenda first
15:02:41 <rhochmuth> #topic reviews
15:02:53 <rhochmuth> https://bugs.launchpad.net/monasca/+bug/1625573
15:02:53 <openstack> Launchpad bug 1625573 in Monasca "monasca-setup does not support service_type & endpoint_type for Keystone Catalog" [Undecided,Fix committed] - Assigned to Pablo Rodriguez (pratid)
15:03:08 <pratid> if you had time to read it
15:03:19 <pratid> there were some objections by joe keen
15:03:34 <pratid> but i replied
15:03:55 <pratid> i consider keystone client is not affected
15:04:04 <pratid> and fix to bug is ready
15:04:11 <pratid> what do you think?
15:04:28 <rhochmuth> i'll have to ping joe
15:04:38 <tomasztrebski> I was thinking on this, but did not manage to reply - however my idea was to consider using openstack's library shade - just a suggestion actually
15:04:40 <rhochmuth> i haven't spend enoguh time on this review
15:05:16 <rhochmuth> what does the shade library do?
15:05:44 <tomasztrebski> it's common interface for openstack, exposing core services in python binding
15:05:57 <tomasztrebski> i wanted just suggest considering it, because of that fact
15:06:12 <tomasztrebski> it is commonly used in openstack ansible modules
15:06:18 <tomasztrebski> AFAIK
15:06:23 <pratid> well, we could consider it
15:06:27 <pratid> in any case
15:06:35 <pratid> we should allow users
15:06:48 <pratid> to refine query to keystone catalog
15:06:56 <pratid> for the monasca_url
15:07:12 <pratid> and not always get the publicURL of service=monitoring
15:07:52 <tomasztrebski> maybe the idea of using direct url is kind of a bad here, for this you have a catalog and endpoints inside - plus different regions may contain different URLs for the same endpoint AFAIK
15:08:33 <pratid> catalog is actually been queried
15:08:42 <tomasztrebski> i'll be back in few minutes
15:08:48 <pratid> but only for publicURL, for instance
15:09:15 <pratid> when getting url from catalog
15:09:23 <pratid> we should consider up to 3 kwargs:
15:09:28 <pratid> service_type
15:09:30 <pratid> endpoint_type
15:09:36 <pratid> and region_name
15:09:47 <pratid> current implementation
15:09:59 <pratid> just ignores them
15:10:04 <pratid> and uses the default values
15:10:21 <pratid> our change solves the problem
15:10:33 <pratid> getting such optional values from agent.yaml
15:11:07 <rhochmuth> pratid: i think we lost tomasz for a few minutes
15:11:12 <pratid> ok
15:11:20 <rhochmuth> and i'm not prepared on this topic
15:11:24 <pratid> yes
15:11:33 <pratid> then we can continue discussion
15:11:41 <rhochmuth> but, i can take a closer look adn also request joe to get involved with this one
15:11:42 <pratid> within the review
15:11:48 <pratid> ok
15:11:51 <rhochmuth> sure
15:12:00 <rhochmuth> i think moving the comments would work well in the review too
15:12:11 <rhochmuth> it sounds like you would like to get this resolved
15:12:12 <pratid> then we could go on with the agenda
15:12:29 <rhochmuth> so i think it is on our radar to start paying attention to this review
15:13:00 <pratid> ok
15:13:11 <rhochmuth> thanks pratid
15:13:15 <rhochmuth> will move onto another topic
15:13:27 <rhochmuth> if you aren't getting enough progress on this, then please bring up again
15:13:34 <pratid> sure
15:13:52 <rhochmuth> hopefully, you'll get some comments in the review and we can make some progress
15:14:09 <rhochmuth> #topic HA improvements for the openstack-ansible-os_monasca role and development of the openstack-ansible-os_monasca-agent role
15:14:18 <rhochmuth> 1.	https://blueprints.launchpad.net/openstack-ansible/+spec/monasca-ha
15:14:18 <rhochmuth> 2.	https://blueprints.launchpad.net/openstack-ansible/+spec/monasca-agent
15:14:31 <rmelero> Hello I'm with mirantis and proposed those changes.
15:14:54 <rhochmuth> hi rmeiro
15:15:02 <rhochmuth> thanks for working on this
15:15:16 <rmelero> We're currently working on the spec and would like any feedback. I should have the full spec laid out by today.
15:15:32 <tomasztrebski> sorry, daughter alert
15:15:54 <tomasztrebski> cry notification method...you get what I am talking about ;-)
15:15:54 <rhochmuth> rmelero: ok, thanks for bging to our attention
15:16:13 <rhochmuth> are you developing kafka roles too
15:16:21 <rhochmuth> as well as influxdb?
15:16:26 <rmelero> One question i have is that influx .9 is going to be deprecated soon, and cassandra is not supported by grafana, any plans for other TS dbs?
15:16:29 <rmelero> Yes
15:16:51 <rhochmuth> monasca works with .10 too
15:17:01 <tomasztrebski> cassandra is now WIP in monasca
15:17:02 <rmelero> clustering support is not free on .10
15:17:18 <tomasztrebski> monasca-grafana-datasource communicates with api which knows how to talk with cassandra
15:17:22 <rhochmuth> i thought it was still free in .10
15:17:22 <tomasztrebski> but that's WIP
15:17:33 <rhochmuth> they had gone closed source on clustering in 0.11 i thought
15:17:41 <tomasztrebski> AFAIK it is still free in 10
15:17:51 <tomasztrebski> witek: don't we use influx .10 ?
15:18:15 <rmelero> ok i'll double check that I thought i saw that .9 was going to be last free cluster release
15:18:23 <witek> yes, we do
15:18:41 <rhochmuth> here is the link, https://www.influxdata.com/update-on-influxdb-clustering-high-availability-and-monetization/
15:18:42 <witek> I think 10 is the last free one
15:19:12 <witek> oh, even 11
15:19:18 <rhochmuth> Next week we’ll be cutting the first release candidate for the 0.11.0 release. While this release includes significant improvements in the query engine and the clustering code base, it will be the last open source version that includes clustering. In April, we’ll release 0.12.0, which will be fully open source and have great new features, but it will be focused on the standalone InfluxDB server. Both of these will be drop in repl
15:19:48 <rhochmuth> witek: agree
15:20:00 <rhochmuth> 0.10 and 0.11 should support clustering and still be open-sourced
15:20:11 <rhochmuth> 0.12 went closed source for clustering
15:20:21 <witek> rmelero: are you going to continue the work on openstack-ansible-os_monasca?
15:20:23 <rhochmuth> cassandra is a work in progress
15:20:51 <rhochmuth> so, please don't consider the cassandra driver production ready
15:21:01 <rmelero> yes I am going to extend it.
15:21:27 <tomasztrebski> nice ;-)
15:21:51 <rhochmuth> are you going to have ansible roles for each component in monasca?
15:22:37 <rmelero> does the role have to have single host only as well? I was thinking of adding some variables that if present, activate the clustering templating and deployment
15:23:06 <rmelero> I'll be basing alot of my work on the hp-cloud and bcom ansible playbooks that already do that, but came across problems implementing it on OSA
15:23:07 <rhochmuth> it would be nice to support single node and multi-node deployments
15:23:37 <rmelero> Ok, I'll keep that in mind.
15:24:02 <rhochmuth> also, i should point out that it is possible to use the non-clustered version of InfluxDB, but replicate all the data to stand-alone InlufxDB instances via Kafka
15:24:11 <tomasztrebski> taking monasca components into account only api's and thresh would be more difficult because of the need of load balancer and how the fact that for thresh this is the storm that's actually clustered
15:24:33 <tomasztrebski> for the rest this is rather straighforward
15:25:28 <rhochmuth> if you replicate the metrics to multiple stand-alone InfluxDB clusters, that would be the poor man's version of clustering
15:25:43 <rhochmuth> it woudl be better than deploying https://github.com/influxdata/influxdb-relay
15:25:52 <tomasztrebski> oh...forgot influx...that I am no export in telling anything ;D
15:26:09 <rmelero> Huh ok I'll look into that.
15:26:09 <rhochmuth> there is still a possiblity of inconsistent data between nodes
15:26:33 <rhochmuth> but, because we have Kafaka in the middle, that probability of inconsisten data shoudl be low
15:26:52 <rhochmuth> as long as the time to live on messages exceeds any down time
15:27:04 <rmelero> there might be some changes i'll have to make to monasca-agent as well because monasca-reconfigure doesn't correctly find the configuration files and doesn't configure services like apache or rabbit correctly
15:27:53 <tomasztrebski> for the agent I would look first if there's no way to supply those detections with custom arguments, there are some plugins where you could define that
15:28:09 <tomasztrebski> that wouldn't work in auto-mode, but manually it should
15:29:22 <rmelero> Ok I'll look into that as well. Also regarding containers, does anyone see a problem with sticking all the services in containers?
15:29:40 <rmelero> or should we is_metal all of them?
15:29:41 <tomasztrebski> did anyone try that
15:29:44 <tomasztrebski> ?
15:29:48 <rhochmuth> we are looking into that now ourselves
15:29:51 <rhochmuth> but i don't have any data
15:30:01 <tomasztrebski> I am aware of an attempt of monasca-agent docker image
15:30:03 <rhochmuth> the biggest issue is how to handle storage
15:30:18 <rhochmuth> yes, we actually had all of monasca running in a single container
15:30:26 <rhochmuth> monasca-docker repo
15:30:40 <rhochmuth> https://github.com/hpcloud-mon/monasca-docker
15:30:49 <rhochmuth> that was done for demo purposes, not production
15:30:58 <tomasztrebski> didn't that run ansible under hood ?
15:31:14 <rhochmuth> yes it did
15:31:21 <rhochmuth> to build the container
15:32:02 <rhochmuth> i think there are already containers avaialble fo zk, kafka, storm, influxdb, ...
15:32:48 <rhochmuth> for systems in whcih you are writing a lot of metrics, kafka and inclufxb would require access to direct storage
15:32:53 <rhochmuth> not network attached storage
15:32:55 <rmelero> from speaking with the OSA people, seems like everything should be defaulted to be in a container, and then move them out if they run into a problem
15:33:37 <rhochmuth> unfortunatley, we dont' have a lot of experience running in conatiners, so other than the one issue i've mentioned, there could be others
15:34:14 <rmelero> copy that. I'll have to experiment with it.
15:34:36 <rhochmuth> if i hear about any over the next couple of weeks i'll let you know
15:34:46 <rhochmuth> most of our work in this area is internal though
15:34:54 <rhochmuth> and investigation focused
15:36:19 <rhochmuth> shoudl we move onto other topics?
15:36:45 <rmelero> sounds good to me.
15:36:56 <rhochmuth> ok, thanks
15:37:05 <rhochmuth> we'll start watching your reviews
15:37:21 <rhochmuth> and comment on them, hopefully
15:37:32 <rhochmuth> but, it sounds like a great project
15:37:39 <rhochmuth> so, thanks for working on that
15:38:15 <rmelero> np monasca really needs ha support if its to be used as a production monitoring system
15:38:57 <tomasztrebski> I think that a thing worth considering here would be multi-node devstack deployment
15:39:06 <rhochmuth> rmelero: yeah, right now we have vertica and influxdb support, but obviousely when influxdb went closed source that was unexpected
15:39:10 <tomasztrebski> to have "community" example of that ?
15:39:28 <rhochmuth> so, cassandra is what we are working on to address that
15:39:46 <rhochmuth> but, there is still a lot of work
15:40:17 <rhochmuth> #topic open mic
15:40:45 <rhochmuth> so, i would like to talk about summit prep
15:41:01 <rhochmuth> this is the list i have so far
15:41:03 <rhochmuth> 1.	Monasca/Cassandra
15:41:03 <rhochmuth> 1.	https://blueprints.launchpad.net/monasca/+spec/cassandra-tasks
15:41:03 <rhochmuth> 2.	Locked alarms
15:41:03 <rhochmuth> 1.	https://blueprints.launchpad.net/monasca/+spec/locked-alarms
15:41:03 <rhochmuth> 3.	Logging
15:41:03 <rhochmuth> 4.	Properties/attributes for metrics/logs and sporadic metrics
15:41:04 <rhochmuth> 1.	https://blueprints.launchpad.net/monasca/+spec/publish-logs-to-topic-selectively
15:41:04 <rhochmuth> 5.	Monasca Analytics
15:41:05 <rhochmuth> 1.	Anomaly Detection
15:41:05 <rhochmuth> 2.	Alarm correlation/clustering
15:41:06 <rhochmuth> 6.	Growing Monasca in the OpenStack community and beyond
15:41:19 <rhochmuth> that didn't copy/paste correctly
15:41:25 <tomasztrebski> readable
15:41:48 <rhochmuth> we have 6 slots
15:42:11 <rhochmuth> if there are other topics please let me know
15:42:20 <rhochmuth> i just i should put that list in an etherpad
15:43:08 <witek> multitenancy for logging?
15:43:30 <rhochmuth> https://etherpad.openstack.org/p/monasca-barcelona-design-summit
15:43:42 <rhochmuth> witek: yes, that sounds like a good topic
15:43:42 <witek> what is 'growing monasca' about?
15:44:12 <tomasztrebski> aggressive expansion
15:44:16 <rhochmuth> i think we shoudl have a discussion about growing the eco-system around monasca
15:44:23 <rhochmuth> and how we can approach that
15:45:03 <rhochmuth> so, it isn't a technical topic, but an important one nonetheless
15:45:26 <rhochmuth> tomasztrebski: does that sound good to you?
15:45:51 <witek> rmelero: will you attend the summit?
15:46:03 <witek> perhaps we could discuss ansible
15:46:24 <rhochmuth> sure
15:46:32 <shinya_kwbt> It seems rmelero has left IRC
15:46:32 <tomasztrebski> rhochmuth: well, since more companies are getting involved and technologies too, it's good idea to consolidate here, though not sure what exactly would be covered, I mean in details
15:47:22 <rhochmuth> i agree
15:47:51 <rhochmuth> i would alos like to discuss linkages with other areas outside of openstack, such as containers, kuberntes and prometheus
15:48:46 <rhochmuth> if rmelero attends, then we could cover the ansible topic
15:48:53 <rhochmuth> in barcelona
15:49:35 <rhochmuth> So, please review the proposed agenda at, https://etherpad.openstack.org/p/monasca-barcelona-design-summit, and feel free to add more topics and details
15:49:47 <rhochmuth> i would like to wrap-up this next week
15:50:10 <rhochmuth> so, i can update the official schedule
15:51:34 <rhochmuth> shinya_kwbt: sorry i didn't get to your updates to the presentation
15:52:06 <rhochmuth> i will start reviewing and contributing hopefully today, but tomorrow and firday i've blocked out to work on the summit session
15:52:07 <shinya_kwbt> you mean you cant see presentation?
15:52:16 <rhochmuth> i can see it
15:52:35 <rhochmuth> i just didn't have time to review/comment on it
15:52:44 <shinya_kwbt> No problem
15:53:04 <rhochmuth> thanks for working on it
15:53:09 <rhochmuth> you are ahead of me
15:53:12 <witek> rhochmuth: shinya_kwbt: it would be good to schedule the call for next week to sync
15:53:23 <rhochmuth> witek: yes, agree
15:53:28 <rhochmuth> how about monday
15:53:32 <shinya_kwbt> me too.
15:53:51 <rhochmuth> we have a cassandra discussion
15:53:57 <rhochmuth> on monday
15:54:02 <witek> yes
15:54:19 <rhochmuth> i could meet after
15:54:22 <rhochmuth> or before
15:54:28 <witek> good for me
15:54:39 <shinya_kwbt> ok for me
15:54:45 <rhochmuth> ok, so let's do a review after the cassandra discussion
15:55:04 <rhochmuth> is that going to be too late for you
15:55:17 <shinya_kwbt> No problem
15:55:24 <rhochmuth> sake
15:55:28 <witek> :)
15:55:56 <tomasztrebski> btw, can someone take another look at xenial change, been trying to test but failed with big bang ;-(
15:56:01 <shinya_kwbt> ha ha
15:56:15 <rhochmuth> tomasztrebski: sure i can look at it again
15:56:24 <rhochmuth> tomasztrebski and kamil: how is bootcamp going
15:56:32 <rhochmuth> should i be reviewing and contributing
15:56:42 <Kamil_> a lot to do
15:56:44 <shinya_kwbt> https://review.openstack.org/#/c/366024/
15:56:45 <tomasztrebski> rhochmuth: take a look at presentation ;-)
15:56:52 <rhochmuth> will do
15:56:58 <Kamil_> but i'ts going forward
15:57:11 <rhochmuth> thanks guys
15:57:15 <shinya_kwbt> tomasztrebski: thanks
15:57:36 <witek> poll: how much RAM your laptop has?
15:57:43 <rhochmuth> 16 GB
15:57:50 <Kamil_> I want to send attendees information in this week.
15:58:11 <witek> shinya_kwbt?
15:59:01 <shinya_kwbt> witek: tomasztrebski is reviewing xenial change
15:59:31 <rhochmuth> thanks everyone, i got to close the meeting down
15:59:47 <rhochmuth> bye-bye
15:59:51 <pratid> bye
15:59:53 <witek> thanks, bye
15:59:55 <shinya_kwbt> thanks bye
15:59:57 <ddieterly> ciao!
16:00:08 <rhochmuth> #endmeeting