14:01:04 <acabot_> #startmeeting watcher
14:01:05 <openstack> Meeting started Wed Nov 18 14:01:04 2015 UTC and is due to finish in 60 minutes.  The chair is acabot_. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:07 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:10 <openstack> The meeting name has been set to 'watcher'
14:01:13 <acabot_> hi
14:01:18 <bzhou> hi
14:01:20 <alexstav> o/
14:01:27 <alexchadin> o/
14:01:27 <jwcroppe_> o/
14:01:34 <tpeoples> o/
14:01:35 <samarinvv> hi
14:01:53 <acabot_> our agenda for today #link https://wiki.openstack.org/wiki/Watcher_Meeting_Agenda#11.2F18.2F2015
14:01:56 <sudipto> o/
14:02:05 <jed56> hi
14:02:09 <sballe> o/
14:02:14 <brunograz> o/
14:02:15 <acabot_> feel free to add anything you like to discuss
14:02:53 <acabot_> is Nishi here ?
14:03:37 <acabot_> lets move on to our first topic
14:03:43 <acabot_> #topic annoucements
14:04:38 <acabot_> as mentioned in the agenda, we have 2 ad hoc meetings planned regarding the POC and academic contributions
14:04:57 <acabot_> any other announcement ?
14:05:18 <sballe> I am still working with BU in the meetup
14:05:27 <sballe> I should know tomorrow if they can host it.
14:05:33 <acabot_> ok great
14:05:38 <jwcroppe_> cool
14:05:42 <sballe> They have a workshp tomorrow and that has taken all the nergy they have
14:05:44 <acabot_> can you remind us the dates ?
14:05:57 <sballe> one sec
14:06:31 <alexchadin> Hello from Moscow, Servionica. We are working on rst-specs for overload-underload algorithms.
14:07:03 <sballe> 1/20-21 or 2/3-4 0r 1/26-27
14:07:06 <acabot_> alexchadin : great thanks, please send a public review as soon as you can
14:07:20 <sballe> alexchadin: +1
14:07:32 <alexchadin> Where should we publish it?
14:07:39 <alexchadin> launchpad?
14:08:13 <jwcroppe_> review.openstack.org (watcher-specs project) and launchpad
14:08:46 <alexchadin> jwcroppe_: yeap, thanks!
14:08:57 <sballe> here is an example: https://review.openstack.org/#/c/246363/
14:09:24 <acabot_> in /specs/mitaka/approved/watcher-overload-underload.rst
14:09:47 <sballe> acabot_: approved?
14:10:11 <sballe> Does it have to be reviewed before it is approved to move in "under approve"?
14:10:26 <jwcroppe_> yes, approved ... it won't be in repo until it's merged
14:10:31 <nishi> what has been approves
14:10:35 <acabot_> sable : yes sorry we only have a "implemented" folder
14:10:53 <jwcroppe_> sballe: so it's implicitly approved if we merge it :)
14:10:56 <sballe> nishi: our poc is approved to move n
14:11:04 <bzhou> merged means approved:-)
14:11:06 <nishi> ok
14:11:08 <acabot_> alexchadin : can we say your proposition is already "implemented" ?
14:11:17 <Kevin_Zheng> sorry Im late
14:11:41 <bzhou> welcome Kevin
14:11:42 <tpeoples> acabot_:  if it's not integrated into watcher i don't think it's implemented
14:11:59 <acabot_> ok so lets do it in "approved" and we will merge it in the future
14:12:18 <tpeoples> +1
14:12:23 <sballe> bzhou: I understand. I just wanted to make sure it was in line with the watcher pylosophy around being just a plug-in before it is approved.
14:12:24 <jwcroppe_> +1
14:12:27 <alexchadin> acabot_: implemented in watcher?
14:13:09 <acabot_> alexchadin : forgive it ;)
14:13:30 <sballe> alexchadin: your stuff will have to be integrated inthe watcher framework. right?
14:13:56 <sballe> I am looking to seeing the specs that describe how everything fits within watcher
14:14:08 <acabot_> sballe : +1
14:14:26 <acabot_> #topic review action items
14:15:15 <acabot_> jed56 : can you give us an update regarding telemetry integration ?
14:15:21 <alexstav> sballe: For now it works with OpenStack Nova Compute, but we are planing to integrate in with watcher. The problem is to get stats from compute-nodes, but it is another watcher-blueprint.from
14:15:22 <bzhou> A general comment for Telemetry integration:
14:15:22 <bzhou> The code is going to remove datasource (concept) and influxdb store and ONLY depends on ceilometer to provide meters.
14:15:38 <jed56> yes
14:15:54 <jed56> we have a first version we need to check that everying is working
14:15:54 <sballe> +1
14:16:15 <jed56> we also start cleanup a bit the code
14:16:18 <sballe> jed56: one you have checked it let bzhou and junjie know
14:16:37 <sballe> mann too many typos this morning. I need more coffee
14:16:42 <jed56> sballe: yes I hope before the end of the week
14:16:49 <sballe> perfect
14:17:06 <bzhou> jed56: is there any background that we remove datasource and influxdb as store?
14:17:07 <jed56> bzhou : yes exaclty
14:17:34 <bzhou> I'm not sure if we could get every meters only from ceilometer
14:17:49 <jed56> yes but we can custom metrics in ceilometer
14:17:53 <jed56> add
14:17:59 <sballe> alexstav: we cannot merge your code if it doesn't fit in the wathcer framework which is why I am insisting on that point. But I am sure you understand that
14:18:09 <acabot_> telemetry v2 allows us to create new metrics and store them
14:18:24 <sballe> +1
14:18:44 <jed56> we don't to maintain influxdb store
14:18:57 <jed56> because ceilosca project and ceilometer
14:19:18 <bzhou> so are we also going to drop watcher-metering-drivers?
14:19:39 <sballe> jed56: next cycle I would like for us to add support for Monasca as the source of telemetry too
14:19:39 <jed56> we can still use metering driver to push metrics into ceilometer
14:19:53 <jed56> but we have a overlap with the project monasca
14:19:56 <alexchadin> sballe: we are going to make patch with our spec-improvements
14:19:57 <jed56> an
14:20:36 <jed56> I think we don't to focus our work in the monitoring
14:20:38 <acabot_> bzhou : this is something we will keep to inject additional metrics in telemetry but for devstack support, we need to give up our metering chain
14:20:45 <sballe> alexchadin: will you contribute to the watcher framework too?
14:20:52 <jed56> want
14:21:05 <alexchadin> sballe: yes, we will
14:21:17 <sballe> alexchadin: great!
14:21:47 <jed56> sballe : this the idea to support Monasca and Ceilometer
14:22:27 <sballe> Monasca will eventually take over IMHO
14:22:35 <acabot_> a spec is available for review https://review.openstack.org/#/c/246363/
14:22:37 <sballe> But not today
14:22:38 <alexchadin> sballe: Before we do that, we would like to make sure our specs have been approved by you:)
14:22:47 <sballe> :)
14:23:01 <bzhou> acabot_: why for devstack support, we need to give up our metering chain? Because watcher-metering-drivers is not managed by openstack?
14:23:55 <sballe> bzhou: we also thought that Ceilometer was the way to go for devstack since it is an openstak project and already part od devstack
14:24:17 <acabot_> bzhou : because deploying Watcher today is too complex and we need to provide a quick install running on top of telemetry (available in devstack)
14:25:09 <jwcroppe_> acabot_: +1
14:25:09 <acabot_> bzhou : watcher-metering-drivers will never be maintained by OpenStack as Ceilometer agent exists already
14:25:14 <sballe> I am assuming if you want you can still use the watcher-metering-drivers with devstack but on a do it yourself basis
14:25:16 <bzhou> So the publisher is also going to be dropped, right?
14:25:21 <Kevin_Zheng> I agree, we it will be better to focus on the main task
14:26:18 <Kevin_Zheng> We can make use of the existing project
14:26:26 <bzhou> just want to make sure I understand the architecture change
14:26:26 <tpeoples> speaking of devstack, i have finally freed up from my ibm work and can start working on the devstack plugin shell script
14:26:28 <jed56> bzhou : we can still use it to push it in ceilometer for specific metrics like kwapi
14:26:34 <acabot_> bzhou : again we won't drop it, we will simply add Telemetry integration as a default solution for collecting metrics
14:27:50 <acabot_> bzhou : the idea is to have Telemetry working as a "common framework" for metrics collection
14:28:13 <bzhou> acbot_: got it, thanks. Will the project watcher-metering be maintained by OpenStack?
14:28:46 <acabot_> bzhou : watcher-metering is maintained by b-com (on our own github repo)
14:29:03 <sballe> I am assuming that if wathcer needs it we can fold it in under watcher
14:29:12 <sballe> acabot_: ^^^
14:29:14 <tpeoples> +1
14:29:26 <acabot_> sable : of course
14:30:02 <acabot_> please review the proposed code here https://review.openstack.org/#/c/246363/
14:30:50 <sballe> this is the spec.
14:30:50 <acabot_> next topic is loading optimization strategies into watcher
14:31:01 <bzhou> acabot_: thanks. So we may need to change the architecture doc:-)
14:31:21 <sballe> +1
14:31:28 <acabot_> sorry, here is the link https://review.openstack.org/#/c/246534/
14:31:43 <sballe> would it make sense to start a new arch doc kind of a v2?
14:31:51 <acabot_> bzhou : yes vamp is working on it
14:32:04 <acabot_> bzhou : vamp sorry
14:32:05 <bzhou> great
14:32:49 <acabot_> sable : a new wiki page ?
14:33:02 * edleafe joins late
14:33:03 <sballe> my nick is sballe
14:33:28 <sballe> acabot_: ^^ FYI
14:33:41 <sballe> yes a new wiki page would work
14:34:21 <sballe> I like to keep the history of the architecture and how it evolves
14:34:30 <vmahe_> FYI vamp is vmahe :-)
14:34:35 <sballe> :-)
14:34:41 <bzhou> :-)
14:34:50 <acabot_> #action acabot_ start a new wiki page for Watcher architecture v2
14:35:07 <sballe> acabot_: get all our nicks wrng this morning. he needs more coffee'
14:35:34 <jed56> lol
14:35:35 <jwcroppe_> As we discussed, we want Watcher to be able to be fed custom metrics - but the reference arch should support basic telemetry/ceilometer.   Agree?
14:35:41 <acabot_> sorry for that, my IRC client is cleaning all your nicks after sending the message...
14:36:06 <acabot_> jwcroppe_ : agree
14:36:13 <sballe> +1
14:36:16 <Kevin_Zheng> +1
14:36:20 <bzhou> +1
14:36:23 <vmahe_> +1
14:36:29 <nishi> +1
14:36:30 <jwcroppe_> cool
14:36:35 <jed56> +2
14:36:35 <edleafe> +1
14:36:35 <bzhou> but one question: we need to define a framework
14:36:45 <sballe> bzhou: +1
14:37:28 <sballe> My thoughts were taht whatever is in Ceilomter can be used.
14:37:35 <acabot_> bzhou : not sure to understand, what do you mean by a framework ?
14:37:52 <sballe> but we do need a way to bring extra telemetry in
14:38:01 <bzhou> I mean people can use it to easily feed custom metrics
14:38:09 <bzhou> maybe not from Telemtry
14:38:36 <sballe> so are you saying we need to keep the wathcer metric agrent for that use case?
14:38:36 <acabot_> bzhou : to my understanding with Telemetry v2 API, anyone can add any custom metrics in Telemetry
14:38:48 <jed56> bzhou : you can use ceilometer to add custom metrics or
14:39:00 <jed56> monasca
14:39:15 <acabot_> bzhou : so except if you dont want to use Telemetry, you will never need our Watcher metering chain
14:40:01 <bzhou> agree
14:40:20 <acabot_> ok lets move on loading strategies in Watcher
14:40:27 <jed56> exactly
14:40:29 <dtardivel> acabot_ +1
14:41:09 <sballe> +1
14:41:22 <acabot_> the review is here https://review.openstack.org/#/c/245903/
14:41:32 <acabot_> I will add it to the agenda
14:42:18 <Kevin_Zheng> Looks cool
14:42:31 <vmahe_> FYI : we'll provide more details regarding the integration of Telemetry v2 in the associated spec and the impacts on the current Watcher components
14:42:46 <sballe> +1
14:43:19 <vmahe_> See https://review.openstack.org/#/c/246363
14:44:03 <acabot_> thanks vmahe_, can you jump on documentation topic ?
14:44:30 <vmahe_> all right
14:44:50 <vincentfrancoise> So regarding the loading of new strategies
14:45:09 <vmahe_> I'm currently writing a Glossary to make sure that every one speaks the same language
14:45:09 <vincentfrancoise> it's still in progress
14:45:26 <vincentfrancoise> waiting for proper testing
14:45:29 <sballe> vincentfrancoise: I am assumign doc ill be added too
14:46:16 <vincentfrancoise> sballe: Yes, there's a plugins.rst added in this changeset explaining how to implement a new strategy
14:46:17 <sballe> I believe we will be testing it. taht was one of the actinos items we wee given at the poc meeting.
14:46:46 <sballe> can you ping us when it is ready to be tested?
14:46:50 <bzhou> sballe:+1
14:47:13 <vincentfrancoise> In parrallel, we are thinking about adding a gloassary to the documentation to lay down the right terminology to use throughout the code / doc
14:47:30 <vincentfrancoise> see https://review.openstack.org/#/c/246370/
14:48:04 <vmahe_> We need to mae sure everyone agrees on words such as "Action", "Meta-Action", ...
14:48:13 <tpeoples> i will take a look at that vincentfrancoise
14:48:40 <vmahe_> I'm currently reviewing it and will push a new patch set very soon
14:49:39 <vmahe_> I also did some testing with the tool used by the Nova team to generate some architecture diagrams
14:49:55 <vmahe_> Which is named Dia
14:50:24 <vmahe_> they have put in the repo a source directory for the diagrams which is used to generate some SVG or PNG images
14:50:59 <vmahe_> The issue with Dia is that it is not human readable source code.
14:51:21 <vmahe_> I need to do some testing also with PlantUML (which is also free)
14:51:46 <vmahe_> and makes it possible to do any UML diagrams : class diagrams, sequence diagrams, ...
14:52:02 <sballe> sounds interesting
14:52:12 <acabot_> sballe : you plan to submit a spec for energy minimization POC ?
14:52:55 <sballe> acabot_: yes Iwill be working with nishi, bzhou and the team on it. After yesterday's meeting we just needed to do some last minute adjustement to the poc
14:53:22 <acabot_> sballe : great, please fill it in approved folder for review
14:53:23 <sballe> I'll work on it as soon as I am back from SC'15
14:53:29 <vmahe_> there is an extension for Sphynx which makes it possible to include PlantUML source code directly in the RST files : https://pypi.python.org/pypi/sphinxcontrib-plantuml
14:54:00 <acabot_> #topic Blueprint/Bug Review and Discussion
14:54:03 <nishi> what do we need to fill
14:54:38 <acabot_> nishi : you need to feel a spec file based on templates availble on watcher-specs
14:54:47 <acabot_> fill not feel ;-)
14:54:48 <nishi> ok
14:55:11 <acabot_> regarding blueprints
14:55:32 <acabot_> vmahe_ has submitted a new one https://blueprints.launchpad.net/watcher/+spec/watcher-add-actions-via-conf
14:55:34 <sballe> nishi: https://github.com/openstack/watcher-specs
14:55:41 <sballe> I'll work with you
14:56:14 <nishi> great
14:56:32 <vmahe_> we need to agree on the mechanism for adding new actions to Watcher
14:56:38 <nishi> when do you come back from SC
14:56:47 <sballe> tomorow
14:56:50 <nishi> ok
14:57:06 <bzhou> nishi: I could help too if you need
14:57:45 <nishi> ook
14:57:50 <Kevin_Zheng> I'm also available :)
14:57:57 <sballe> lol
14:58:05 <acabot_> jwcroppe_ sballe : can you look at this BP before next meeting to fix priority ?
14:58:41 <acabot_> #action jwcroppe_ sballe acabot_ set priority for BP https://blueprints.launchpad.net/watcher/+spec/watcher-add-actions-via-conf
14:58:56 <acabot_> sorry but we wont have time for open discussions today
14:58:56 <bzhou> acabot_: Kevin_Zheng wants to work on bp: https://blueprints.launchpad.net/watcher/+spec/make-watcher-use-graduated-oslo-lib
14:59:10 <jwcroppe_> will do
14:59:26 <acabot_> bzhou : right there is review in progress on it, I will set it as started
14:59:32 <sballe> acabot_: will do
14:59:49 <bzhou> thanks
15:00:08 <acabot_> bzhou : https://review.openstack.org/#/c/246153/ right ?
15:00:19 <Kevin_Zheng> acabot: thanks
15:00:36 <acabot_> thank you guys for your time
15:00:39 <sballe> we are out of time
15:00:44 <acabot_> #endmeeting