14:01:54 <dtardivel> #startmeeting watcher
14:01:55 <openstack> Meeting started Wed Dec 16 14:01:54 2015 UTC and is due to finish in 60 minutes.  The chair is dtardivel. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:56 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:59 <openstack> The meeting name has been set to 'watcher'
14:02:11 <jwcroppe> hi
14:02:16 <sballe> o/ I won;t be able to join :-( I am hosting a meeting
14:02:22 <jed56> hello
14:02:34 <jwcroppe> o/
14:02:34 <brunograz> hi
14:02:39 <dtardivel> o/
14:02:42 <tpeoples> o/
14:02:51 <cdupont> o/
14:02:55 <edleafe> \o
14:03:10 <cdupont> \o/
14:05:09 <dtardivel> agenda is available on #link https://wiki.openstack.org/wiki/Watcher_Meeting_Agenda#12.2F16.2F2015
14:06:03 <dtardivel> ok let's go :)
14:06:16 <dtardivel> #topic Announcements
14:06:56 <cdupont> Hi dtardivel, I have some requests for this meeting :)
14:07:17 <cdupont> Could we store more informations in the minutes using #info ?
14:07:39 <dtardivel> reminder: Mid-cycle eventbrite registration is opened
14:07:46 <cdupont> because after each meeting I am obliged to browse the full log to create my report
14:08:04 <dtardivel> cdupont: yes i will try
14:08:11 <cdupont> thanks!
14:08:35 <dtardivel> #info Mid-cycle eventbrite registration is opened
14:08:42 <jwcroppe> dtardivel: did acabot send on mailing list?
14:08:45 <cdupont> Also I was wondering if this meeting could be longer, one hour is short in a launch phase like now
14:09:20 <cdupont> it's always very rushed and no time for questions...
14:09:25 <edleafe> cdupont: meetings are standard at 1 hour. We can always continue in #openstack-watcher
14:09:28 <dtardivel> for b<>com, acabot and jed56 will go to the meeting
14:10:04 <jed56> jwcroppe I don't think so
14:10:22 <dtardivel> for bcom, acabot and jed56 will go to the meeting
14:10:32 <jwcroppe> jed56: ok, I think we need to send out on mailing list - I updated the wiki with details as well and he also created the eventbrite
14:10:56 <jed56> great job thanks
14:11:26 <jwcroppe> #link https://wiki.openstack.org/wiki/Watcher_mitaka_mid-cycle_meetup_agenda
14:11:26 <cdupont> this is April 25 - 29, 2016 in Austin, right
14:11:32 <cdupont> ?
14:11:40 <jwcroppe> cdupont: Feb 2-4
14:12:06 <jed56> I have my ticket!
14:12:08 <tpeoples> #info mid-cycle meetup in Austin Feb 2-4, 2016
14:12:24 <cdupont> Ok
14:13:38 <dtardivel> #action dtardivel send (again ?) the announce to the mailing list
14:14:23 <dtardivel> next: we released the mitaka-1 milestone on Watcher
14:14:33 <cdupont> Could you circulate again the mailing list address?
14:15:18 <cdupont> its not in the wiki
14:15:37 <dtardivel> all fixes committed and targeted into this milestone have been automatically set to 'fix released' state
14:15:38 <tpeoples> openstack-dev
14:16:05 <jed56> #link http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
14:16:10 <tpeoples> with a [watcher] tag
14:16:12 <cdupont> OK, there is no mailing list specific to watcher?
14:16:21 <cdupont> this is a very high volume mailing :(
14:17:53 <dtardivel> cdupont: you can use the tag as a filter ;)
14:18:05 <tpeoples> No, that's just how the openstack projects do it
14:18:44 <seanmurphy> #info - announcements pertaining to watcher typically distributed on openstack-dev list with [watcher] as tag
14:19:03 <cdupont> I see them as [openstack-dev] [watcher]
14:19:25 <cdupont> I'll create a filter ;)
14:19:55 <jwcroppe> next topic?
14:20:32 <dtardivel> acabot has posted a video about Watcher UI #link https://drive.google.com/drive/u/0/folders/0B1KilbTYF00uanZCRE9kQmgzQ28
14:20:49 <dtardivel> Please check if you can access to it
14:21:05 <cdupont> nope
14:21:23 <jwcroppe> no access
14:21:28 <tpeoples> Looks like we'll have to wait for acabot
14:21:34 <cdupont> :(
14:21:53 <seanmurphy> the link appears to be a user specific gdrive link
14:22:52 <dtardivel> yes I afraid. ok I will find a way to put the video on another location ...
14:23:05 <vmahe__> I can't read the video either
14:23:40 <edleafe> *cough*youtube*cough*
14:23:50 <dtardivel> #action dtardivel give access to video Watcher UI to the community
14:24:32 <dtardivel> #topic Review Action Items
14:24:34 <tpeoples> Maybe there's some sweet copyrighted music on it edleafe
14:25:06 <tpeoples> dtardivel: did you have a chance to give the devstack plugin another shot?
14:25:17 <edleafe> tpeoples: well, that wouldn't be in the open spirit now, would it? :)
14:25:48 <jed56> :)
14:26:29 <dtardivel> tpeoples: I validated the devstack installation with the wather plugin. I need more time to validate the NFS settings for live migration as documented in the patchset
14:26:41 <tpeoples> cool, good to hear
14:26:51 <dtardivel> so it is in good progress
14:27:25 <dtardivel> review on i18n translation. What is the status ?
14:27:29 <tpeoples> i will try to check my email over the holiday break to see if we need a new patchset
14:27:59 <vincentfrancoise> Well the first patchset has been merged
14:28:05 <brunograz_> tpeoples: in a related topic, we've been working on putting Watcher in a container. this is very much a work in progress and I believe that by beginning of next year we have it done
14:28:13 <vincentfrancoise> which added the french locale to Watcher
14:28:32 <vincentfrancoise> #link https://review.openstack.org/#/c/256591/
14:28:44 <tpeoples> vincentfrancoise: i will re-review before i leave for vacation
14:28:48 <jwcroppe> dtardivel: FYI, I need to drop in 2 mins... I will review the pending watcher-specs as well
14:29:06 <vincentfrancoise> And there is another one in waiting for reviews
14:29:24 <vincentfrancoise> #link https://review.openstack.org/#/c/257475/
14:29:29 <tpeoples> brunograz_: I believe bcom already has a container-based deployment of watcher working
14:29:35 <dtardivel> brunograz_: we have project on github named watcher-tools where you will find a complete setup of watcher based on Docker container and ansible :)
14:29:47 <tpeoples> let's save that for open discussion
14:29:56 <dtardivel> #link https://github.com/b-com/watcher-tools
14:30:46 <tpeoples> vincentfrancoise:  i am not too worried about the content of the log messages for the i18n work, just the way they are being translated.  we can worry about the content later... jfyi
14:31:05 <brunograz_> tpeoples dtardivel that's pretty cool, it would be good to have it as well in the github description. but yeah true let's leave it for the right topic :)
14:31:57 <dtardivel> there is again some review on progress about refactoring #link https://review.openstack.org/257249
14:32:07 <vincentfrancoise> tpeoples: you mean the actual messages that are currently logged?
14:33:05 <tpeoples> Yeah, I'm saying we can improve the actual log strings later, but for the i18n work let's just get them translatable
14:33:30 <vincentfrancoise> For the refactoring bit, most of the work has already been merged
14:33:47 <tpeoples> i will review 257249 again before i leave jed56, thanks for respinning
14:34:24 <vincentfrancoise> #link https://review.openstack.org/#/c/257249/ is the only one remaing
14:34:30 <jed56> tpeoples thanks
14:36:43 <jed56> next topic ?
14:36:44 <tpeoples> wrt the intel thermal PoC - I think that needs some rework before we can merge.  I identified some concerns I had with the approach in the review
14:38:08 <tpeoples> though the concerns I had might also apply to the consolidation algorithm.  I think we need to be using the nova-scheduler to help make decisions on where to migrate to.  Otherwise we lose the benefit of all of the other filters
14:38:35 <tpeoples> so for a PoC it's fine not to do that, but IMO it should be
14:38:44 <jed56> tpeoples : I agree
14:39:14 <cdupont> are you refering to https://blueprints.launchpad.net/watcher/+spec/outlet-temperature-based-strategy ?
14:39:28 <jwcroppe> tpeoples: right, scheduler for sure needs to be involved
14:39:37 <tpeoples> yes cdupont, review is here #https://review.openstack.org/#/c/257189
14:40:26 <tpeoples> i also don't like that the resources used is being calculated on the fly (instead of relying on the scheduler ... or some sort of cache within watcher) to get that info
14:40:30 <cdupont> Is it possible to link the BP to its spec, even if in progress?
14:41:05 <cdupont> I noticed the spec field is never set, for all the BPs I checked
14:41:33 <tpeoples> that sort of thing ^ won't scale very well
14:41:43 <tpeoples> but again, fine for PoC...
14:42:03 <dtardivel> cdupont: you can push a first patchset , and set it into Work In Progress (ie workflow = -1)
14:42:14 <cdupont> Ok
14:42:35 <jed56> tpeoples : I think that can be done through a blueprint ( cache system for nova and ceilometer )
14:43:04 <dtardivel> vmahe__, what about your review on dynamic action #link https://review.openstack.org/#/c/257494/
14:43:07 <tpeoples> #action junjie bzhou incorporate review comments into https://review.openstack.org/#/c/257189
14:43:14 <tpeoples> yes jed56 , probably
14:43:32 <vmahe__> I will push a new patch-set very soon
14:43:35 <tpeoples> don't want to tell the intel folks they need to solve all of our problems :)
14:44:10 <tpeoples> that's why i said good for a proof of concept
14:44:12 <vmahe__> following some discussion with jed56 and vincentfrancoise
14:44:43 <dtardivel> ok
14:44:45 <tpeoples> have you guys had a chance to poke into taskflow any more?
14:44:58 <jed56> tpeoples : yes a little bi t
14:45:03 <vmahe__> yes
14:45:11 <vmahe__> it looks great
14:45:20 <jed56> this a great tool but some feature are missing
14:45:41 <vmahe__> but we need to also have a look to other frameworks such as Mistral or other Python frameworks
14:45:57 <tpeoples> jed56: what features and are they deal breakers?
14:47:09 <jed56> for example you can describe your Flow but you cannot describe some constraints to be solve by the workflow engine ( e.g A need to before B )
14:47:58 <jed56> you have to describe your workflow by yourself
14:48:26 <dtardivel> remains 10mn :)
14:48:27 <jed56> morever there is not a clear separation of concern between the planner and applier
14:48:28 <dtardivel> #topic Blueprint/Bug Review and Discussion
14:48:29 <tpeoples> i imagined we'd have to build the workflow ourself.
14:48:58 <jed56> tpeoples : yes but with vmahe
14:49:16 <tpeoples> the planner would essentially build the flow with all of the dependencies, and then the applier would just run the flow using an engine
14:49:19 <jed56> we think that some tools can do to that built in
14:49:42 <jed56> but this is WiP
14:49:43 <jed56> :)
14:49:57 <jed56> I have a PoC with taskflow
14:50:05 <tpeoples> ok cool
14:50:51 <tpeoples> wrt the tempest blueprint / work (https://blueprints.launchpad.net/watcher/+spec/tempest-basic-set-up), i passed it to vincentfrancoise  while I am out on vacation
14:51:19 <vincentfrancoise> yup
14:51:23 <cdupont> regarding the BP https://blueprints.launchpad.net/watcher/+spec/watcher-consolidation-grammar
14:51:28 <dtardivel> about tempest basic set up #link https://blueprints.launchpad.net/watcher/+spec/tempest-basic-set-up, vincentfrancoise will be able to work on it very soon
14:51:39 <cdupont> I completed a first draft
14:51:59 <cdupont> I will upload as soon as I understand how gerrit works :)
14:52:10 <jed56> cdupont : great  can you push a patchset ?
14:52:16 <cdupont> yep
14:52:35 <jed56> cdupont : we can help you on openstack-watcher
14:52:42 <cdupont> OK
14:52:47 <cdupont> thanks
14:52:55 <dtardivel> cdupont: we can assist you if you need help
14:53:44 <jed56> cdupont : http://docs.openstack.org/infra/manual/sandbox.html
14:54:38 <cdupont> #action cdupont  push a patchset for spec https://blueprints.launchpad.net/watcher/+spec/watcher-consolidation-grammar
14:54:48 <tpeoples> any other blueprints or bugs to discuss?
14:54:49 <dtardivel> I think we have to make effort on spec in the few weeks, because refactoring and enhancement coding tasks will be finished soon...
14:55:12 <dtardivel> and we need spec to make coding :)
14:55:28 <jed56> in the few weeks : vacations!
14:55:41 <cdupont> yeah
14:55:41 <vincentfrancoise> as you may have noticed, Watcher is now python 3.4 compatible
14:56:08 <dtardivel> vincentfrancoise: +1
14:56:26 <vincentfrancoise> So we have to make sure our tests pass both py27 and py34 gates ;)
14:56:29 <tpeoples> sweet
14:56:49 <tpeoples> open discussion quick?
14:57:19 <dtardivel> brunograz_: containers ?
14:57:48 <brunograz_> dtardivel: just wanted to pointed out that would be good to have a link in the git repo
14:58:03 <brunograz_> but we can follow up this in the watcher chat
14:58:27 <vincentfrancoise> brunograz_: you mean in the readme file?
14:58:37 <tpeoples> in regards to accommodations for Austin for the midcycle - a lot of people who visit IBM stay at hotels near the Arboretum.  I'm not sure where jwcroppe is staying specifically.
14:58:48 <brunograz_> vincentfrancoise: yes
14:59:07 <tpeoples> The Domain is literally across the street from IBM so you'd be able to walk, but the hotels there will be more expensive
14:59:16 <brunograz_> vincentfrancoise: particularly, I didn;t know abt this other repo
14:59:20 <tpeoples> Feel free to message me with any questions
14:59:22 <jed56> tpeoples : ok
14:59:48 <dtardivel> ok, it's time
14:59:49 <dtardivel> bye
14:59:52 <brunograz_> bye
14:59:53 <vincentfrancoise> brunograz_: create a wishlist bug and feel free to do it I guess
14:59:54 <vmahe__> bye
14:59:56 <vincentfrancoise> bye
14:59:57 <tpeoples> And I am out on vacation until next year, so ttyl :)
15:00:01 <dtardivel> #endmeeting