13:57:25 <uli-k> #startmeeting Octopus weekly meeting
13:57:25 <collabot> Meeting started Mon Jun 15 13:57:25 2015 UTC.  The chair is uli-k. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:57:25 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
13:57:25 <collabot> The meeting name has been set to 'octopus_weekly_meeting'
13:57:36 <uli-k> #topic roll call
13:58:02 <uli-k> Hi everybody, please type #info if you are here to attend the octopus meeting
13:58:09 <uli-k> #info Uli
13:58:31 <fdegir> #info Fatih Degirmenci
13:58:36 <zhifeng> #info zhifeng
13:59:07 <MR_Sandvine> #info Manuel Rebellon
13:59:13 <radez> #info Dan Radez
13:59:45 <MatthewLi> #info Jun Li
14:00:24 <uli-k> Great to see also new names! We still need more help in octopus :D
14:00:32 <uli-k> So let's start.
14:00:39 <uli-k> #topic Agenda bashing
14:00:44 <uli-k> I have:
14:00:48 <dneary> Morning
14:01:03 <uli-k> Morning Dave!
14:01:13 <uli-k> Agenda: Action item review / Arno maintenance - High prio Jira issues / Starting work on B-Release / AoB
14:01:24 <uli-k> Anything else you want to discuss?
14:02:27 <uli-k> ok. Agenda seems fine.     (was not so difficult ...)
14:02:37 <uli-k> #topic Action item review
14:03:02 <uli-k> #info First one: ulik_ and vlaza check and improve description how to comment on documents
14:03:33 <uli-k> I have to confess that I didn't find time to check. vlaza is not here.
14:03:39 <uli-k> So we have to carry on.
14:04:08 <uli-k> #action uli-k and vlaza check and improve description how to comment on documents
14:04:33 <uli-k> #info next: uli-k_ to send the link to R2 etherpad to community to trigger feedback
14:04:45 <uli-k> #info I've done that.
14:05:20 <uli-k> #info next: MatthewLi2 to prepare zuul demo - we have another week. How is it going?
14:05:57 <MatthewLi> #info sorry, I have lots of work recently, so maybe in next weeks...
14:06:24 <uli-k> ok. Just let me know when to put it on agenda.
14:06:37 <uli-k> #info next: trozet to prepare RDO/TripleO demo
14:06:45 <MatthewLi> ok when it's ready, I will info you
14:07:00 <uli-k> trozet is not here and we still have time.....
14:07:11 <uli-k> That was all actions....
14:07:29 <uli-k> Just let me document the two others to carry on....
14:07:54 <uli-k> #action MatthewLi2 to prepare zuul demo
14:08:04 <uli-k> #action trozet to prepare RDO/TripleO demo
14:08:18 <uli-k> #topic Arno maintenance
14:08:59 <uli-k> #info There are several patches in progress for this. But I didn't see any far enough to get +2.
14:09:22 <uli-k> Any idea how to speed this up?
14:09:54 <radez> uli-k: I'm working with trozet on the RDO/Triple-O demo, just FYI
14:09:54 <MatthewLi> anyone can feel free to do the taskes
14:11:23 <fdegir> #info We're quite slow fixing maintenance issues
14:11:36 <fdegir> #info The important jobs are merge and daily jobs
14:11:55 <fdegir> #info  merge and daily jobs produce stuff and upload to artifacts.opnfv.org
14:12:17 <MatthewLi> sorry, i am a little busy recently, anyone can feel free to do them
14:12:29 <fdegir> #info and in order to +2 things, changes should fulfil basic functionality
14:13:51 <uli-k> I am sure we all agree... we still need to improve.
14:14:06 <fdegir> another thing we need is
14:14:09 <fdegir> the input from BGS
14:14:12 <fdegir> radez: ping
14:14:38 <fdegir> #info Especially for genesis-daily jobs, we need to set the triggers and go back to automatic runs
14:14:57 <fdegir> #info and we can't do this without input from BGS as they might be doing some work on the PODs
14:15:18 <fdegir> #info So the question to BGS is: should we enable automatic build/deploy/test runs on POD1 and POD2
14:15:34 <fdegir> #info if so, what should be the schedule per branch per POD?
14:16:02 <fdegir> #info We propose to have at least 1 run for master and 1 run for stable/arno on both PODs
14:16:39 <fdegir> uli-k: this could be an action on you or me to ask this to BGS
14:16:43 <chenshuai> agree
14:17:13 <uli-k> agree, but I was hoping, radez would have a suggestion, since he's here....
14:17:55 <uli-k> ok. I'll action me for that.
14:18:28 <radez> fdegir: would a daily run be enough?
14:18:44 <radez> afaik we should be good with automatic runs
14:19:01 <fdegir> radez: it depends on how much work is happening
14:19:17 <fdegir> if foreman/fuel getting changes then 1 run a day is not enough
14:19:26 <uli-k> and we have to run both , master and arno.....
14:19:44 <radez> the dailys would be in addidtion to the runs for changes though wouldn't it?
14:19:49 <fdegir> and trozet suggests running deployment for each and every patchset - which is a bit streching
14:20:02 <fdegir> we don't run deployment and functest for changes
14:20:33 <radez> ah gotcha... so why don't we devise a schedule
14:20:39 <radez> that gives some leeway
14:20:55 <radez> but keeps it continuous... say every 3 hours or 4 hours?
14:21:10 <fdegir> but it takes so much time to run full cycle - 3 hours maybe
14:21:28 <radez> so how about a new one kicks off every 4 hours?
14:21:29 <fdegir> so with the current infra, we are not able to run stuff in parallel
14:21:41 <fdegir> radez: it depends on you
14:21:53 <fdegir> last week someone was doing some vIMS(?) work on pod2
14:21:54 <radez> are you saying a full run takes 3 per installer?
14:21:59 <fdegir> and fuel work on POD1
14:22:25 <fdegir> so if we start deployments automatically, it might mess up the things for the people who are working on PODs
14:22:47 <radez> seems like we should use the non-LF labs for investgation work and do automatic deployments on the LF hardware
14:23:09 <fdegir> that could be a way forward for the short term
14:23:09 <uli-k> some plan like that might be the solution.
14:23:32 <radez> I wasn't part of the work that was being done on POD2 so we may have to wait for trozet to report back when he's abck online
14:23:36 <radez> he's out today
14:23:41 <fdegir> and we also will start working on deploying virtualized once maintenance work is done
14:23:54 <fdegir> and this should give us more freedom
14:24:01 <radez> we have enouugh labs floating around... we shoudl use the LF for the oficcial testing and the others for dev
14:24:11 <fdegir> but how much the others up?
14:24:25 <fdegir> I see most of them going up and down
14:24:39 <radez> we can rebuild in the amount of time it takes for our deploy  to run
14:24:56 <uli-k> I guess the other labs are also working on investigative stuff....
14:24:58 <radez> so if the foreman stuff needs invesitgation we can just rebuild and continue dev
14:25:17 <radez> fdegir: is the fuel stuff able to do the same?
14:25:26 <fdegir> need to check with fuel guys
14:25:32 * radez nods
14:25:33 <fdegir> but they were fine with at least 1 run a day
14:25:38 <fdegir> per branch
14:25:52 <uli-k> on LF?
14:25:54 <fdegir> yes
14:26:06 <fdegir> we can start with this and see how much people scream
14:26:18 <fdegir> if noone says anything, we increase to 2,3 or whatever per branch
14:26:42 <fdegir> and another difference is that instead of triggering stuff even there is no changes in git
14:26:57 <fdegir> this time we will poll SCM - meaning that if there is no change in SCM, run won't proceed
14:27:18 <fdegir> (did timer based runs for arno for checking deployment stability)
14:27:37 <uli-k> What about the proposal to have the official jobs on LF and investigative work on other labs?
14:27:49 <uli-k> Where you already thinking on that base?
14:28:03 <fdegir> fine by me but this needs to be checked with rest of the bgs guys
14:28:09 <uli-k> right.
14:28:17 <fdegir> they might want to verify some stuff on LF HW since it is the official place
14:28:18 <radez> is there another BGS meeting anytime soon?
14:28:32 <fdegir> radez: there should be one right after this meeting
14:28:37 <uli-k> could we directly action radez and JonasB to check whether we can go that way?
14:28:45 <fdegir> uli-k: agree
14:28:48 <radez> I hadn't see an invite, maybe i just missed it
14:28:57 <radez> I'll hang around :)
14:29:09 <fdegir> I don't look invites :D
14:29:19 <fdegir> I watch when a meeting starts
14:29:33 <uli-k> #action radez and JonasB to clarify whether we can activate official automatic builds on LF PODs
14:29:35 <radez> fdegir: that makes you mr. OPNFV :)
14:29:50 <fdegir> :D
14:29:52 <radez> I'll communicate with trozet and we'll let you know
14:29:54 <fdegir> ok
14:30:04 <fdegir> so uli-k should we pound our proposal as well
14:30:12 <fdegir> start with 1 run per branch per POD
14:30:19 <uli-k> per day...?
14:30:23 <fdegir> yes
14:30:30 <fdegir> and evaluate while we move on
14:30:47 <uli-k> yes. And thats not #action, but #agree ....
14:30:54 <fdegir> yes
14:31:16 <uli-k> #agree to run once per branch per POD per day and evaluate while we go
14:31:49 <uli-k> ... so we really need to have our jobs ready ....
14:31:58 <uli-k> ok.
14:32:07 <fdegir> chenshuai: and meimei have foreman job almost ready
14:32:27 <fdegir> and I'll apply trigger to fuel job so it looks fine for installer stuff
14:32:37 <fdegir> but documentation stuff is still the pain
14:33:06 <chenshuai> yes, we will start the doc when the other works are finished
14:33:16 <uli-k> yes. that's about the pipeline for arno and b-work, but we need to document what people need to do.
14:33:29 <fdegir> chenshuai: I'll +2 foreman change once you update the commit message
14:33:29 <uli-k> Do we have the jira's ready for doc?
14:33:41 <fdegir> and wait for submit until we hear back from radez
14:34:04 <fdegir> uli-k: we have the jira issues based on job names
14:34:14 <chenshuai> ok
14:34:20 <fdegir> the jira issues for merge and daily jobs are related to docs
14:34:40 <radez> sry, fdegir are you asking for my feedback onsomething now or on the ok to reenable auto deploy?
14:34:57 <fdegir> radez: you said you'll check with trozet
14:35:01 <MatthewLi> fdegir: i will check it that by spending some time....
14:35:13 <uli-k> I mean don't we need a doc telling people how to backport a patch from master to stable/arno
14:35:29 <fdegir> uli-k: that's the other story
14:35:33 <radez> yes, ok... just making sure that's what we were talking about fdegir thx. I'm sending him an email right now, we should hear back from him later today
14:35:47 <fdegir> Story	OCTO-76
14:36:09 <fdegir> is about documenting how to work with stable/arno
14:36:16 <fdegir> and taken by chenshuai
14:36:38 <fdegir> radez: just write it the answer from trozet to #opnfv-octopus
14:36:40 <fdegir> and we fix it
14:36:52 <uli-k> ok. Thanks. question answered.
14:36:53 <radez> fdegir: ack
14:37:14 <fdegir> so; installer jobs, merge/daily jobs, and document how to work with stable/arno
14:37:20 <fdegir> these are the 3 main items
14:37:49 <fdegir> #link https://jira.opnfv.org/issues/?filter=10200
14:39:21 <fdegir> I think we covered all maintenance stuff
14:39:33 <uli-k> my feeling was that we are now pretty clear about installer jobs and merge/daily jobs.
14:39:52 <fdegir> just infoing the line
14:40:00 <uli-k> I was not sure, everything is clear about what we want to write in the document.
14:40:11 <fdegir> #info Main items to finish for maintenance support are
14:40:15 <fdegir> #info installer daily jobs
14:40:16 <uli-k> (how to work with stable/arno)
14:40:26 <fdegir> #info merge/daily jobs for automatic document generation
14:40:36 <fdegir> #info documentation reagrding how to work with stable/arno
14:40:46 <fdegir> about how to work document
14:40:54 <fdegir> we can perhaps steal from OpenStack/ODL
14:40:56 <fdegir> and adjust it
14:41:27 <uli-k> .. if we want to work the same way......
14:41:59 <uli-k> but let's check the OpenStack/ODL documents then first.
14:42:28 <uli-k> ok. should we move on and have a look at R2 stuff?
14:43:15 <fdegir> would be good to have some fun :)
14:43:20 <uli-k> #topic Starting work on B-Release
14:43:45 <uli-k> #link https://etherpad.opnfv.org/p/octopusR2 is still our brainstorming
14:46:07 <uli-k> We should take one of the topics and start thinking how to best go forward.
14:46:20 <uli-k> Anybody feeling one of the issues should be top priority?
14:47:45 <uli-k> What I don't see explicitely on the etherpad is to make more parallel jobs possible
14:49:09 * uli-k feels everybody is gone and reads the etherpad (but there are only 5 people there....)
14:49:21 <fdegir> agree - and I think that should be our target to improve the feedback from CI
14:49:29 <fdegir> the parallel job stuff
14:49:57 <uli-k> #info let's start with a discussion on parallel jobs then
14:50:39 <fdegir> and this one probably should be part of it :optimize by using virtualization
14:50:51 <fdegir> I don't see us doing stuff in parallel with the current resources
14:50:52 <uli-k> parallel jobs for me means: Either we have to run in virtualized environment, or we have to run with more flexibility to where to run a job.
14:51:14 <fdegir> that was my thinking as I don't see existing labs stable enough
14:51:43 <fdegir> so virt is the way I suppose
14:51:49 <uli-k> Do we need BGS support to make a build job run virtualized?
14:52:50 <fdegir> I think we just need build servers - virt or metal or container
14:53:07 <fdegir> which I already raised this issue
14:53:21 <radez> uli-k: fdegir: the foreman stuff can build baremetal or virtualized, trozoet has been finishing up virt support
14:53:46 <MatthewLi> i have experience of container, docker
14:54:02 <uli-k> radez, where do you run it virtualized?
14:54:17 <MatthewLi> however I don't quite understand what we will do with it
14:54:44 <fdegir> MatthewLi: we build ISOs and in order to build them, we need some resources
14:54:46 <[1]JonasB> Nested KVM is supported by Fuel
14:54:55 <fdegir> curently we use ericsson-build server for all builds
14:55:26 <uli-k> JonasB: do you have build run virtualized?
14:56:26 <[1]JonasB> Yes, for build we can almost do anything. We're normally running build inside docker
14:56:57 <uli-k> radez: what about foreman and docker?
14:57:45 <uli-k> Do I need to ask LF to buy build server?
14:58:15 <fdegir> I think we should have build servers in LF lab
14:58:19 <fdegir> at least 2
14:58:28 <fdegir> it can be dockerized/virtual or whatever
14:58:48 <uli-k> OK. I think we have to close for BGS meeting. But I will action myself for the build server before.
14:58:49 <fdegir> and I think Chris and Frank have been talking about this already
14:59:01 <uli-k> I think so too.
14:59:01 <fdegir> ok
14:59:05 <radez> uli-k: we have development labs we use
14:59:17 <MatthewLi> docker itself may still have some problems in using, it should be careful using it
14:59:20 <radez> uli-k: what about foreman and docker? we're not doing any containerization right now
14:59:33 <fdegir> #info Start thinking about running jobs in parallel and in virtual environments
14:59:43 <uli-k> #action uli-k request LF build server
15:00:19 <uli-k> we need to close. Urgent discussions, please listen on #opnfv-octopus
15:00:33 <uli-k> #endmeeting