12:59:43 <[1]JonasB> #startmeeting Fuel weekly sync meeting
12:59:43 <collabot> Meeting started Thu Feb 18 12:59:43 2016 UTC.  The chair is [1]JonasB. Information about MeetBot at http://wiki.debian.org/MeetBot.
12:59:43 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
12:59:43 <collabot> The meeting name has been set to 'fuel_weekly_sync_meeting'
12:59:52 <lmcdasm_> #info Daniel Smith
13:00:04 <[1]JonasB> #topic roll-call
13:00:05 <z0d> #info Peter Barabas
13:00:16 <[1]JonasB> Only IRC today :-)
13:00:20 <z0d> all right
13:00:29 <[1]JonasB> Hi everyone
13:00:35 * lmcdasm_ waves!
13:00:58 <lmcdasm_> i see that Szi is back in action - happy to see you! :)
13:01:17 <[1]JonasB> #info Jonas Bjurel
13:01:24 <mskalski> #info Michal Skalski
13:01:47 <[1]JonasB> Hi Szi, how are things?
13:03:25 <[1]JonasB> s_berg: Lets exersize your jabber set-up
13:04:20 <z0d> <-:
13:04:37 <z0d> he is not here though(?)
13:05:00 <z0d> 12:54 -!- s_berg [~sfb@quintus.consultron.com] has quit Ping timeout: 268 seconds
13:05:36 <fdegir> he's always here, watching every move we make
13:05:39 <[1]JonasB> z0d: consultron is his spy bot:-)
13:05:50 <z0d> Sauron's eye
13:05:58 <z0d> Sauron Berg <-:
13:05:58 <[1]JonasB> Lets move on
13:06:11 <[1]JonasB> #topic RC2 readiness
13:06:26 <billyoma> #info Billy O'Mahony
13:06:32 <[1]JonasB> mskalski: RC3 is in stable, right
13:07:01 <mskalski> [1]JonasB: yes
13:07:25 <[1]JonasB> #info Beryllium RC3 is in stable/brahmaputra
13:08:11 <[1]JonasB> So we need J�rgens yardstick patch in so we can start pass ODL-L3 CI runs.
13:09:24 <[1]JonasB> #info Jonas will ask J�rgen to mergehttps://gerrit.opnfv.org/gerrit/#/c/10151/ in order to pass ODL-L3 CI runs.
13:09:43 <mskalski> one question I'm looking at https://build.opnfv.org/ci/view/OPNFV%20Platform%20CI%20-%20Alternative%20View/ does apex deployments are also checked by yardstick?
13:10:28 <[1]JonasB> mskalski: It doesnt look so in my eyes, fdegir?
13:10:46 <mskalski> [1]JonasB: this change should also contain small fix to security groups, as described in mail
13:11:08 <fdegir> mskalski: they don't follow the same job structure
13:11:15 <fdegir> mskalski: like they do for the rest
13:11:18 <lmcdasm_> i dont see any YS, FUNTEST or rally (Tempest?) jobs associated.
13:11:21 <fdegir> mskalski: they only show deployments
13:11:26 <lmcdasm_> ahh.. fdegir knows best :)
13:11:31 <fdegir> mskalski: so please disregard them
13:11:44 <[1]JonasB> mskalski: I will ask J�rgen to add your stuff
13:11:53 <fdegir> meaning disregard whatever apex job you see on jenkins
13:12:02 <[1]JonasB> #info SFC scenario is checked in
13:12:11 <mskalski> fdegir: ok, thanks for explanation
13:12:52 <[1]JonasB> #info Ferenc have a patch for tacker, I will ask him to move it to another directory (/prototypes/sfc-tacker)
13:13:38 <[1]JonasB> Question is how is the tacker script moved to a controller and executed?
13:14:44 <[1]JonasB> A question, Fernce is probably sleeping, he worked like hell, maybe better to merge it and I move it in a separate patch?
13:15:19 <[1]JonasB> I'll do so!
13:16:10 <lmcdasm_> hey tehre.
13:16:15 <[1]JonasB> billyoma: OVS failed yesterday, I gues it was because of the ubuntu packet issue that you fixed yesterday?
13:16:16 <lmcdasm_> right now its a tarball that you can install
13:16:31 <lmcdasm_> i dont think the Plugin wrapper is there yet (answering your tacker quetion - im doing it right now with his patch)
13:16:35 <billyoma> JonasB: failed to build?
13:16:40 <szilardcserey> Hi everybody, sorry I was not at my desk :)   thank God I'm well, Praise the Lord!
13:16:42 <lmcdasm_> the next step (now that it can be done manual) would be to wrap it in a plugin
13:17:09 <billyoma> yes that would be it. I am running it on a cron job here every 15 minutes so hopefully I'll spot build break before you do!
13:17:25 <[1]JonasB> szilardcerey: Welcome back:-)
13:17:41 <szilardcserey> thanks Jonas, good to be back :)
13:17:52 <[1]JonasB> billyoma: That is great, Thanks!
13:18:06 <billyoma> np
13:18:34 <[1]JonasB> Any news on KVM and failing yardstick test cases?
13:18:58 <fdegir> pulling Jörgen in
13:19:25 <fdegir> the latest mail was from me to Ruijing, sending logs/details etc
13:19:49 <fdegir> so the same state as last friday
13:21:58 <[1]JonasB> mskalski: J�rgen posted this now https://gerrit.opnfv.org/gerrit/#/c/10151/
13:22:21 <mskalski> [1]JonasB: yep he got +1 from me ;)
13:22:28 <[1]JonasB> mskalski: As far as I can tell it includes your stuff
13:23:31 <[1]JonasB> So, apart from Ference patch which I will expedite, I dont know of anything missing for the release, do anyone else?
13:25:05 <mskalski> ODL does not produce stable results, we trying make workaround but I don't what else we can do
13:25:32 <mskalski> that is why I was intersted how stable it is for example in apex
13:25:43 <[1]JonasB> mskalski: Agree, nothing we can really do about it.
13:26:33 <[1]JonasB> mskalski: Can you write a Jira bug for ODL, with the information you have collected?
13:27:34 <jnon> Do you think I can merge the patch now
13:27:57 <[1]JonasB> jnon: Yes for sure :-)
13:28:17 <mskalski> [1]JonasB: yes I think that we should gather information about failed deployments together with Stefan and put it in one place
13:28:44 <jnon> I submit to master and stable and we'll see how it goes :)
13:29:06 <lmcdasm_> have we seen any improvements in RC3
13:29:15 <[1]JonasB> mskalski: Great, Jira tickets will automatically be reflected in the release notes.
13:29:16 <lmcdasm_> at all?  2.3 still seems to be running very hot (even with J8)
13:29:41 <mskalski> maybe not failed deployments because usually its deploy, but failed tests, exceptions etc.
13:30:01 <lmcdasm_> hmm.. i wonder if the failures arent load related still
13:30:28 <lmcdasm_> for me, all my deployments (as Michal says) work fine, but the controller doesnt stay up very long, has to be restarted and seems (i think maybe we should pput a script to gather metrics)
13:30:41 <lmcdasm_> but seems to have some sort of memory leak - or the JVM flags are completely ignored
13:30:54 <lmcdasm_> cause regardless of the RAM i give the box, ODL takes 80% off the start
13:31:05 <lmcdasm_> so 8Gb ODL - 6GB
13:31:10 <lmcdasm_> 12- ODL 10, etc, etc
13:31:30 <[1]JonasB> lmcdasm: Wow, maybe rewrite it all together?
13:31:40 <lmcdasm_> i think Michal a good idea of maybe trying a "seperate deployment" of the ODL on it own node .
13:32:01 <lmcdasm_> and see how that behaves..   cause while we are only testing 5 nodes, this problem will really get bigger as we go along with larger clusters ;)
13:32:11 <lmcdasm_> (since we are production / carrier grade platform :P)
13:32:26 <[1]JonasB> ;-)
13:32:37 <lmcdasm_> questin
13:32:47 <lmcdasm_> is the co-location of the ODL on the controlleres a OPNFV requiremnts?
13:32:57 <lmcdasm_> requirement*
13:33:34 <[1]JonasB> lmcdasm: I think there is an early ppt stating this, but we should maybe question that for SR1
13:33:58 <mskalski> I wonder how it will look when we start deploy 3 ODL controllers  in cluster :p
13:33:58 <lmcdasm_> i think we should try a couple different deployment types..
13:34:05 <lmcdasm_> (exactly Michal)
13:34:15 <lmcdasm_> im worried about what happens with HA ODL - the controllers are gonna fall over
13:34:40 <lmcdasm_> and if we look "down the road"... is the ODL on the controllers feasible for large deployments - how big will our controllers have to be
13:35:09 <lmcdasm_> AND - when we go the ODL guys with a trouble and say "ya - we are colocated on the controllers"  are we gonna run the risk that they say - well, first thing, get it off there before you do anything else and try again?"
13:35:15 <lmcdasm_> sory didnt mean caps there
13:36:14 <lmcdasm_> sounds like a SR1 scenario to me :)
13:38:24 <[1]JonasB> lmcdasm: YEP
13:39:04 <[1]JonasB> So are we ready to tag RC2, meaning that from now on we only fix serious faults.
13:40:32 <lmcdasm_> +1
13:40:39 <[1]JonasB> +1
13:40:47 <mskalski> +1
13:40:58 <billyoma> +1 (if I have a vote :) )
13:40:58 <szilardcserey> +1
13:41:19 <[1]JonasB> billyoma: Sure you have!
13:41:27 <billyoma> :)
13:41:58 <[1]JonasB> So that is majority!
13:42:01 <lmcdasm_> Billy's vote counts for 3 right?
13:42:02 <lmcdasm_> :P
13:42:17 <[1]JonasB> I will tag before EOB today.
13:42:34 <lmcdasm_> one question
13:42:40 <[1]JonasB> #info: We have an RC2 tag
13:43:21 <lmcdasm_> are you open to accepting a new ODL version / bug fix if they cut it - or you want RC3 to the taget and we deal with it in SR1?
13:43:45 <lmcdasm_> to be the target* (sorry new keyboard).
13:43:56 <[1]JonasB> I don't think we will take a new ODL unless we have very good reasons to
13:44:00 <lmcdasm_> ok
13:44:20 <[1]JonasB> That is for SR1
13:44:37 <[1]JonasB> Ok can we end here?
13:45:14 <[1]JonasB> Thanks for all the good work!
13:45:19 <[1]JonasB> #endmeeting