09:03:10 #startmeeting ha 09:03:11 Meeting started Mon Feb 29 09:03:10 2016 UTC and is due to finish in 60 minutes. The chair is aspiers. Information about MeetBot at http://wiki.debian.org/MeetBot. 09:03:12 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 09:03:15 The meeting name has been set to 'ha' 09:03:23 #topic introductions 09:03:43 as we have a couple of new people, maybe we should do quick introductions 09:03:54 bandini, samP: you wanna quickly introduce yourselves? 09:03:57 sure thing 09:04:00 or if you prefer you can lurk ;-) 09:04:32 actually maybe we all should, for your benefits 09:04:43 my name is Michele Baldessari, I live in the alps and I recently joined RH's HA team. Working on Openstack/HA architecture/installer/bugs whatnot 09:05:01 I will come to Austin, so hopefully I will meet some of you ;) 09:05:11 My name is Sampath and I am working with masa, and kazu at tokyo. mainly fixing masakari now 09:05:15 I'm Adam Spiers, working for SUSE on OpenStack (mainly HA) from London, and I usually moderate this session, and also maintain the openstack-resource-agents project 09:05:21 hey guys 09:05:30 hi beekhof :) 09:05:34 oh and beekhof is my tech lead ;) 09:05:39 ahah 09:05:40 * bandini waves 09:05:41 cool 09:05:48 this bandini guy sounds sketchy, i suggest we stear clear of him 09:05:52 great to have you here 09:05:54 haha :) 09:06:04 lol, that's why you're the lead, clear ;) 09:06:09 *clearly* 09:06:15 that beekhof guy is *definitely* sketchy ;-) 09:06:36 doesn't even own any lego sets 09:06:44 rotfl 09:06:52 I'm Dawid Deja, Wworking for Intel on Openstack from Poland, mainly focusing on instances HA right now 09:06:52 :) 09:07:07 * _gryf is wondering what is happening… 09:07:29 hazing the new guy 09:07:32 _gryf: we're doing introductions for the new people 09:08:08 <_gryf> anyway, my name is Roman Dobosz, working at Intel, Gdańsk, Polan. I use to work closely on the tenants HA automation 09:08:36 #info two first-timers to the meeting: bandini (RH) and samP (NTT) 09:08:37 <_gryf> aspiers, no no, the other things ;P 09:08:44 oh :) 09:08:47 ok 09:09:02 #topic Current status (progress, issues, roadblocks, further plans) 09:09:12 quick update from my side 09:09:39 I've been doing some reviews on code submitted to openstack-resource-agents by Norbert Illes 09:09:54 we're working towards getting bashate passing 09:10:02 so we can add some basic CI to the repo 09:10:28 #info work proceeding on https://bugs.launchpad.net/openstack-resource-agents/+bug/1550203 09:10:28 Launchpad bug 1550203 in openstack-resource-agents "Bashate violations in OCF scripts" [Medium,In progress] - Assigned to Norbert Illes (nilles) 09:10:46 also, on Thursday I crashed Fuel's weekly team IRC meeting 09:10:57 and asked them about their OCF RA forks 09:11:04 which started sometime in 2013, I think 09:11:26 Basically I asked if they would be interested in reconverging with this new upstream project, and they seemed enthusiastic 09:11:51 #action aspiers to send an email to openstack-dev about reconvergence of Fuel RAs 09:11:56 i saw that meeting go past 09:12:08 we need to form a working group and arrange a meeting 09:12:17 that is something I am also interested in 09:12:27 bandini: great! 09:12:33 i tried to get their rabbit agent to converge, but i dont think it went anywhere 09:12:51 ah, well luckily for me, the rabbit RA is out of scope ;-) 09:12:57 since it's not purely openstack-related 09:13:07 although that's slightly sad to hear 09:13:26 also you RH guys should check recent changes upstream to Nova{Compute,Evacuate} 09:13:42 beekhof: I thought I saw a moderately recent mail from bogdan about rabbitmq RA..I need to dig it up again 09:13:46 my colleague Vincent Untz made a bunch of fixes for our next release which might help you 09:13:59 probably not many parallels to draw from that experience. the pure openstack agents are significantly simpier 09:14:17 beekhof: I hope you're right 09:14:53 NovaCompute is gone from ours 09:14:53 I'm also very happy to see that ... 09:15:02 #info http://bugs.clusterlabs.org/show_bug.cgi?id=5269 has been fixed 09:15:02 bugs.clusterlabs.org bug 5269 in Core "DoS: valid authkey should be required for takeover of a Pacemaker remote" [Normal,Resolved: fixed] - Assigned to ygao 09:15:28 the wait logic is now in a more consistently named agent and we use systemd for starting the compute process 09:15:35 beekhof: oh? perhaps you can give a status report and briefly summarise? 09:16:00 there were issues starting nova-compute from anywhere other than systemd 09:16:16 beekhof: ok. so you renamed NovaEvacuate to nova-evacuate or something? 09:16:23 something to do with the way pacemaker does fork+exec 09:16:55 no, NovaCompute -> nova-compute-wait + systemd:openstack-nova-compute 09:17:23 so instead of fighting it, we just switched to systemd 09:18:06 hmm 09:18:40 worth pushing upstream? 09:18:44 yes 09:18:48 ok 09:19:01 can you or bandini take an AI to kick that process off?? 09:19:02 for that start+stop and used OCF for the "wait for evacuations to complete ebfore starting nova 09:19:09 me 09:19:34 interested to see how that works 09:19:42 its not complex 09:19:47 that's good ;-) 09:20:00 #action beekhof to upstream the decomposed NovaCompute agent 09:20:18 #info RH have split NovaCompute into nova-compute-wait + systemd:openstack-nova-compute 09:20:21 great, thanks! 09:20:45 ok. bandini, too early for you to have anything to report? 09:21:11 aspiers: yeah, I guess so 09:21:16 no problem 09:21:23 _gryf? 09:21:46 <_gryf> i've been involved in other tasks recently 09:22:00 <_gryf> so nothing to report 09:22:04 ok 09:22:07 ddeja? 09:22:27 I have finished working on auto-evacuate Pacemaker agent for mistral https://github.com/gryf/mistral-evacuate/blob/master/fence_evacuate.py 09:22:43 cool! 09:22:47 so I can say that PoC stage is finished 09:23:09 awesome :) 09:23:19 #info Mistral PoC has been completed 09:23:31 will there be any kind of demo to share? 09:23:53 that would be good 09:24:20 I can prepare one, but it's hard for me to imagine, how it should look like.. 09:24:33 maybe we can interrupt the regular schedule to have a demo-off between mistral and masakari 09:24:47 lol 09:24:53 next week i mean 09:25:04 demos are welcome in my book 09:25:13 although I don't think they need to be done as a competition ;-) 09:25:30 ddeja: maybe have a think, and let us know if you have any ideas 09:25:37 ddeja: even a 3 min youtube video would be great 09:26:00 aspiers: ok 09:26:03 thanks! 09:26:11 well at some point we need to decide which approach to throw our collective weight behind 09:26:18 #action ddeja will have a think if there's an easy way to do demos 09:26:23 maybe austin is a good time for that 09:26:36 beekhof: agreed, and yeah we can definitely discuss more in Austin 09:26:37 btw. I will not do much work this week, since I'm on medical leave 09:26:53 ddeja: ok. hope you're alright! 09:27:07 samP: you want to report anything? it's optional ;-) 09:27:31 still workin on masakari, sqlalchemy and some patches for masakari to work with centos 09:27:36 * beekhof has to run, i'll read the scrollback 09:27:45 cya beekhof, thanks for joining! 09:28:22 beekhof, bandini: check for recent changes to NovaEvacuate, I think there have been some you could use 09:28:31 aspiers: ack, thanks 09:28:55 samP: sounds good 09:29:02 hopefully I can finish them before next meeting 09:29:07 cool 09:29:12 ok, I think we're done with status updates? 09:29:22 anyone have any other topics they wanna raise? 09:29:31 I think we should talk about Austin briefly 09:29:40 hopefully most of us will be there 09:29:55 * bandini nods 09:30:01 I'm still waiting for confirmation, but if any of my talks get accepted I'm definitely going 09:30:17 #topic Austin summit 09:30:37 <_gryf> aspiers, is there any session reserved for ha topics? 09:30:45 I think we should aim to get a more formal platform for HA BoF design sessions 09:30:50 _gryf: not that I'm aware of 09:30:56 <_gryf> mhm 09:31:04 it's been a complaint of mine for every single summit since San Diego 09:31:16 there was one in San Francisco (Folsom) but that's the last one I can remember 09:32:05 <_gryf> it should be some kind of procedure to have a reservation and time on the schedule I think 09:32:10 any opinions on whether we should aim forr multiple design sessions? 09:32:12 yeah 09:32:21 e.g. we could ask the nova guys for a session on compute HA 09:32:31 and maybe one for neutron too 09:32:42 could even have a cross-project one 09:32:57 for topics like reconnections in oslodb 09:33:05 <_gryf> aspiers, that's what i thinking as well - to have a cross-project session 09:33:59 ok, I'll raise on openstack-dev@ and cc thingee (Mike Perez, cross-project guy) 09:34:13 <_gryf> +1 09:34:17 aspiers: sounds like a plan 09:34:40 we should definitely have demos of masakari and mistral PoC in Austin too :) 09:34:55 <_gryf> aspiers, you mean - live? 09:34:57 <_gryf> :> 09:35:02 sure! :) 09:35:07 <_gryf> oh my :) 09:35:16 don't worry it's the design summit, so noone will expect demos to work ;-) 09:35:37 I don't even expect demos in the keynote sessions to work ;-) 09:35:45 lol 09:35:54 <_gryf> I'm not sure if I'm going to be in Austin - decission is not taken yet by my boss (this is also true for ddeja) 09:36:02 same here 09:36:03 yup 09:36:07 travel budget, blah blah :( 09:36:14 <_gryf> right. 09:36:35 but if I organise some design sessions then it's a lot easier ;-) 09:36:48 <_gryf> heh :) 09:36:55 ok, anything else to discuss on Austin? 09:38:23 * _gryf have nothing than that. 09:38:27 not that I can think of now 09:38:41 #topic AOB (Any Other Business) 09:38:55 the floor is open :) anything else? 09:39:24 BTW openstack-ha-guide project seems very active, I guess more reviewers are always welcome there 09:40:16 yes, I try to review there from time to time 09:40:52 cool 09:41:11 ok, if nothing else we can finish early. going once ... 09:41:31 going twice ... 09:41:54 sold! thanks all for coming, and hope to see you next week :) 09:41:58 <_gryf> ok, thanks! 09:42:04 Thx... 09:42:11 thanks! 09:42:22 #endmeeting