16:30:36 #startmeeting kolla 16:30:37 Meeting started Wed Sep 9 16:30:36 2015 UTC and is due to finish in 60 minutes. The chair is sdake. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:30:38 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:30:40 The meeting name has been set to 'kolla' 16:30:46 #topic rollcall 16:30:47 hi 16:30:49 hi 16:30:50 yo folks o/ 16:31:09 howdy 16:31:20 hey 16:32:08 i'll give it acuple more minutes 16:32:18 afternoon 16:32:29 hi everyone 16:32:51 #topic rc1 planning 16:33:02 no anncuncements for this week btw ;) 16:33:44 <[Leeloo]> Hello 16:33:45 #link https://launchpad.net/kolla/+milestone/liberty-rc1 16:33:50 hey [Leeloo] 16:33:53 hey 16:34:04 you see we have a whole lot of blueprints 16:34:10 the deadline is the 25th 16:34:43 on the 18th I'm going to kick out any low/medium blueprints not in good progress to mitaka 16:35:38 sdake: fyi I dont see myself having time to tackle neutron-third-party-plugins 16:35:49 ok kicking it now then moment 16:35:57 so unless someone wants to grab it kick it forward I guess 16:36:27 are ther eany other blueprnts people want to kick 16:36:43 mickt's ansible-murano is done 16:36:48 will mark as so 16:37:08 going to go through the started ones now and get a status update 16:37:23 #link https://blueprints.launchpad.net/kolla/+spec/logging-container 16:37:43 review is upstream 16:37:47 inc0 will this definately make good progress by the 18th? 16:37:54 is that all the reviews required? 16:38:26 no, we'll need to turn services to use it 16:38:38 ok start working on those reviews please 16:38:46 don't serialize waiting on this review to finish 16:39:03 blocked anywhere or all set? 16:39:17 well, I'd just like confirmation if approach itself is correct 16:39:27 #link https://blueprints.launchpad.net/kolla/+spec/replace-config-external 16:39:33 inc0 i will review your patch after meeting 16:39:39 and provide guidance there 16:39:40 sdake, I'll be ready with that 16:39:51 rhallisey it needs to finish sooner rather then later :) 16:39:51 I'm changing stuff around to work w/ keystone 16:39:54 its a big change 16:39:55 yes 16:40:03 i dont want it landing on the 25th 16:40:06 once I get past keystone it should just be copy paste 16:40:31 ok well please if you are blocked reach out to me 16:40:34 kk 16:40:37 I'm good atm 16:40:41 thanks though 16:40:55 #link https://blueprints.launchpad.net/kolla/+spec/kolla-ansible-script 16:41:23 Hu had indicated he would provide status updates via the mailing list or irc 16:41:26 haven't seen any of that 16:41:34 may need to reassign this 16:41:55 since sam isn't here I won't cover his blueprints 16:42:23 mongodb + ceilometer, coolsvap said he is working on this weekend to wrap up 16:42:32 in my mind, that leaves us in good shape for a release 16:42:42 the one problem spot is cinder 16:42:46 which is totally bust 16:43:06 anyone brave enough to take on ownership of fixing cinder? 16:43:34 I will if I finish the refactor in time 16:43:48 like this week 16:43:52 sdake: can you summarise the problem if possible? 16:44:03 lvm creation works, attachin to vm does not work 16:44:20 but i did alot of work to mke lvm creation work thta may be incompatible with vm attachment 16:44:45 it may be that we kick cinder out of liberty 16:45:08 are there other peices of work people intend to bring into rc1? 16:45:53 jpeeler since your here, mind giving us a quick update on ironic? 16:45:55 i have a review up for ironic, but i need to finish it 16:46:06 it just needs the neutron portion completed 16:46:26 do you end up with a nova that can schedule bare metal and vms? 16:47:16 no, i sent out an email to dev list about it and got no response, so i don't think that is possible without multiple availability zones 16:47:33 jpeeler i'll follow up with the ptl and try to get a response to your email 16:47:52 did you tag with [rionic] ? 16:47:58 i did 16:48:09 ok, i think we want to avoid having ironic take over nova 16:48:31 sound slike your almost blocked on a response on that email 16:48:57 it's disabled by default 16:49:09 ya but if you want to use it, you can only use ironic and nothing else? 16:49:13 that doesn't make any sense ;) 16:49:14 yeah, ironic is required to have kolla-based undercloud 16:49:25 it's main usage case is for undercloud i would suspect 16:49:33 http://lists.openstack.org/pipermail/openstack-dev/2015-September/073530.html 16:49:44 i want to use it to schedule bare metal vms 16:49:56 alongside virtua lmachines 16:50:10 in other words, as an overcloud ;) 16:50:13 would be nice, but i don't know how to do that 16:50:24 ok well lets get some followup on the mailing list 16:50:27 i'll take that action 16:50:37 #topic bug squashing 16:50:47 we have 40-50 bugs in the bug tracker 16:50:50 half of them are unowned 16:50:54 i try to fix 1-2 a day 16:51:03 but it would be helpful if other folks could take on atleast 1-2 bugs for the release 16:51:38 there are about 20 unassigned 16:51:53 after rc1, we have rc2 and rc3, with 1 week intervals 16:52:02 at rc1, I think I'm branching liberty 16:52:16 and then we will backport for the rc2/rc3 tags if bug fixes hit master 16:52:21 that way development can continue on in master 16:52:23 after rc1 16:52:34 we wont be backporting features 16:52:38 just bug fixes 16:54:00 can i get acks from folks in the meeting that eveyrone understands the plans and schedule? 16:54:12 +1 16:54:23 check 16:54:29 hey harmw :) 16:54:31 :) 16:54:31 ack 16:54:33 sounds good 16:54:50 ok thats all I have so :) 16:54:54 #topic open discussion 16:55:10 i know i've been burning up our agenda the last month and left little time for open discussion 16:55:14 so now is 30 mins available :) 16:55:27 when do we move to openstack github? 16:55:34 11th (friday) 16:55:40 cool 16:56:27 would be interested to know what kind of test scenarios people are running these days? 16:56:31 if any? ;) 16:56:43 pbourke i run multinode over and over 3 control 1 network 3 storage 3 compute 16:56:46 works like a champ 16:56:50 for centos +binary 16:56:53 sdake: nice 16:57:03 sdake: im attempting the same but not yet got to booting a vm 16:57:04 I did ubuntu and it works as well 16:57:12 i have noticed one problem, if a container is in restart state, the container contents ins't updated, need to file a bug for that 16:57:13 building behind proxy works like charm 16:57:21 sorry dc'd 16:57:23 sdake: ? 16:57:38 it would be nice to brainstorm painpoints people are noticing during testing at some stage 16:57:44 sounds like you want it to pull again, right? 16:57:51 harmw if a container is restarting, it wont be pulled again 16:57:56 is anyone else running into the haproxy user creation issue? 16:58:04 rhallisey: yes 16:58:05 rhallisey could yo ugo into more detail 16:58:07 it's driving me crazy 16:58:10 doesn't restart policy do the pulling sdake ? 16:58:12 rhallisey: only in AIO though? 16:58:15 i've had ha proxy creation fail a couple times during my 50-100 deploys 16:58:22 harmw no 16:58:23 pbourke, ya that's why I use 16:58:31 ok , please file a bug:) 16:58:42 https://bugs.launchpad.net/kolla/+bug/1491782 16:58:43 Launchpad bug 1491782 in kolla "Fail to create haproxy mysql user" [Critical,Triaged] 16:58:44 rhallisey how consistent is it for you? 16:58:46 takes forever to test cause 1/2 the time haproxy blows up 16:58:59 its intermitent it seems 16:59:07 sdake, It will break like 10 times in a row then stop 16:59:08 its only AIO? 16:59:10 and come back 16:59:13 ya it's AIO 16:59:31 why do we even run haproxy in aio tho? 16:59:40 inc0 simpler to setup playbook wise 16:59:54 check moving parts :) 16:59:56 btw, pleae do not recommend wierd debug sessions with people like turning off haproxy 17:00:08 it makes it much harder to diagnose their actual problem ;) 17:00:27 speaking of diagnostics, we could really use some more docs ;) 17:00:42 people struggle over the meaning of a VIP 17:01:08 anything specific sdake (area)? 17:01:08 and the ha architecture 17:01:10 i get questions on it all the time 17:01:29 people want to understand how he haproxy round robin works 17:01:31 maybe I can get some time to get into /docs :) 17:01:38 i still haven't quite got it pictured in my brain 17:01:44 https://blueprints.launchpad.net/kolla/+spec/sanity-check-container this would help in terms of finding which service broke in the first place 17:02:10 it's often that for example glance endpoint creatin fails because keystone didnt start correctly 17:02:25 sdake: have you experienced any timing issues wrt to maria coming up before keepalived? So it cant access the vip (multinode) 17:02:31 and ansible marked it as success because there is no checking whatsoever 17:03:00 pbourke, keepalived runs before maria in ansible 17:03:00 how do people feel about supplying a handful of helper ansible tasks. for resetting the cluster, checking status etc 17:03:09 i like the idea, but creating separate containers for each services seems like overkill to me 17:03:18 and it's rocket fast...if you remember to enable ip_Vs 17:03:19 inc0: we're hitting instances where that's not the case, I'll follow up with a bug 17:03:23 pbourke haven't seen that but i have really fast hardware 17:03:29 sdake: kk 17:04:37 yeah, we hit all sort of races because ansible just sets up container, not waiting for service to start 17:04:50 we could change that 17:04:52 probably 17:05:06 registering some thing, and check stdout or something 17:05:09 yeah that really sucks 17:07:02 let's brainstorm solution maybe? 17:07:24 harmw's suggestion of checking stdout seems ok 17:07:27 checking docker logs will be ugly 17:07:35 sdake: if the formal meeting is over, perhaps we should just move to #kolla now :) 17:07:46 yup sounds good 17:07:48 thanks folks :) 17:07:51 good session :) 17:07:52 thanks 17:07:52 #endmeeting