20:01:33 #startmeeting kolla 20:01:34 Meeting started Mon Apr 27 20:01:33 2015 UTC and is due to finish in 60 minutes. The chair is sdake. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:01:35 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:01:38 The meeting name has been set to 'kolla' 20:01:50 #topic rollcall 20:01:55 hey 20:01:57 hello 20:02:01 hi 20:02:02 o/ folks 20:02:03 hey 20:02:45 #topic meeting time change 20:02:57 we have many more contributors from apac 20:03:17 but unfortunately they can't attend our irc weekly meeting because its like 4am their time 20:03:35 I would like to alternate meeting times between 1600 utc and 2300 utc 20:03:36 yea, well 10pm isn't all that funny either 20:03:41 :p 20:04:06 this should allow ust o increase our development velocity 20:04:07 anyway, sdake good point 20:04:20 thoughts concerns etc? 20:04:50 mstoly looking for feedback from the core team here :) 20:04:58 I'd vote 1600 20:05:05 who is the core team anyway :) 20:05:20 we need to alternate to hit all parts of the world 20:05:21 well alternating can make it harder to follow 20:05:31 calendar should solve that problem easily 20:05:36 makes sense to alternate 20:05:46 can update channel topic with next meeting scheduled time as well 20:05:48 ya I'll just have to make a reminder 20:06:01 ok here is what i'm going to do 20:06:10 I'll send a doodle poll to openstack mailing list 20:06:11 with some times 20:06:25 sounds good 20:06:29 mon-wed 1400-1700utc and 2200-2400utc 20:06:39 please vote on which ones you can make 20:06:45 and i'll go with critical mass 20:07:29 alternating is fine, just may not make the later time if the examples given above are decided on 20:07:31 any objections? 20:07:43 jpeler what is the latest you think you can make? 20:07:51 and rhallisey since you guys are on the east coast 20:08:03 22:00 UTC, but don't schedule anything around just me! 20:08:15 jpeeler, agreed 20:08:20 aroudn~22 20:08:22 well i'd like the current core at the meetings if possible 20:08:43 ok i'll set 2200 as the end time 20:08:52 rather the later time 20:08:54 altnerating 20:09:20 #topic kilo branching 20:09:29 so there is a problem with kilo branching 20:09:32 which is why we are not yet branched 20:09:43 I'd ask thatfolks not +2 / +a changes until the branch finishes 20:09:50 hopefully today/tomorrow 20:09:56 there are a couple feature exceptions I'd like people to consider though 20:10:26 https://review.openstack.org/#/c/177681/ 20:10:43 to fix the kilo branching, I need to change project-config 20:10:53 still sorting out the chnage, but there is a turnaround on that change of 1-2 days 20:11:20 can folks either -1 or +2 that change 20:11:25 keeping in mind I know its not perfect 20:11:31 and we can address perfection in master ;) 20:11:43 ok 20:12:09 one more change 20:12:11 https://review.openstack.org/#/c/177684/ 20:13:36 one more change 20:13:38 https://review.openstack.org/#/c/170306/ 20:14:04 can we spend 2-3 minutes now reviewing, and once you have reviweed with either -1, +2, please respond in channel so we can proceed 20:15:21 reviewed and +2's them 20:16:01 make sure not to just rubber stamp 20:16:06 but actually look at the changes toop lz :0 20:16:09 ;) I mean 20:17:17 so the current goals is to support both nova-network and neutron? 20:17:31 sdake, what are you trying to do with the demo 20:17:50 why is cirros being downloaded and loaded into glance in a network config script? 20:18:09 maybe not worth refactoring at the current time 20:18:18 jpeeler it is really a misnomer, it should be called "init-runonce" or something 20:18:30 ^^ agreed 20:18:33 but the reason is because it can act atomically 20:18:38 sdake: compose/README.md should probably be updated as to point to the new network setup script as well, no? 20:18:42 it clearly needs a refactor 20:18:51 jtriely cool -1 and add a comment 20:19:09 in my review I noticed there is a "cho" where it should be "echo" 20:19:20 so I -1 and set workflow to -1 20:19:38 the demo launches 20 virtual machines with neutron ports and neutron floating ips 20:20:24 ok after meeting I will udpate any review comments and would appreciate fast reviews so master is in suitable shape for branching 20:20:39 #topic continuous integration 20:20:55 jpeeler has been working on this a bit, have any status update? 20:21:30 sdake: some progress has been made, can now call tox to setup docker properly 20:21:31 https://review.openstack.org/#/c/168598/ 20:21:45 I dont think we want tox setting up docker 20:22:01 I think we want a script in tests which sets up the environment,and tox should call that script 20:22:20 there is another guy from apac - his name escapes me atm 20:22:23 but he is also working on this problem jpeeler 20:22:24 i don't see the difference from what i said 20:22:44 jpeeler do you ahve a review for early looking? 20:22:58 no, but i will soon. it's not really based off your review. couldn't get that going at all 20:23:13 it sure would be nice to go into our demo at summit (#6 on the most popular talks yay:) saying we have some basic functional testing 20:23:20 also, kolla isn't a python project. so i've tried to avoid as much of that as i could 20:23:35 I like how testr works 20:23:36 (while still using all the python test runner stuff) 20:23:42 it would be nice if we could write the test cases in python 20:23:58 yes, that's the plan 20:24:01 nice 20:25:06 ok any qs for jpeeler? 20:25:20 jpeeler if you can get a review up soon that would be great - so poeple can bikeshed it :) 20:25:31 just mark workflow to -1 20:25:38 yeah i know. it's just that i haven't had time to do much 20:26:04 jpeeler I intend to continue work on my patch so it would be nice if we could integrate efforts 20:26:37 sdake: no idea how you could run that patch. got all kinds of errors 20:26:50 ya its a work in progress for sure 20:26:55 I can integrate into your wip patch 20:27:10 lets just get the ideas in the review queue because atm there are 3 people interested in working on it 20:27:12 sdake: what i'm hearing is, hurry up. so i hear ya 20:27:24 and I want image create/start working prior to the 10th of may :) 20:27:54 rather image create --release without push and kolla start 20:27:59 that would be a good first step in the gate 20:28:10 if we could set that as our goal for may 10th that would rock :) 20:28:28 we have 3 people wanting to tackle the problem so lets get all our ideas in the review queue 20:28:33 jpeeler wfu? 20:28:48 yeah 20:28:52 nice :) 20:29:03 #topic kolla final milestone blueprint review 20:29:37 i have been working on #link https://blueprints.launchpad.net/kolla/+spec/cinder-container 20:29:39 #link https://blueprints.launchpad.net/kolla/milestone-4 20:29:45 trying to get cinder-volume to work 20:30:06 daneyon, same 20:30:11 nice - I htink that is probably goingto be master material 20:30:20 with a target of liberty 1 20:30:24 so i'll change that now 20:30:26 i made some progress, able to create a volume but can not attach to a nova instance 20:30:34 rhallisey, any progress on your end? 20:30:58 daneyon, I went though the docs again and I saw that I was missing a few things 20:31:08 I added them in, but still not working 20:31:16 I need to ask a cinder person 20:31:32 https://blueprints.launchpad.net/kolla/+spec/compute-operation-neutron 20:31:33 to try and getting a better understanding of what's going on 20:31:38 I think this is done right daneyon? 20:31:43 rhallisey, if you have time i would like to setup a webex and you can see what i'm seeing. 20:31:57 daneyon, ya that would be great 20:31:57 i don;t think it's a cinder.conf config issue 20:32:11 rhallisey please update the review with your new config items 20:32:12 lets coordinate after this meeting 20:32:20 so everyone is working from same tree :) 20:32:36 ya I have a feeling I'm not mouting all the proper directories 20:32:45 sdake https://blueprints.launchpad.net/kolla/+spec/compute-operation-neutron is done 20:33:25 rhallisey we'll talk after the meeting, i don;t think that's the issue either. maybe we can get this done if we combine our brains ;-) 20:33:29 daneyon mark it as such so you get credit in stackalytics pls 20:33:43 sounds good 20:34:25 daneyon this is done right? https://blueprints.launchpad.net/kolla/+spec/container-set-compute-operation-neutron 20:34:55 sdake. yes. i just marked as implemented 20:35:43 marking https://blueprints.launchpad.net/kolla/+spec/container-set-storage-operation for liberty 20:36:20 marking https://blueprints.launchpad.net/kolla/+spec/compose-multios for liberty as well 20:36:24 sdake can you wait till tomorrow to set as Liberty? 20:36:44 rhallisey and I may get it working in the next 24 hrs 20:36:53 ok can you changeit back then 20:36:55 I think we're close 20:37:01 I think this is the last issue before cinder is functional 20:37:04 ya 20:37:26 so major features for milestone #4 are neutron and possibly cinder 20:37:37 sounds about right 20:37:38 plus we did a whole lot of refactoring to made the code base tidier 20:37:50 and fixed a slew of bugs :) 20:38:11 next week we will make our announcement release notice 20:38:16 live on irc 20:38:25 ok 20:38:37 the process is I will branch stable/kilo 20:38:50 I will tag rc1 from stable/kilo master 20:39:08 we will test this week/next make sure we are golden 20:39:21 if rc1 is solid prior to the 10th I'll cut our final release 20:39:32 if we have updates to fix bugs after the branch, there could be an rc2/rc3 etc 20:39:39 each commit will result in a new rc 20:39:50 make sense to folks? 20:39:52 will rc1 fixes go directly into stble/kilo branch or get backported from trunk? 20:40:06 they get backported from trunk 20:40:11 ok 20:40:32 sounds good 20:40:52 #topic improving process goingforward 20:41:02 we do a whole lot of work 20:41:22 but its really difficult for me to track the work because we are not using implements blueprint and closes bug tags in the git log 20:41:41 * rhallisey is guilty of that all the time 20:42:04 i was wondering when that would come up 20:42:06 for liberty I'd like folks to -1 reviews any non-documentation changes that don't have a closes bug or implements blueprint documented 20:42:30 for documentation changes I don't think we need to document "improving quickstart guide" as the docs are a continual evolution 20:42:51 since this is a big change for how we operate, I'd like a unanimous vote from the core 20:42:54 +1 is my vote 20:43:11 it's time, +1 20:43:14 +1 20:43:17 it is 20:44:08 +1 20:44:15 cool well thats all the cores here 20:44:20 i'll sync up with mandre tonight 20:44:28 but lets assume he will vote +1 and operate under this model 20:44:38 just after liberty dev starts 20:44:44 i.e. may 10th and later 20:44:50 until may 10th proceed as you currently do 20:45:10 i'd like to point out daneyon does a real nice job of documenting his changes :) 20:45:14 so not everyone is guilty 20:45:20 but I'm guilty as well 20:45:23 the only way it works is if everyone suffers the pain :) 20:45:31 sdake thx 20:45:49 #topic open discussion 20:46:21 so for folks not staying up until midnight, we have a large contingent of apac folks who look to be devs 20:46:25 so we will likely get more help :) 20:46:26 yay :) 20:46:38 ya I've been talking to some of them 20:46:45 it would be nice ot somehow tackle sync'ing commits with kollaglue images during the L cycle 20:47:01 daneyon I agree 20:47:26 we may hae to do it out of band, since the ci jobs havea 1 hour limit and push takes 5 hours on my machine/network 20:48:07 are people not pushing image changes? 20:48:10 ok 20:48:27 jpeeler its more like we want a fresh push per commit 20:48:29 i am, but their needs to be syncronization 20:48:54 compose-multios will fix alot of the problems we have with overwriting other poeples images/etc 20:48:56 what if you push to latest before your review is merged and you intro a bug? 20:49:12 the image should be built locally and pushed by ci when the assoc patch is merged 20:49:21 I'd like to ask some questions around some of the design principles you guys have (probably in #kolla after this) so I can wrap my head around some things 20:49:34 xarses we will be there :) 20:49:43 jpeeler do u see where i'm coming from? 20:49:53 daneyon: yep! just hadn't really thought about it 20:50:09 the big problem is the gate has a 1 hour timeout 20:50:37 i personally wouldn't love per commit pushes. doesn't each push add a new fs layer to download? 20:50:43 we can setup a vm somewhere which reads the changes and commits them via a docker push 20:51:01 jpeeler we would push with --no-cache 20:51:20 which eliminates that problem 20:51:25 ah didn't know about that, cool 20:51:40 what if we pushed to our own registry, then that could help? 20:52:04 you mean setup a registry in openstack-infra? 20:52:06 sounds painful :) 20:52:10 ya 20:52:45 sdake can you help me understand what would be painful? 20:53:01 setting up a new service in openstack-infra - no idea how to do that 20:53:25 this is a conversation we need to have at summit 20:53:36 I guess we can explore the diff options in more detail when we get pass Vancouver 20:53:36 pehraps I can get the infra guys to give us a session on the topic 20:53:40 we could turn the problem upside down and create a git wrapper to checkout to a certain commit that works with image in the registry 20:53:49 sdake not a bad idea 20:54:32 another simpler option is to allow us a post-commit hook to push via infra 20:54:35 not sure if that is possible 20:54:40 but they have a "push tarballs" job 20:54:50 maybe that could be a psuh images job 20:55:24 daneyon: sdrake, can't you guys use auto-builds from the docker registry? 20:55:27 sdkae i was looking at docker-hook to do something like that 20:55:31 sdake, no r in my name :) 20:55:48 haha 20:55:56 xarses we investigated that early on, problem with that is we would need a repo per dockerfile 20:55:58 which is like 20 repos 20:56:00 do u go by sdkae too? 20:56:02 and we would have to keep that in sync 20:56:42 sdake: the last time i was playing with it, I was able to do a couple in diff branches, but it looked like multiple could be done in the same repo & branch 20:57:09 i think the best course of action is to investigate how post tarballs work 20:57:14 and do the same but with docker 20:57:33 maybe there is no timelimit on that 20:57:58 let me try to get a session from the infra guys 20:58:11 see if they have bandwidth for it in their schedule 20:58:34 if not I may be able to steal one of the magnum sessions 20:58:40 since its semi-related to magnum as well 20:58:50 magnum has 12 sessions 20:59:40 meeting time out thanks guy :) 20:59:41 s 20:59:43 #endmeeting