23:01:55 #startmeeting kolla 23:01:56 Meeting started Wed Apr 13 23:01:55 2016 UTC and is due to finish in 60 minutes. The chair is rhallisey. Information about MeetBot at http://wiki.debian.org/MeetBot. 23:01:57 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 23:01:59 o/ 23:02:00 The meeting name has been set to 'kolla' 23:02:02 first :) 23:02:06 #topic rollcall 23:02:09 o/ 23:02:16 hello 23:02:22 o/ 23:02:32 hi 23:03:14 this meeting slto always so slow 23:03:16 no Jeffrey4l__ ? 23:03:21 ya it is.. 23:03:28 he was burning the midnight oil with me 23:03:48 #topic Announcements 23:03:53 sdake, go ahead 23:03:57 I don't have much 23:04:10 ya we barely have a quroum 23:04:15 I guess peopel can read the logs 23:04:18 we can run though quick 23:04:27 no NNOUNXWMWNRA 23:04:31 no announcements 23:04:46 speaking off by one there :) 23:04:58 ha 23:05:04 alright moving on 23:05:05 i didn't update the agenda, but things we need to talk about are 23:05:05 #topic Liberty backport planning 23:05:29 soo 23:05:33 1) final schedule for summit ODS sessions, 2) final 2.0.0 tag and planning for 2.0.1 and 3) there wa soething elses 23:05:43 roger that 23:05:54 ya backports 23:05:56 inc0, do you have anything regarding this topic? 23:05:57 thats the osmething else 23:06:03 yes I acutally hace 23:06:42 so it seems one big patch is best way to go 23:07:01 inc0 according to dims? 23:07:07 according to dhellman 23:07:24 cool 23:07:26 wfm then :) 23:07:37 either that or we talk to infra 23:07:55 whats the infra option 23:08:02 don't know yet 23:08:09 do we want to mention this to the ML? 23:08:14 ok they have a ialing list 23:08:15 add it on to that thread from before 23:08:18 opentack-infra 23:08:34 lets use the infra miaing list if we ask them soemthing sepcific 23:08:42 i guess i'd like to know whaat we would sk 23:08:52 inc0 : that's not quite what I said. I suggested making your mitaka version work with liberty, too, but you didn't like that. Merging a bunch of commits from your mitaka or master branches into liberty is going to cause you problems, which cherry-picking will avoid. 23:08:54 if doug said superpatch was way to go 23:08:57 didi doug suggestconsulting infra? 23:09:35 dhellmann, well, sorry for misunderstanding 23:09:47 it'll cause other problems, namely not being able to get the stable tag if you want it 23:10:03 yes we want thestable tag 23:10:09 what I meant is coexisting liberty and mitaka will be really messy 23:10:12 inc0: nah, it's what we settled on as the best of a bunch of bad options but I didn't want anyone to have the impression I thought it was a good idea 23:10:21 howeer is tht dependent on all branches being well managed? 23:10:30 because going forwrd all branches will be wel managed 23:10:30 yeah, it is better of bunch of evils 23:10:35 I agree on that 23:10:42 sdake : backporting a bunch of feature work will make that unlikely. you'll have to check with tonyb about those policies, though. 23:11:01 dhellmann its a one time thing 23:11:12 we are never doing this again ever in the history of mankind 23:11:25 even in future of mankind;) 23:11:29 it woulndl't bee necessary except kolla 1.0.0 has a fatal flaw 23:11:51 sdake : you should read http://governance.openstack.org/reference/tags/stable_follows-policy.html and http://docs.openstack.org/project-team-guide/stable-branches.html 23:11:57 and then talk to tonyb before making your decision 23:12:15 dhellmann ack i'll red that 23:12:28 eventually we kill stable branches right? :) 23:12:48 well in any case, one big patch will cause least of a mess from a technical standpoint 23:12:53 sdake: I don't think so 23:12:55 tonyb is hte stable branches maintainer? 23:13:03 sdake: ACK 23:13:20 tonyb sorry dont htink so which 23:13:25 but we'll lose history but we can point to history of stable/mitaka, this will be effectively it 23:14:43 the problem is 1.0.0 has a fatal flaw whih causes data loss which we cant fix 23:14:54 without esentially bringing in the entire new ocde base 23:14:57 there is no surgical fix 23:15:30 in the future there will be no feature backports 23:15:39 only critical and high severity bug fixes imo 23:15:58 tonyb do yo uthink that owuld meet requirements for stable tag? 23:16:29 sdake: okay /me bogging dow the meeting probably isn't productive. Add me to the review/discussion and I'll come up to speed 23:16:47 tonyb ok we can dooffline if you like 23:16:52 tonyb, let me dig out ML thread 23:16:54 but we can bog down the meeting now - its ok with me :) 23:17:14 bogging down meeting might be better as we have core team here 23:17:20 right 23:17:28 this isn't a rathole form my perspective 23:17:28 okay 23:17:28 and I think everyone agrees that this is actually fairly important thing 23:18:18 yes, I feel this is important 23:18:30 the entire core team wants this done tonyb 23:18:35 tonyb, http://lists.openstack.org/pipermail/openstack-dev/2016-March/090813.html 23:18:46 there isn't a single detractor - unimous agreement 23:19:01 o/ sorry for late. 23:19:04 if i could spell after being awak for 35 hrs 23:19:37 Jeffrey4l__ all good 23:19:39 sleep isn't actually optional, you know 23:19:49 sdake: I sleep on the keyboard 23:19:53 https://review.openstack.org/#/c/299563/ tonyb this is review that makes all the changes required 23:19:53 yea, take care sdake 23:19:56 guys let's move on and let tonyb catch up 23:19:59 so tonyb 23:20:02 read thread 23:20:06 we will come back to this topic ok ? 23:20:10 our plan was to rebase this PS to stable/mitaka 23:20:11 if you don't mind being interrupted that is 23:20:19 * tonyb is reading the thread and the review 23:20:20 and somehow make stable/liberty have exactly this code 23:20:28 tonyb, thanks 23:20:37 ok next up.. 23:20:42 #topic final schedule for summit 23:20:50 sdake, do you have a link 23:20:53 * rhallisey looks 23:21:06 https://etherpad.openstack.org/p/kolla-newton-summit 23:21:13 #link https://etherpad.openstack.org/p/kolla-newton-summit 23:21:26 we have 4 fishbowl sessions 23:21:33 10 design sessions 23:21:40 wow, that's cool space 23:21:47 I offered to work withthe security team on one of our design sessions 23:21:56 so we really hve 9 free deign sessions 23:21:59 +1 to that 23:21:59 I have analyzed the votes 23:22:12 and I'll jut go bsaed upon the otes 23:22:38 if there are objections to that 23:22:41 let me know 23:22:44 if you have conflicts 23:22:46 let me know 23:22:56 so I can keep the important sessions conflict free 23:23:07 the full schedule is posted on the mailing list from ttx 23:23:34 sdake: the link has topics ? where is the info about speakers of those topics ? 23:23:46 these are design sessions 23:23:48 no speakers 23:23:52 vhosakot, design sessions aren't really talks 23:23:58 we all speak.. got it :) 23:23:59 we just sit by the table and argue 23:24:06 raretly hit each other 23:24:09 someone typically takes notes and everyone else talks 23:24:13 last time pual took notes 23:24:19 but this time I'll likely be doingthat 23:24:31 cool 23:24:35 hitting or taking notes ? ;) 23:24:44 vhosakot, notes :) 23:24:50 hitting with notebooks 23:25:11 ok cool that's covered 23:25:18 #topic final 2.0.0 tag and planning for 2.0.1 23:25:24 is it covered 23:25:24 wait 23:25:32 can we get a rollcall vote 23:25:40 what's the undo 23:25:41 #undo 23:25:42 to my proposed plan 23:25:43 Removing item from minutes: 23:25:45 there ya go 23:26:19 so biovusl yI am +1 to my own plan since its the best i'vegot :) 23:26:24 note I need to enter this data last week 23:26:30 so i need a decision now :) 23:26:43 +1 23:26:45 sdake, where is this proposed plan? 23:26:55 i just stated it in irc 23:27:07 see lines 32 and 33 23:27:19 and comments about avoiding conflicts in your schedules 23:27:28 sdake: what are we voting for again ? for this --> "I'll jut go bsaed upon the otes" 23:27:34 ah, about sessions 23:27:37 to avoid conflicts in yur schedules, I need to know when your schedules conflict with our ods sessions 23:27:48 vhosakot right 23:28:04 and that people will actually send me their scheduled conflicts 23:28:07 that didn't happen last time 23:28:15 and we had multiple core reviewers missing from critical sessions 23:28:50 sdake, can you reach out to Mirantis folks if they still want kolla-mesos sessions? 23:29:08 inc0 waiting on mail to the ml 23:29:24 ok 23:29:25 but at this point i'll schedule them if they are voted on 23:29:37 and if things change we can adapt those that are not fishbowl 23:29:40 fishbowls get locked in 23:29:51 sdake: when are the design sessions ? tue-fri ? 23:29:56 the other stuff gets burried ithe schedule 23:30:02 wed-tue 23:30:03 tuesdady is cross project 23:30:05 fri is meeting 23:30:07 meetup 23:30:10 wed and thur are our sessions 23:30:14 we he ful day meetup on friday 23:30:16 ah, true 23:30:25 sorry:( 23:30:34 no apologies needed :) 23:30:40 so, you are checking conflicts for which day ? wed and thur ? 23:30:45 right 23:30:53 we have presenttion with inc0 and elemoine_ on Wed 27 11:00am-11:40am 23:30:54 mail me personally 23:30:59 mail me 23:31:00 oh ok 23:31:09 asap :) i'll enter the data in tomorrow morning 23:31:17 when i recover from that webinar :) 23:31:46 sdake: send us the recording of the webinar 23:32:01 vhosakot, I'll link it at the end 23:32:14 coo, thanks! 23:32:17 ok lets go over tagging 23:32:27 sdake, what do where you gong to tag? 23:32:46 huh 23:32:48 I think you mentioned it earlier, I forgot though 23:32:56 you mean the releases repo? 23:33:01 we will cover that friday 23:33:04 of ods 23:33:18 releases repo training and gate training friday 23:33:26 no tagging M 23:33:33 oh 4/15 23:33:40 ok cool 23:33:42 i'll do that 23:33:45 but ya thats the deadline 23:33:47 this is 2.0.0 tagging right ? 23:33:47 right 23:33:52 anything to mention about 2.0.1 23:33:54 yes 2.0.0 23:33:57 here is the idea 23:34:24 we fix as mmany high and critical bugs in mitaka as possible by friday 23:34:32 we only ahe a few of those left that aren't already in master 23:34:45 the rest we can handle in a z stream 23:34:54 as in 2.0.1 in 1 month or so 23:35:00 we can have bugbash session in Austin 23:35:11 inc0 possibly 23:35:34 point is we hae 2 dys 23:35:38 to fix the critical/highs 23:35:49 i'll tag friday evening 23:35:55 stable/mitaka is looking good in my tests 23:36:24 same 23:36:36 if you are going to fix a medium/low thats fine too 23:36:48 Jeffrey4l__ you ohave a bunch of bugs in high 23:36:54 any you want to give up to distribute the load 23:37:01 sdake, got 23:37:53 #topic Open Discussion 23:38:00 M was a great cycle 23:38:00 tonyb 23:38:05 :) 23:38:09 its not over yet rh:) 23:38:18 roger 23:38:21 tonyb did you hae a chance to catch up to where we are ;) 23:38:21 Phew that's a lot of context to try and grok (pre-coffee) 23:38:30 sorry 23:38:32 if you need more time 23:38:34 sdake: np 23:38:36 we can provide it 23:38:49 sdake: Well we'll see *if* I grokked it. 23:38:49 but we want to get moving quckly 23:39:30 as soon as mitaka is done, we want to do the liberty backport in the sanest way possible that doesn't excludee us from governance tags 23:40:54 tonyb one approach is to try to repeat it, another is to query us on the details 23:41:02 tonyb or maybee there areother appraoches - up to you ;) 23:41:07 I think what you're proposing would invalidate stable:follows-policy 23:41:24 permanently? 23:41:45 I'm not sure I understand the difference from the current plan and "abandoning liberty" isn't that what you're doign by stealth? 23:41:55 sdake: nothing in permanent ;P 23:41:58 no we are keeping liberty 23:42:13 we are abanodning 1.0.0 as a dead tag 23:42:19 tonyb, since we're deployment tool, it's really about what version of openstack we deploy 23:42:36 and right now stable/liberty deploys liberty, only with critical flwa 23:42:41 idea is to tag 1.1.0 23:42:53 so we want to give operators a chance to deploy liberty without critical flaw 23:42:56 inc0: Yeah deployment tools lok kolla/osa require different thinking ;P 23:43:07 kinda, true 23:43:14 sdake: but *what* will that tag be ? 23:43:16 different on our part or governance erpo part? 23:43:23 some of deployment tools explicitly deploy N-1 version 23:43:32 like Fuel 23:43:33 tonyb the 1.1.0 tag will be tagged from the stable/liberty branch 23:43:44 we would like to deploy N, so liberty kolla deploys liberty 23:43:45 the stable/liberty branch will contain the current contents of mitaka made to work with liberty 23:44:44 only reason we go to this lengths is to provide ops good way to deploy liberty while keeping our "liberty deploys liberty" policy 23:45:21 ahh okat that's a point I was missing 23:46:12 so stable/liberty would install liberty BUT basically be the same kolla code as stable/mitaka + work to make it install liberty 23:46:33 yeah 23:46:41 and be tagged 1.1.0 23:46:45 correct 23:46:51 and this is one time thing mind you 23:46:58 inc0: sure 23:47:05 ya after this no more o that nonsense 23:47:05 mitaka will not be ever deployed by newton code 23:47:34 dependencies killed us :( (damn you ansible!) 23:47:54 tonyb so if it requires governance changes to handle deployment projects which are a special beast 23:47:56 i can do that 23:48:09 to unblock us from not being taggable 23:48:12 inc0: tangent .... how is this not a problem for OSA id it's ansible that's the root cause? 23:48:24 its docker thats the root cause 23:48:33 sdake: okay. 23:48:36 osa doesn't use docker 23:48:37 tonyb, there was bug in ansible 1.9 in docker module 23:49:02 that's not what sdake said 23:49:05 the issue is docker data containers don't work as advertised 23:49:05 and ansible refused to fix it in 1.9 branch, and 2.0 is not backward compatible and wasn't even released back then 23:49:16 inc0 is talking about a different side-beenfit :) 23:49:20 Ahhh "refused to fix" 23:49:27 another issue is docker add more new feature recently and backport the code is very difficult. 23:49:28 data conianers cause data loss 23:49:35 sdake, but reason we couldn't use docker 1.9 with volumes was because of ansible bug 23:49:43 inc0 oh right 23:49:53 okay. I think I'm gettign it now. 23:49:59 tonyb, by refused to fix I mean they said "it will be fixed by 2.0" 23:49:59 so we need to backport kolla_docker 23:50:03 which is sort of a feature 23:50:16 we wrote our own docker module in Mitaka;) 23:50:18 *can* it be fixed in 1.9? 23:50:31 we hae fixed it by writing our own docker module 23:50:38 yeah, by replacing it by our own code 23:50:40 ansible module to integrate with docker 23:50:49 Ahh and that's the main thing you're backporting 23:50:57 this requires massive changes to all of the playboosk 23:50:59 yeah, and that was a lot of work 23:51:07 and it affected a lot of our code 23:51:09 (to use this new module) 23:51:45 so instead of trying to get this logic out of mitaka (we tried and failed) we decided to make mitaka deploy liberty, and that worked without big effort 23:52:15 inc0: and that's basically https://review.openstack.org/#/c/299563 23:52:25 t-8 minutes 23:52:31 tonyb, correct 23:52:37 which shows up on master because in March stable/mitaka didn't exist? 23:52:43 yup 23:52:55 I'll rebase it to stable mitaka as soon as we tag 23:52:57 that patch isn't going into master 23:53:02 okay 23:53:10 no, it will not go to stable mitaka as well 23:53:16 it would go into stable/liberty 23:53:20 but it will represent code which we want to be stable liberty 23:53:37 after a megapatch is applied on the table/liberty branch 23:53:44 so at some point you'll do the equivielent of git co stable/liberty ; git reset --hard stable/mitaka ; git push 23:53:57 pretty much, yeah 23:53:58 and then add that review 23:54:29 it might require a couple extra reviews to enable centos binary as well 23:54:32 in fact that would be best if we could do that because that would retain git history 23:54:42 sdake: sure. 23:54:51 well yeah, and maybe something on this 23:54:54 but tha'ts besides point 23:55:34 another way we can do it is git co stable/liberty, git diff stable mitaka -> make massive patch out of it and merge it as one massive patch 23:55:38 dhellmann thanks for pointing u at tonyb ;) 23:55:43 but git history is lost 23:55:44 Okay I'm pretty sure I understand what and why you're doign it 23:55:55 don't hit us too hard:( 23:56:15 inc0: well you can tag the liberty branch with liberty-early-demise or similar to not loose the history 23:56:25 inc0: :) 23:56:50 well, history will be lost from liberty to mitaka 23:56:59 tonyb you mean i should tag prior to this backport 23:57:04 because patch will be massive and will not contain commits it was build by 23:57:26 3 minutes 23:57:37 t-3 minutes - may have to overflow into #openstack-kolla 23:57:41 tonyb, if you have time we can move it to #openstack-kolla 23:57:51 just want to link one thing 23:57:53 sdake: Yeah you need to tag stable/liberty (with early-demise or similar befoer you do the reset) 23:57:56 inc0 i have a meeting conflict but i trust you cn handle it 23:58:00 inc0: ACK 23:58:05 #link https://vimeopro.com/midokura/345kl392 23:58:11 ^ sdake's video 23:58:15 thanks rhallisey! 23:58:22 nice! 23:58:39 tonyb the one quetion is does that stop us from applying for specificc tags? 23:59:32 sdake: it doesn't stop you *applying* but I'd -1 your application for stable:follows-policy 23:59:41 sdake: while liberty was a thing) 23:59:54 thats 3 cycles? 00:00:08 0 minutes 00:00:18 sdake: that'll be decided at the Austin summit 00:00:28 ok let's flow over to openstack-kolla 00:00:30 ya les end meeting and overflow in openstack-kolla 00:00:34 * tonyb moves 00:00:38 thanks guys 00:00:41 #endmeeting kolla