16:00:59 #startmeeting kolla 16:00:59 Meeting started Wed Jan 11 16:00:59 2017 UTC and is due to finish in 60 minutes. The chair is inc0. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:01:01 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:01:04 The meeting name has been set to 'kolla' 16:01:16 #topic rollcall, w00t please! 16:01:17 woot o/ 16:01:18 O/ 16:01:20 you know what to do 16:01:21 ;) 16:01:25 o/ 16:01:27 o/ 16:01:30 woot woot! 16:01:31 :) \Ô/ 16:01:35 o/ 16:01:37 0/ 16:01:37 o/ 16:01:42 hello 16:01:48 o/ 16:01:50 o/ 16:02:06 zhubingbing did a double wave :) 16:02:17 ;) 16:02:42 o/ 16:02:46 o/ 16:02:46 0/ 16:02:50 o/ 16:02:57 o. 16:03:01 o/ 16:03:48 #topic Announcements 16:04:18 1. We have voting open for new deliverables. Cores, please cast a vote, you should get link on your mailbox 16:04:39 any community announements? 16:04:49 just a heads up - if you don't get a voting link please ping inc0 16:04:56 its important that eveyrone have a vote ;) 16:05:06 well, in this case core team;) 16:05:10 meow meow 16:05:13 right - just core team will get link :) 16:05:18 core teams :) 16:05:29 wirehead_: woof! woof! 16:05:41 behave you two! 16:05:55 who are good boys? 16:06:05 ok, back to business;) 16:06:13 I guess no other announcements? 16:06:24 things pretty slow becaue of the holiday shutdown inc0 16:06:25 moving on 16:06:33 yeah, trust me I know 16:06:36 although we are tagging 3.0.2 of kolla 16:06:37 Oregon is shut down today 16:06:40 or i t has been tagged 16:06:42 i have one, newton 3.0.2 is release 16:06:44 it has been tagged 16:06:46 and also kolla-kubernetes 0.4.0 is tagging soon 16:06:51 hopefully its ready today 16:06:52 cool. 16:06:58 cool 16:07:02 Jeffrey4l you will have to submit hte erquest pls :) 16:07:09 we can add that to agenda i htink 16:07:12 Jeffrey4l: feature freeze for Ocata is in two weeks right? 16:07:15 sdake, my pleasure. 16:07:17 need some hub images for 3.0.2 16:07:19 egonzalez, yep. 16:07:24 ok, actually let's start with this one 16:07:27 around Jan 25. 16:07:27 and there's about 3 ps's about to go in to finish 0.4.0. 16:07:33 #topic releases, 3.0.2 and 0.4.0 16:07:56 my question is 16:08:08 does 0.4.0 need to be based on 3.0.2 images? 16:08:25 or this is just because gate lacking of new images? 16:08:36 inc0 my take is kolla-kubernetes *should* be building images from kolla repo and loading into a docker hub locally in the gate 16:08:55 you mean master? 16:09:02 inc0 however that is not the case with the images 16:09:17 I'd like master rather then newton for kolla-kubernetes yes, however, kfox1111 seems to want newton 16:09:23 seems kolla-k8s is using 3.0.2 hub.docker.com images 16:09:25 we should always be gating on master 16:09:26 there are some hacks in kolla-kubernetes to work around bugs in 3.0.1 16:09:33 we'd really like to get rid of them for 0.4.0. 16:09:47 inc0 for the short term, we need the 3.0.2 docker hub images 16:09:54 perhaps Jeffrey4l can build them and push them 16:10:00 since coolsvap is out of the office for the moment 16:10:08 Jeffrey4l said he didn't have the bandwidth. 16:10:13 ok 16:10:14 sdake, actually no. ( have a bad net in China ) 16:10:18 portdirect does but doesn't have a machine available. 16:10:23 kfox1111 i'll push them today then 16:10:31 k. thanks. 16:10:34 portdirect, promise to push them today. 16:10:36 but can't make a long term commitment to pushing consistently 16:10:39 thanks sdake 16:10:44 I'll push buntu 16:10:50 if you coudl do the centos-binary and ubuntu-binary first, then I can test sooner. 16:10:53 inc0 cool - splitting up work sounds great :) 16:10:53 no, we need to handle it automagically 16:11:02 at the end of the day 16:11:05 +1 to automation in the future. 16:11:07 inc0 agreed 16:11:10 I can help with that I think. 16:11:22 When our network better I will take the responsibility but auto is better ideal 16:11:23 we should talk to infra 16:11:38 if you are struggling - I'll have access back in about 4 hours 16:12:02 anyway, https://review.openstack.org/#/c/418550/ <- working on bringing fresh images to gates 16:12:03 i can definately push the centos images - I have a pushing vm just for i t 16:12:25 just can't commit long term to pushing or maintaining docker hub images because of time constraints with $$DAYJOB 16:12:54 no commitments 16:13:15 sdake: awesome thanks. yeah. long term it really belongs in the gate somehow. I have some ideas after talking with infra that might work. 16:13:26 inc0 that work looks relaly good 16:13:45 well, I hope to make it work sooner rather than later 16:13:53 kfox1111 what that work does is use master images (rather then newton) to gate kolla-kubernetes 16:13:59 this is how kolla-ansible operates until end of cycle 16:14:24 yeah, I'd like to have same logic for that across deliverables 16:14:36 otherwise explaining it to users will be hell 16:14:36 inc0 i think where that might struggle is xenial binary since i'm not sure they have a DLRN feature 16:14:37 (delorean) 16:14:56 why would that be an issue? 16:15:03 I mean we have xenial binary 16:15:03 kfox1111 do you have issue with that? 16:15:25 sdake: my plan was to get both working. 16:15:25 inc0 the reason is because kfox1111 stated above he needs 3.0.2 containers for ubuntu binary (the gate is voting iirc or will be soon) 16:15:32 newest stable and trunk. 16:15:35 then we can test upgrades. 16:15:40 kfox1111 cool - thats a fantastic idea 16:15:56 I'd be soft against voting on binary gates 16:16:01 kfox1111 so no conflict then with the idea? 16:16:06 as binary gates arent voting in kolla 16:16:09 so that ps is great, but I'll add another set of tests for it. 16:16:15 inc0 ya i feel same way - maybe nonvoting gates for some time 16:16:29 inc0: I agree with not voting on binary trunk. 16:16:30 so technically there is slight chance of have them polluted in tarballs.o.o 16:16:39 binary stable should always work though. 16:16:39 source gates are voting and should be stable at all times 16:16:50 yeah, not our fault most of the time 16:16:55 yeah. 16:17:00 inc0 source gates on deploy are voting? 16:17:04 for ansible? 16:17:31 we should talk about making kolla-kubernetes gates voting on kolla container changes too at some point. 16:17:38 no, but after registry happends there (https://review.openstack.org/#/c/413720/) I hope to make them such 16:17:46 inc0 sweet :) 16:17:53 Yeah, that's already regressed at least once or twice, kfox1111 16:17:53 speaking of which, review this change polease 16:17:55 it's important 16:18:10 Jeffrey4l, how long till it's no longer wip? 16:18:16 kfox1111 what we need there is cross repo gating 16:18:22 sdake: yeah. 16:18:27 will update it tomorrow. 16:18:36 kfox1111 i can possibly help implement that since Jeffrey4l has figured it out already 16:18:47 kfox1111 and its just a different set of scripts and gate code to run 16:18:55 sdake: I think the kolla-kubernetes gate should be reusable for that. just needs containers prebuilt before the run. 16:19:16 so review I proposed is to kolla-k8s 16:19:18 kfox1111 right - that is how kolla-ansible does cross-repo gating 16:19:20 and this is kolla-ansible 16:19:26 so both repos already have wip 16:20:13 anyway, keep an eye on these changes please, we can have voting gates before end of the year with them 16:20:36 you mean 2017 :) 16:20:45 release 16:20:49 omg, need caffeine 16:20:53 sorry 16:20:58 ya - that conufsed me - why i asked for clarification :) 16:21:10 let's move on 16:21:26 #topic Deprecation of Debian images during Ocata cycle, removal during Pike cycle 16:21:30 berendt, you're up 16:21:42 yes i want to revisit this topic 16:21:51 we already discussed the removal of debian last cycle and delayed it 16:21:58 is there any activity on the debian images now? 16:22:06 there are still not gate jobs 16:22:10 i see nothing 16:22:23 i already opened a review and a bug report for the deprecation 16:22:37 if you think it is fine to move forward i will open a vote on dev ml tomorrow 16:22:47 here is the thing 16:22:57 we agreed already we would deprecate if there was not further implementation 16:23:02 and i have not seen further implementation 16:23:09 nor have I seen anyone step up to maintain it 16:23:15 this means we do not need an other vote? 16:23:16 those were the two things we agreed upon 16:23:42 #link https://review.openstack.org/#/c/417874/ 16:24:07 berendt i think we agred we would vote on it again this cycle 16:24:14 although i don't recall 16:24:28 ok i will open a vote tomorrow 16:24:34 berendt, so to follow full deprecation path we should also send mails to openstack-dev and ops mailing lists 16:24:57 inc0: i will cc openstack-ops, this should be sufficient 16:24:57 berendt right - that is part of the process - especially the ops list 16:24:59 also, I'd say if we find volunteer to pick up the work and create gates, we shouldn't deprecate it 16:25:15 inc0: yes, but we already tried this last cycle. 16:25:19 here is what I'd suggest 16:25:28 send email to ops list, if the ops list responds with crickets 16:25:38 sdake: there's a chance the email will dig someone up. so lets leave the door open a crack for that. 16:25:39 then send vote to ml for core review team 16:25:41 yeah, so just a remider to them;) 16:25:46 kfox1111 agreed 16:25:52 but if no one shows up we do follow through. 16:26:02 yeah I'd hate to screw people who actually used it 16:26:08 berendt - i'd highly recommend doing them staged tho - email to ops list - wiat a week, then email to dev list 16:26:10 ok, mail to ops this week, mail to dev with vote next week if nobody rpolies 16:26:14 if ops say "NO WAY" 16:26:19 then we are stuck with it :) 16:26:28 and have a volunteer 16:26:51 if someone says no way to me it's equivalent to "I'll take care of these images" 16:26:53 ok i will do this tomorrow, we can revisit the feedback from ops next meeting and move forward with the vote after the next meeting 16:27:03 you can point that in email too;) 16:28:12 can we move on? 16:28:14 i think we can move forward to the next topic 16:28:24 #topic PTG planning 16:28:36 I wanted to leave it to the end, so we can keep doing it for rest of meeting 16:28:40 who will attend? 16:28:55 berendt inc0 has organized remote participation 16:29:07 berendt so hopefully eveyrone from the core review team can make it that can't make the travel 16:29:10 #link https://etherpad.openstack.org/p/kolla-ocata-ptg 16:29:17 yeah, we'll have teleconferences going 16:29:26 that's great 16:29:27 Nice 16:29:34 I'll be attending. 16:29:40 i'll also be there in person 16:29:52 mon-wed 16:30:00 you'll be able to disconnect me :) 16:30:11 i leave wed at some point, although I saw tripleo wants to have a collaboration effort 16:30:13 so I think this PTG will have more k8s sessions than ansible 16:30:24 depends on our decision 16:30:30 sdake: +1 to talking with tripleo 16:30:37 i will not be there. 16:30:42 kfox1111 unfortunately i can't change my flight 16:30:48 but personally think at this stage of project we need more talking in k8s 16:30:54 kfox1111 so if we do anything live it will have to be wednesday morning 16:30:55 sdake: thats ok. I'll be there all week, so I can do it. 16:31:00 kfox1111 cool 16:31:05 well, wed is common day I think 16:31:13 and I can try to extend my stay 16:31:15 inc0 wish that was clear from the schedule :( 16:31:19 so we'll see 16:31:37 yeah, I need to pick the brain of organizers regarding time/space 16:32:17 comon folks plz open the ptg etherpad :) 16:32:36 there is only 7 people on it atm, and we had a whoel bunch of waves at the meeting start :) 16:32:42 so far 5/5 is kolla-k8s;) 16:36:01 even if you can't make the ptg, please add your suggestions for sessions 16:37:00 SamYaple, around? 16:37:44 talk about flaunted? 16:37:51 egonzalez the red hurts my eyes :) 16:37:57 fluentd 16:37:58 egonzalez any chance you can change your colors :) 16:38:07 yeah, me too ;) 16:38:36 thanks egonzalez :) 16:40:59 rhallisey: i am 16:43:19 SamYaple, have you ever considered having multiple playbooks in kolla-ansible. A playbook for bootstrap, deploy, upgrade, ect... 16:44:03 or having one playbook with good use of tags 16:44:13 for easier browfield deploys 16:44:37 rhallisey, I did 16:44:54 inc0, what are your thoughts on the matter? 16:44:56 we did action variable because there is no way to share parts of roles 16:45:18 like, I don't want to have role nova-upgrade because it won't be able to share any tasks from nova-deploy 16:45:26 right 16:45:34 and in reality in 90% of cases they're just subset of deploy tasks 16:45:44 yup that's true 16:45:45 with potentially few extra upgrade specific tasks 16:45:56 rhallisey: thats supposed to be dont with tags 16:46:07 rhallisey: at least in the ansible world, kolla does it different 16:48:36 wow 19 people on the pad 16:48:36 like a record ;-) 16:56:00 ok looks like people are sort of done :) 16:56:10 yeah 16:56:13 I think we can refine this next meeting a bit as well - typically in the past we have 2-3 sessions on it 16:56:16 let's have few minutes of open convo 16:56:17 just sorting out the agenda 16:56:19 #topic open discussion 16:56:41 thanks inc0 for the agenda time - I know its expensive :) 16:57:13 yeah, this part of our meeting process is not ideal 16:57:41 maybe etherpad would be easier 16:57:57 zhubingbing is finishing fluentd, review/test are welcomed ;) 16:57:59 etherpad can be multi thread ;) 16:58:02 #link https://review.openstack.org/#/c/407392/ 16:58:02 zhubingbing ++ 16:58:12 yes 16:58:23 fluentd most log analysis work has been basically completed; 16:58:33 help me review it 16:58:36 zhubingbing is syslog sorted out? 16:58:47 yes 16:59:11 about fluentd is differnet 16:59:22 zhubingbing cool - i'll pull down the patchset today and give it a go before next meeting 16:59:37 we have many DocImpact bugs filled in launchpad:/ 16:59:40 i'd encourage others to do the same, so we can get this work merged - it seems pretty important 16:59:46 Flunentd syslog with heka there is a bit different 16:59:46 fluentd do not support syslog /dev/log socket and it listen on a tcp port. 16:59:46 guys 16:59:50 we ran out of time 16:59:54 please overflow to kolla 16:59:58 thanks everyone for attending! 17:00:00 thanks inc0 17:00:00 bye guys 17:00:03 #endmeeting Kolla