16:30:54 #startmeeting kolla 16:30:55 Meeting started Wed Oct 14 16:30:54 2015 UTC and is due to finish in 60 minutes. The chair is sdake. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:30:56 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:30:58 The meeting name has been set to 'kolla' 16:31:04 #topic rollcall 16:31:09 o/ 16:31:13 hello 16:31:15 hi 16:31:15 o/ 16:31:16 hi 16:31:18 \o/ 16:31:26 hi 16:31:28 hi 16:31:38 hi 16:32:06 hey folks looks like a full house 16:32:11 sure does :) 16:32:29 #topic liberty release announcement 16:32:42 ok time to break out the etherpad and write a liberty release announcement with all the goodies for liberty 16:33:11 #ink https://etherpad.openstack.org/p/kolla-liberty-announcement 16:33:18 o/ 16:33:23 i fell asleep 16:33:47 #link https://etherpad.openstack.org/p/kolla-liberty-announcement 16:36:00 shouldn't this be very similar to the l3 announcemnt? 16:36:17 i would think so 16:37:00 shouldn't it be "we made whole thing"? 16:41:28 i got bounced i guess 16:47:49 can anyone see this? 16:47:56 no 16:48:01 thanks harmw 16:49:54 ok looking pretty good ;-) 16:50:03 i will make the announcement tomorrow 16:50:22 so feel free to modify the release announcement prior to then if you want to make further changes 16:50:24 my irc is no good :( 16:50:32 #topic liberty release 16:50:37 SamYaple_ irc server imploded 16:50:52 so we hae some bugs remaining 16:51:05 i'd like to make a decision to either bounce them or fix them 16:51:10 lets go through them now 16:51:38 #link https://bugs.launchpad.net/bugs/1460129 16:51:38 Launchpad bug 1460129 in kolla "Cinder needs tgtd" [Critical,Confirmed] - Assigned to Sam Yaple (s8m) 16:51:44 cinder works with ceph 16:51:49 that is good enough for me 16:51:53 i havent been able to get tgtd to be stable 16:52:02 it works but crashes after a while unknown reasons 16:52:06 anyone opposed to bouncing? 16:52:08 do you want me to put a patch up with my tgt stuff that use to be there? 16:52:11 i hate tgt btw 16:52:15 rhallisey: no 16:52:20 that ran in the cinder-volumes container 16:52:43 basically tgtd needs to be its own container 16:52:51 but its just not working well 16:53:09 we ahe 24 hours to fix the remaining bugs 16:53:13 is this one we are going to fix? 16:53:21 I didn't have issues when it was in that contianer? Will it do harm to anything though? 16:53:53 rhallisey: i dont want to have cinder-volume no longer be a microservice 16:54:01 and tgt doesnt need to be running for ceph 16:54:03 #link https://bugs.launchpad.net/bugs/1492736 16:54:03 Launchpad bug 1492736 in kolla "TechDebt: add config section for heat for trustees" [High,Confirmed] 16:54:12 sdake: that ones all you :) 16:54:14 ok sam - focus dude :) are you going to have it fixed by toomororw 9am 16:54:15 well it's a matter if the trade off is worth it 16:54:25 sdake: nope 16:54:31 ok boucning it then 16:55:06 #link https://bugs.launchpad.net/kolla/+bug/1500699 16:55:06 Launchpad bug 1500699 in kolla "cinder: attach volume doesn't work" [Critical,Confirmed] - Assigned to Sam Yaple (s8m) 16:55:12 I would like to see that heat bug fixed 16:55:15 but I dont know how to fix it 16:55:24 ^ that would be fixed w/ tgt 16:55:27 is this cinder volume attach - does it happen with ceph as well? 16:55:41 nope 16:55:52 ok boucnign it 16:55:52 thats because we allow it to be deployed with lvm as a backend 16:56:13 #link https://bugs.launchpad.net/kolla/+bug/1489676 16:56:13 Launchpad bug 1489676 in kolla "Leftover /var/run files prevent Keystone start" [Critical,Triaged] - Assigned to Sam Yaple (s8m) 16:56:23 maybe 16:56:26 is this going to be fixed in the next 24 hours 16:56:35 basically apache2 just needs to not have a pid file 16:56:36 ok i'll leave it for now 16:56:43 ill tell you by tonight 16:57:02 #link https://bugs.launchpad.net/kolla/+bug/1504606 16:57:02 Launchpad bug 1504606 in kolla "Horizon instance console fails to connect instance with more than one controller" [Critical,In progress] - Assigned to Sam Yaple (s8m) 16:57:06 yes 16:57:25 #link https://bugs.launchpad.net/bugs/1504800 16:57:25 Launchpad bug 1504800 in kolla "kolla issue with multiple nova_consoleauth and HA" [Critical,In progress] - Assigned to Sam Yaple (s8m) 16:57:30 same as above 16:57:33 its teh same bug 16:57:36 are you assigned to all of them? 16:57:43 I didn't notice lol 16:57:44 #link https://bugs.launchpad.net/bugs/1505306 16:57:44 Launchpad bug 1505306 in kolla "Kolla should be using openstack/requirements to ensure more repeatable builds" [High,In progress] - Assigned to Paul Bourke (pauldbourke) 16:57:44 only the ones i have code for :P 16:57:46 ya SamYaple is superman 16:58:06 there wa ssome debate in teh review if this patch is correct 16:58:08 pbourke: are you here? 16:58:12 Yes 16:58:13 pbourke comments? 16:58:30 I would not merge the above for release 16:58:33 Too risky 16:58:39 ok boucning 16:58:44 agree. but it is needed sooner rather than later 16:59:03 Will split it up and merge piece by piece to master tomorrow 16:59:11 cool 16:59:23 need some kind soul to own https://bugs.launchpad.net/bugs/1492736 16:59:24 Launchpad bug 1492736 in kolla "TechDebt: add config section for heat for trustees" [High,Confirmed] 16:59:44 i dont think that can be done in 24 hours sdake 16:59:51 i think it may even need an ansible module 17:00:04 bummer 17:00:13 ok well i'd like someone to magicaly fix it soon so we can backport it 17:00:17 otherwise wawit conditions don't wor in heat 17:00:22 which meaks heat fairly unusable 17:00:35 youre the heat guy 17:00:41 i just poke it until it builds 17:01:12 as far as backports go 17:01:23 please get backports in for any bugs you have fixed 17:01:44 I have purposed all of the backports that have merged to master but not to liberty 17:01:50 built a oneliner for it 17:02:05 nice so its semi-automated? 17:02:11 yea 17:02:18 https://review.openstack.org/#/c/234843/ 17:02:30 that patch _must_ merge otherwise centos is a bust in liberty 17:03:18 to all - please take a look a t the review queue after the meeting and clean things up there 17:03:26 check 17:03:29 most are backports which have already gone through a thorough review and test cycle 17:03:38 Kk 17:03:56 #topic bare metal deployment 17:04:13 rob hirschfield has offered to make some of his tooling he has produced deploy kolla on baremetal 17:04:27 also dan prince has offereed to do something similar with tripleo 17:04:34 using ironic 17:04:39 another option is we could roll our own 17:04:43 how will that work? 17:04:48 docker image on baremetal? 17:04:52 is anyone interested in working with rob and dan? 17:04:53 or baremetal+docker 17:04:55 I dont know the details 17:05:01 yes i am very interested 17:05:01 don't we have session on that? 17:05:02 it involves working with the two engineers doing the work 17:05:19 inc0 they want to get started asap from what I can tell 17:05:21 anyway, count e in 17:05:24 sam i'll make an email introduction 17:05:27 and you can handle it from there 17:05:31 awesome 17:05:32 inc0 i'll also introducey ou as well 17:05:40 I know Dan 17:05:42 it is more then a one person job 17:06:22 #topic open discussion 17:06:33 rhallisey, didnt you already work with Dan on this? 17:06:43 liberty is really shaping up, fantastic work folks! 17:06:44 inc0, ya already doing so 17:06:46 over 400 bugs fixed 17:06:48 that is huge output 17:06:57 inc0, merged some stuff intro tripleo 17:07:08 I have two things to say 17:07:21 first thing, docker 1.8.3 and above is all broken 17:07:38 they changed that API without bumbing versions and broke the ansible docker module 17:07:55 the patches have landed upstream ansible to fix this _but_ we have to wait for a new ansible tagged release 17:08:08 so for all intents and purposes kolla only works with upto docker 1.8.2 17:08:22 additionally it requires ansible 1.9.2 17:08:29 any questions on that? 17:08:42 sounds legit 17:08:42 needs to be in the docs if not already pls 17:08:57 it annoys me tha tdocker did that 17:09:00 * sdake groans 17:09:10 i brought it up in docker-dev 17:09:13 they are aware 17:09:16 now 17:09:23 ok moving on 17:09:38 second thing is I have had several successfully ansible decploys in the gate 17:09:49 so we will have an ansible dpleoying openstack gate very soon here 17:09:57 all of the dependant patches for this have merged 17:09:58 \o/ 17:10:14 we will be openstack smoke-test 17:10:18 its very... touchy at teh moment and we have no logs 17:10:38 good news is we can get full container logs pretty easy for review 17:10:42 but thats a few patches aware 17:10:46 away* 17:10:57 now question for all of you 17:11:12 we have to do functional testing of the DEPLOYED openstack environemnt, how do we go about that? 17:11:25 we can write a bash script to do it.... but im open to ideas 17:11:43 is there no functional openstack testkit available already? 17:11:49 I was thinking if we could rig tempest to run against it 17:11:49 a long time ago we talked about tempest 17:11:58 tempest comes to mind indeed :) 17:12:02 will that test all the apis? 17:12:20 don't know 17:12:24 rally? 17:12:30 if we are doing tempet, i would suggest rally 17:12:31 i mean launching an instance should work, but we need to test the apis and connectivitey and all 17:12:32 isn't rally more for performance? 17:12:37 I think few clever rally tasks should be enough 17:12:45 Thats what i heard too 17:12:45 harmw: suprisingly it kinda evolved from that 17:12:50 hm ok 17:12:52 interesting 17:12:55 Oh 17:12:58 harmw: its great for just plain testing 17:13:03 but i'd personally like to see a smokescreen that kolla can use internally 17:13:05 it is, but it has few nice functional scenerios 17:13:09 +1 for using simple rally tasks for that 17:13:28 ok so we will look into rally for this 17:13:29 what if we'd just start out with rally 17:13:32 anyone want to head that up? 17:13:38 I will 17:13:43 cool pbourke_ you got the job 17:13:50 ill work on getting the logs out 17:14:01 I'll help with code reviews, I had some exposure to this project 17:14:16 SamYaple, rsyslog is a no-no? 17:14:16 final question from me. in the gate is everyone ok with deploying using ceph to excerise that code base as well? 17:14:26 ceph+1 17:14:31 inc0: its perfect! we just need to be able to view those logs 17:14:31 b bvgbvvvbv 17:14:33 yes ceph 17:14:46 yeah, I prefere ceph above lvm anyway 17:14:55 i mean ceph for nova and cinder and glance 17:15:17 we dont have the ability to do just cinder but not glance atm 17:15:37 I wonder if anyone would want that 17:15:41 so we wont be exersicing some non-ceph stuff like the file backend for glance 17:15:45 inc0: people are wierd 17:15:55 inc0: yes people do 17:16:05 i want it all 17:16:07 and I want it now :) 17:16:18 well im going to implement it with ceph 17:16:22 That stuff can be added later 17:16:22 no one spoke against it 17:16:39 use ceph 17:17:04 ya we will gate against ceph, but also make an attempt to ensure the other code paths work as well ;-) 17:17:08 awesome. well give it a few weeks for infra to give us our new gates and functional testing here we come 17:18:14 thats it from me everyone 17:18:17 oh wait! 17:18:20 Sam do you feel the --profiles patch goes against sdakes existing bp? 17:18:22 pbourke_: your profiles patch! 17:18:24 yea 17:18:27 :) 17:18:27 so no 17:18:30 i dont 17:18:40 i think that existing blueprint can be implemented as profiles 17:18:51 so a main infra and aux profile 17:18:56 between those everything builds 17:19:02 Sure that would be fine 17:19:07 sdake: ^^? 17:19:10 wfm 17:19:24 awesome. so we just need the ability to specify multiple profile pbourke_ 17:19:27 Need to tie it into ansible deploy profiles too? 17:19:37 As a separate patch 17:19:43 i dont think so 17:19:52 main and infra would build all ansible can deploy 17:20:19 Its the same use case though i think 17:20:30 what do you mean? 17:20:46 Build regexes are like ansible --tags 17:21:02 Would be nice to easily deploy a subset 17:21:27 we do that with the enable_service varibles 17:21:35 True... 17:21:40 if that variable isnt set you cant deploy that anyway 17:21:43 even with a tag 17:22:19 Ok sure dont worry about that part 17:22:42 so multiprofile support and were good 17:22:49 fixes it for everything i think 17:22:56 Cool 17:23:13 awesome. great idea on the profile thing pbourke_ 17:23:23 Thanks :) 17:24:04 ok well does anyone else have something do discuss or shall we return to #kolla? 17:24:45 nada 17:24:49 ok folk well thanks for coming :) 17:24:53 #endmeeting