15:00:42 #startmeeting kolla 15:00:43 Meeting started Wed Sep 19 15:00:42 2018 UTC and is due to finish in 60 minutes. The chair is egonzalez. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:44 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:46 The meeting name has been set to 'kolla' 15:01:07 #topic rollcall 15:01:14 o/ 15:01:21 o/ 15:01:39 o/ 15:03:05 #topic announcements 15:03:56 Im planning to tag rc2 this week 15:04:26 a few changes are missing to merge due broken gates the last few days 15:04:43 now are green, please check https://review.openstack.org/#/q/project:openstack/kolla+branch:stable/rocky+status:open and https://review.openstack.org/#/q/project:openstack/kolla-ansible+branch:stable/rocky+status:open 15:05:00 any other announcement? 15:05:55 none from me 15:06:07 #topic Berlin summit 15:06:24 o/ 15:06:29 Topic propossal for berlin summit is open until 26th september 15:07:03 If have any topic want to discuss with devs and ops, please make a propossal 15:07:29 process is the same as call for presentations for a summit talk https://www.openstack.org/summit/berlin-2018/call-for-presentations 15:08:13 if the topic is related to kolla, add it to the etherpad list https://etherpad.openstack.org/p/kolla-forum-stein 15:09:13 other than this, are you guys planning attend the summit? 15:09:33 unsure yet 15:09:52 I will attend 15:10:16 no from me 15:10:18 hope we can get more people than in denver 15:11:40 #topic Denver PTG recap 15:12:34 I was not able attend the PTG until friday, probably mgoddard can help us more on this topic 15:13:13 etherpad list if want to take a look are at 15:13:14 PTG was pretty quiet for kolla. I was the only core present, and spsurya and egonzalez joined in when they could. There were a few non-cores in the room also 15:13:15 #link https://etherpad.openstack.org/p/kolla-stein-ptg-recap-priority-setting 15:13:50 I also could only join on thurs & friday 15:14:17 we had some useful discussions, but some where difficult when the person who proposed the topic was unknown or not present 15:14:52 so I think we need to make sure each topic has a proposer listed, and that they either attend or arrange for someone else to attend on their behalf 15:15:45 it seemed like the timezone & travel was the main issue with attendance, although there does seem to have been a drop-off in contribution during rocky also 15:16:06 will be interesting to see if this changes at the summit 15:16:47 these meetings are often not well attended, perhaps we need to shift to a more asia-friendly time? 15:16:56 for many contributors travel to europe and US is hard because of the VISA 15:17:33 thats true, meetings were already changed one hour before to be more asia friendly 15:17:35 yeah. I think we need to consider a fully virtual PTG next time 15:17:58 because the timezone still makes it difficult 15:18:21 will be at denver again right? but merged with the summit or a week before 15:18:35 yes, 3 day summit then 3 day PTG 15:19:21 the main discussions with outcomes were around testing 15:19:42 we have some volunteers to help improve CI testing in kolla - scenarios and upgrades mostly 15:19:55 this is important if the project is becoming more maintenance mode 15:20:52 also had some good discussions with GoDaddy people. I'd like to make sure we get their patch chain merged this cycle 15:21:02 agree, and to become more stable for non core services which probable there are many broken 15:21:12 first one is haproxy refactor, looking good: https://review.openstack.org/601648 15:21:55 there is a cross-community goal for python3, we should aim to start building py3 images 15:22:05 suggested to start with ubuntu & fedora 15:22:21 mandre already have a testing patch for py3 with fedora 15:22:32 ok cool 15:22:37 https://review.openstack.org/#/c/582290/ 15:22:41 there is another community goal to support upgrade checks, we should support these 15:23:39 placement will be split from nova this cycle, need to support that, hopefully will be a small change 15:24:11 spsurya is investigating dashboards, possibly leveraging something like ansible AWX or Rundeck 15:24:23 the placement change would be easier at first time, problem will came when the remove support to share the nova_api database and need data migration path 15:24:48 +1, I think they are planning to build a migration tool 15:25:05 I guess it would require downtime though? 15:25:32 that would be helpful, if not i would propose to build something cross deployment project to share the process 15:25:46 +1 15:27:01 yankcrime is planning to complete the DB backup/recovery work this cycle 15:27:57 I'd like to try to add support to kolla for generating enough information to configure a firewall 15:28:05 https://etherpad.openstack.org/p/kolla-stein-ptg-firewall-config 15:28:21 also to list which images need building 15:29:01 I think SamYaple was going to look into supporting LOCI images 15:29:14 I'm a bit meh on the above 15:29:47 egonzalez is planning to look at FFU 15:29:55 I heard there was feedback that kolla is not suiting users as they can't bring their own images 15:30:05 which is semi related to loci 15:30:30 pbourke what does (they can't bring their own images) means? 15:30:34 how many people have that issue? 15:30:40 mgoddard: that's what Im not sure 15:31:03 nvm, got it 15:31:04 egonzalez: because the start scripts are baked into kolla images, only those images can be used with kolla-ansible 15:31:21 which can make it seem less flexible 15:31:36 yep, users need to pass their dockerfiles to kolla build process 15:32:43 its something to think about, particularly if we're seeing lower rates of participation 15:32:55 anything that can bring more people to the project is good in my view 15:32:55 it's not something that bothers me, but would be interested to hear who it does bother 15:33:51 tripleo is using the start scripts too, no heard a complain about it 15:34:06 I'd like to know whether adoption is falling, or just contribution 15:34:18 perhaps they go hand in hand 15:34:23 if we change the behaviour of the images, is something that affect many projects, not only k-ansible 15:34:47 seems to be common to openstack - it's as popular as ever, but companies have moved resources to shinier projects 15:35:00 agreed 15:35:19 harder to justify maintenance than development 15:36:48 I'm not against supporting LOCI, but we have a large base of images and years of testing. It could add significant complexity to support both 15:37:50 anyways, that's probably enough PTG coverage, unless anyone has questions 15:38:37 i'm waiting for translations of the user survey to talk about it, but there are downstream projects which also uses kolla images with their own orchestrations on top, would be nice if some of them can join forum discussions and talk about it 15:39:54 thanks for the coverage mgoddard 15:40:51 Do you feel is time to set priorities/goals for stein with few people in the conversation? 15:41:54 0/ 15:41:55 Also, we does not have that many goals, if we prioritize some over others, may feel like some poeple work is not that important 15:41:57 we're still quite low on numbers. 15:42:10 most of the goals already have people assigned 15:42:59 it's mostly about saying what do we *need* to do in stein vs. what do we want to do 15:43:18 I don't mind 15:43:43 maybe we need to use the ML to manage priorities and goals? 15:44:15 would be nice to give everyone a chance to put forward a proposal and sign up for work 15:44:38 agree 15:45:08 +1 15:45:09 i'll do a resume of each goal and send an email 15:45:17 maybe open an etherpad for proposals, until next week's meeting 15:45:39 then agree goals & volunteers 15:45:43 more volunteers may be needed for testing scenario one to cover all services 15:46:41 https://etherpad.openstack.org/p/kolla-stein-ptg-scenario-testing 15:47:29 yep, that goal need to be splitted into smaller items 15:49:15 mgoddard thanks for handling the discussions in denver 15:49:23 #topic open discussion 15:49:33 egonzalez: np, did what I could 15:49:41 any other topic to discuss? 15:49:45 quick one 15:49:45 do we need to visit the core list? 15:50:23 pbourke go ahead 15:50:34 could people have a look at my patches to update rabbitmq? (https://review.openstack.org/#/q/topic:update-rabbitmq+(status:open+OR+status:merged)) 15:51:29 pbourke: put it on my list 15:51:33 the ceph jobs seem to be failing on https://review.openstack.org/#/c/584427/ and I haven't been able to figure out if its related. I've been testing with these patches for a month or so now and they seem good 15:51:59 bit of a pain I know but it would be good for us to get onto a more recent version of rabbitmq 15:52:20 some ceph jobs are broken due to recent bluestore changes, and we haven't had new kolla images published yet 15:52:25 maybe was when the bluestore labels were renamed and jobs failed until new images were pushed 15:52:26 it fixes a number of bugs, removes the deprecated clusterer, and has a much easier to use config format 15:52:52 thanks for that, all from me :) 15:53:12 one from me, could someone review these patch about rolling-upgrade,TIA https://review.openstack.org/#/q/topic:bp/apply-service-upgrade-procedure+status:open 15:53:17 pbourke thanks, is a good improvement 15:53:32 donghm_: will do 15:53:54 pbourke: thanks :) 15:54:02 i'd like to point out my revised backup spec as well: https://review.openstack.org/#/c/573311/ 15:54:06 pbourke: donghm_ will check 15:54:42 spsurya: yeah, thank you too 15:54:43 donghm_ yankcrime sure 15:54:45 it's been updated in light of ptg discussions about what operators may or may not want 15:55:06 as well as folding in previous comments 15:55:31 mgoddard regarding check core list, thats something i've been thinking for a while 15:55:54 there are definitely some who are no longer active 15:55:57 http://stackalytics.com/report/contribution/kolla-group/90 15:55:59 the polite method will be to send an email to non active core members 15:56:04 that is fairly useful 15:56:16 +1 15:56:32 then let them time to ask for core removal, if no answer in x days/weeks, remove them 15:56:44 it's also a good time to consider if anyone should be added to the core group 15:58:11 yes, agree 15:58:18 should we approach redhat to ask for more resources, given that they have a product using kolla? 15:58:23 mgoddard: +1 15:59:17 sometimes it is ok to ask for help - it is not always obvious that you need it 15:59:34 the TC can be helpful 15:59:56 will check that out 16:00:01 we're running out of time 16:00:07 thanks for comming 16:00:16 enjoy your day 16:00:17 thanks egonzalez 16:00:24 #endmeeting