15:01:11 <ttx> #startmeeting releaseteam 15:01:12 <openstack> Meeting started Fri Sep 1 15:01:11 2017 UTC and is due to finish in 60 minutes. The chair is ttx. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:13 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:01:16 <openstack> The meeting name has been set to 'releaseteam' 15:01:37 <ttx> Agenda @ https://etherpad.openstack.org/p/pike-relmgt-tracking -- scroll down to R+0 15:01:45 <dhellmann> o/ 15:01:52 <ttx> #topic Release postmortem 15:02:07 <ttx> so... I think it went well overall 15:02:21 <ttx> There was the issue with release announcements 15:02:33 <ttx> Anything else we shoudl change before next time ? 15:02:47 <fungi> what was the announcements issue again? 15:03:03 * fungi either wasn't paying attention or already killed off those brain cells 15:03:12 <dhellmann> fungi : the script to update the deliverable files for the final releases had a bug that set the wrong field so the announcements didn't include the full list of changes 15:03:14 <smcginnis> A minor type. 15:03:17 <smcginnis> *typo 15:03:18 <ttx> typo in the change-from entry causing bad start of interval ? 15:03:19 <smcginnis> Irony 15:03:21 <dhellmann> it was litterally _ instead of - 15:03:42 <fungi> aha, got it. i think i saw that fix blow by on wednesday but didn't pay it enough heed. thanks! 15:03:44 <dhellmann> although that points to a need for better schema validation for that yaml file 15:03:59 <smcginnis> I was a little surprised that wasn't caught. 15:04:16 <dhellmann> the field isn't required, and I thought the existing schema file would have caught it, but nope 15:04:29 <ttx> Anything else ? 15:04:55 <dhellmann> I don't think so 15:05:03 <dhellmann> very smooth, overall 15:05:28 <dhellmann> oh, well, we did have some late releases for libs, didn't we? 15:05:37 <ttx> Do we know of some projects which ended up shipping with a pretty old intermediary-release ? 15:05:38 <smcginnis> Yeah, was a lot more peaceful than I was expecting. 15:05:55 <ttx> that could be indicative of beingt dead/dying 15:06:05 <smcginnis> Hah 15:06:06 <dhellmann> I haven't looked at that 15:06:23 <ttx> yeah, it's not that easy to figure out from files 15:06:38 <ttx> before the branch we could sort files by date 15:06:52 <ttx> Probably a good exercise for the TC, not for us 15:06:58 <ttx> I'll have a look 15:07:08 * dhellmann looks at the intersection of the membership of both teams 15:07:12 <ttx> as part as my graveyard exercise 15:07:21 <smcginnis> dhellmann: Huh, big venn diagram. 15:07:22 <ttx> dhellmann: yeah, that's what I meant 15:07:40 <fungi> is the goal to catch releases which tagged all old/stale files, or just releases consisting of old-ish tags? 15:07:46 <dhellmann> I've wanted to put dates on releases.o.o for a while. I'll bump that up my low priority list a bit 15:08:03 <ttx> fungi: the goal is to spot dead projects before they start embarassing us 15:08:08 <fungi> or looking for projects which saw significant churn after their last release but failed to propose another one before the coordinated release? 15:08:33 <fungi> yeah, i guess all three of those scenarios is a possible indicator of some problem 15:08:42 <ttx> yeah, at least worth a deeper look 15:08:46 <smcginnis> ++ 15:08:54 <ttx> I'll run some analysis 15:09:03 <ttx> ok, next topic then 15:09:08 <ttx> #topic Tasks for trailing projects over the coming two weeks 15:09:27 <ttx> I reviewed them and spotted a few potential issues 15:09:53 <ttx> We have a grey area around trailing projects, as some follow milestones and some don't 15:10:05 <ttx> and it's hard to determine which is which 15:10:25 <ttx> I assumed they were all intermediary-released now, but that was a mistake 15:10:26 <smcginnis> Should we try to enforce for Queens that they at least follow one milestone prior to end of cycle? 15:10:32 <dhellmann> yes, that has been an issue in the past 15:10:42 <ttx> So for example the Gerrit ACL reset for cycle-trailing on R+2 15:10:46 <ttx> Only tripleo has a $PROJECT-release-branch thing 15:10:57 <ttx> the aclmanager command won't fix it for you 15:11:11 <ttx> since it only acts on cycle-with-milestones things 15:11:26 <ttx> can patch it, or we can fix it manually since it's only one team 15:11:40 <ttx> That said, Kolla should probably have a kolla-release-branch ACL 15:11:46 <dhellmann> for the tool to propose final releases, I added a separate mode flag for the trailing projects 15:11:49 <ttx> and a stable-maint team 15:12:09 <dhellmann> both kolla and tripleo have milestone-based versions going right now 15:12:23 <ttx> Since we'll be at the PTG, we could fix that all in a group exercise 15:12:36 <ttx> since I won't have much cycles before now and then 15:12:44 <dhellmann> makes sense 15:12:48 <ttx> and I'm the expert on the ACL dance 15:12:56 <smcginnis> I'd appreciate being able to look over your shoulder too. 15:13:00 <ttx> will be the occasion to pass down that knowledge 15:13:05 <dhellmann> maybe you can do the talking and someone else can do the driving, ttx 15:13:10 <smcginnis> ++ 15:13:11 <ttx> yap 15:13:20 * ttx adds to etherpad ptg 15:13:48 * fungi wonders if we could eventually drive that script from release jobs as well 15:14:07 <dhellmann> oh, I like that 15:14:32 <fungi> worth mulling over at ptg at least 15:15:09 <ttx> ok added 15:15:33 <ttx> I also used pretty relaxed lan,guage in my countdown email, compared to what those tasks show 15:15:49 <ttx> Assuming they were mostly intermediary and not needing a final patch 15:16:07 <ttx> I told them they could push additional RCs during the R+1 and R+2 weeks 15:16:21 <ttx> as long as they hit the deadline on Thursday of PTG week 15:16:22 <dhellmann> kolla hasn't done their rc1, yet 15:16:34 <ttx> The taskplan hints at a final addition deadline on R+1 15:16:54 <ttx> Given the number of projects on that list, and the fact that nobody cares if they are a day late... 15:17:07 <ttx> I think we can afford to close the list on Tuesday for a release Thursday 15:17:18 <ttx> (rather than close the list on the Friday before) 15:17:41 <dhellmann> didn't we have some confusion last cycle about these trailing deadlines? 15:18:01 * ttx looks up ocata-tracking 15:18:41 <ttx> Fuel proposed their final early Tuesday 15:19:00 <ttx> maybe that's what you mean ? 15:19:31 <dhellmann> maybe 15:19:47 <ttx> anyway, now that my email is out, if we want to be stricter we have to retract 15:20:45 <dhellmann> I don't have a problem being relaxed, as long as we're clear what the actual final date is 15:21:00 <ttx> I was clear on that date in the countdown email 15:21:14 <smcginnis> I thought the email was clear on dates. Hopefully there are no surprises. 15:21:23 <dhellmann> ok 15:21:28 <ttx> "need to post 15:21:30 <ttx> their Pike final release before the cycle-trailing release deadline 15:21:32 <ttx> (September 14)". 15:21:48 <dhellmann> that's clear :-) 15:21:57 <ttx> Basically on Sept14 we can produce the patch and process it 15:22:17 <ttx> but yeah, would be great to have brnaches for everything on R+1 15:22:45 <ttx> let's discuss that in the "unbranched projects" topic 15:22:51 <smcginnis> Only thought I had was to list the projects themselves, just so if they actually read the email, there's a better chance they see it and think "oh, that's me" and pay attention. 15:23:11 <ttx> #topic unbranched projects 15:23:29 <ttx> patrole IIRC was to be considered like tempest 15:23:35 <dhellmann> ok, I wasn't sure 15:23:37 <ttx> (last I asked) 15:23:43 <smcginnis> That's my understanding. 15:23:58 <dhellmann> I think I've already made a note that list-deliverables needs to skip over the branchless projects 15:24:04 <dhellmann> in this query, at least 15:24:07 <ttx> So we shoudl chase down kolla for RC1 and tripleo-quickstart for a stable/pike branch 15:24:15 <ttx> That should happen next week 15:24:33 <ttx> I should have listed/shamed them in the countdown email 15:24:35 <dhellmann> I'm more worried about the rc1 than the branch 15:24:46 <ttx> yep 15:24:59 <ttx> Addin that as a team action for next week to pester Kolla folks for a RC1 15:25:01 <smcginnis> Public shaming can be effective. :) 15:25:01 <dhellmann> at least for the projects not using pre-release versions 15:25:45 <ttx> task added 15:26:03 <ttx> Anything else on that topic ? 15:26:19 <dhellmann> nothing from me 15:26:23 <ttx> #topic PTG prep 15:26:29 <ttx> So we have an etherpad up 15:26:36 <ttx> #link https://etherpad.openstack.org/p/queens-PTG-relmgt 15:26:56 <smcginnis> It's a good start. 15:27:06 <ttx> I think we should cover that agenda on Wed-Fri in some room cprner, rather than try to get something done in Mon-Tue 15:27:23 <dhellmann> yes, my monday and tuesday are full of other things already 15:27:32 <ttx> Let's use our limietd free time on Mon-Tue to field other team's questions 15:27:37 <ttx> (in the "helproom" 15:27:38 <smcginnis> Sure, works for me. 15:27:38 <ttx> ) 15:28:01 <dhellmann> i'm committed for wednesday morning, too, but free after that 15:28:23 <ttx> OK, let's see how other things are being scheduled before we make a call 15:28:31 <smcginnis> I have Cinder and wanted to drop in for a bit with Glance, but this will have priority. 15:28:33 <ttx> I feel like we'll likely need half a day 15:28:39 <fungi> i'll hang out in the relmgmt helproom monday/tuesday when i'm not busy in the infra, qa, stable or requirements helprooms ;) 15:28:40 <ttx> a day max 15:29:03 <lbragstad> fungi: you have a lot of rooms to be in :) 15:29:18 <fungi> they're conveniently close to each other 15:29:22 <smcginnis> fungi: It's all one room, right? Rotate chairs every few minutes? :) 15:29:36 <ttx> The way I see the Mon-Tue helproom is that we should have some presence to answer questions that project teams may have about "infrastructure" in a large sense 15:29:46 <ttx> Obviously the infra team will field zuul v3 questions 15:29:58 <fungi> if something comes up wednesday through friday, i'm happy to run over to wherever stuff is being discussed that could use input, just give me a heads up 15:30:01 <ttx> It's OK if only one of us is present in the room at any time 15:30:19 <ttx> and even if nobody is present, if someone can redirect the questions to a time when we'll be present 15:31:01 <smcginnis> We can also ping in the -ptg channel if someone's presence is required. 15:31:05 <fungi> across those 5 teams, i expect that there's someone from one of the other 4 who knows the answer to >90% of teh questions anyone's likely to ask anyway 15:31:11 <ttx> dhellmann: we could also (ab)use the room for an informal presentation on release management (annoucned by ptgbot) for interested projects 15:31:33 <smcginnis> I need to give one of those internally. 15:31:41 <dhellmann> we could 15:31:57 <ttx> like how we drive things from the releases repo, the different models, etc. Paraphrasing the project-team-guide chapter 15:32:12 <smcginnis> Any existing material out there I could steal^wborrow? 15:32:15 <ttx> Not necessarily a recruitment talk, just a level-set talk for new liaisons and stuff 15:32:19 <dhellmann> ttx: were you thinking https://github.com/dhellmann/presentation-openstack-release-team-onboarding ? 15:32:27 <dhellmann> ah, ok 15:32:42 <smcginnis> dhellmann: Oh nice, I missed that one. 15:32:48 <ttx> I was thinking more about https://docs.openstack.org/project-team-guide/release-management.html 15:32:59 <dhellmann> yeah, that would make more sense, probably 15:33:21 <ttx> Reality is, not so many people understand that stuff as well as they think :) 15:33:30 <ttx> Reno could use some publicity too 15:33:47 <ttx> like "openstack is better if you get into the habit of using reno extensively 15:34:29 <dhellmann> yeah, I'll see if I have some time to put together some notes for something about that 15:35:00 <ttx> Can be prett informal / workshoppy 15:35:21 <ttx> I fear that if we wait for people to ask questions we might never get one 15:35:45 <dhellmann> having a few topics ready for discussions on tuesday seems like a reasonable way to go 15:35:46 <ttx> Such a "presentation" is one way to answer FAQ 15:36:09 <ttx> also we can do it when we end up with a hole in our busy multitasking Mon-Tue schedule 15:36:19 <dhellmann> I've been meaning to write a blog post on reno anyway 15:36:25 <ttx> if we do it in the helproom itself it doesn't require that much booking of space or anything 15:36:41 <smcginnis> No projector though, right? 15:37:03 <ttx> OK, added a "Presentations / explanations we could give in the infra helproom on Monday-Tuesday" section to the etherpad 15:37:09 <ttx> smcginnis: actually most rooms have AV 15:37:15 <smcginnis> Oh great. 15:37:18 <ttx> (integrated AV from the hotel) 15:37:30 <ttx> (that's one of the reasons why we picked this hotel actually) 15:37:31 <smcginnis> I think folks will appreciate that. 15:37:40 <ttx> I think /some/ rooms won't have it 15:37:48 <ttx> but /most/ rooms supposedly have 15:37:56 <smcginnis> Better than the very few in ATL at least. 15:38:01 <ttx> that's why we haven't been saying "there will be AV everywhere" 15:38:23 <ttx> Pretty sure the large Infra helproom will have AV 15:38:30 <fungi> exciting 15:38:34 <ttx> fungi: something we could check with Erin I guess 15:38:52 <fungi> sure, i'll ask her now 15:39:06 <ttx> Last thing on PTG, we should probably organize a RelMgt + friends dinner 15:39:13 <smcginnis> ++ 15:39:22 * smcginnis is going to gain so much weight in Denver 15:39:28 <ttx> our small group + tonyb, prometheanfire (requirements and stable folks) 15:39:54 <ttx> swpanil maybe 15:40:01 <ttx> swapnil* 15:40:27 <ttx> Those teams are just too small to have a "team dinner" so I think we should include them. They work so close to us anyway 15:40:33 <dhellmann> ++ 15:40:37 <dhellmann> that worked ok in atl 15:41:30 <ttx> smcginnis: want to handle that one ? I'm trying to organize the TC dinner already :) 15:41:43 <smcginnis> ttx: Yep, got it. 15:42:01 <ttx> https://wiki.openstack.org/wiki/PTG/Queens/CityGuide will hopefully have lots of suggestions next week 15:42:21 <ttx> So far only reflects my very limited Denver knowledge 15:42:37 <smcginnis> We have a few locals that should be able to help fill that out. 15:43:15 <ttx> I have great suggestions for small places, but they are likely to not be happy if we show up in parties of 7+ 15:43:19 <dhellmann> lyft is going to be very busy coming and going to the hotel that week 15:43:51 <smcginnis> We need the armada of shuttle buses we had in Paris. 15:43:51 <ttx> Yes, beyond Station 26 everything else interesting seems to be 45-min walk away 15:44:43 <ttx> Anything else on PTG prep ? 15:45:03 <ttx> #topic Meeting / countdown email handover 15:45:15 <smcginnis> \o/ 15:45:20 <ttx> So that was my last countdown email and meeting leading 15:45:36 <dhellmann> thanks for a smooth cycle, ttx 15:45:52 <ttx> I'm traveling next Friday, so will miss 1st meeting then 15:45:59 <dhellmann> and big thanks for stepping up, smcginnis 15:46:04 <smcginnis> ttx: IIRC, you suggest a one time direct email to all PTLs just stating how the release communication will go, then just the weekly ML posts? 15:46:08 <dhellmann> do we need to meet next week? 15:46:16 <smcginnis> dhellmann: Glad to. 15:46:29 <ttx> No need for a countdown email next week, I posted for R+1 and R+2 15:46:37 <smcginnis> I would guess there will be a lot of folks travelling. We could probably skip since we will have all the face to face time. 15:46:43 <smcginnis> meeting, I mean 15:47:05 <dhellmann> that's what I was thinking, smcginnis 15:47:14 <ttx> smcginnis: that's how we did it past cycles. First an email explaining how things will be done (you have received the Pike one as PTL), then ask PTLs to formally ack it, then use ML 15:47:39 <smcginnis> ttx: I was checking if I still had the pike one but apparently don't. 15:48:02 <smcginnis> ttx: I remember it was informative. If you still have a copy, I'd love to take a look again and follow suite. 15:48:09 <ttx> that was openstack-dev + cc all PTLs apparently, let me find it 15:48:27 <smcginnis> Yeah, couldn't find it on the archives, but I can look some more. 15:48:31 <ttx> http://lists.openstack.org/pipermail/openstack-dev/2017-March/113303.html 15:48:40 <smcginnis> Dang you're quick. 15:48:49 <smcginnis> Thanks! 15:48:58 <ttx> Nobody can beat my crazy firefox shortcuts 15:49:18 <smcginnis> re: meeting, I will send out an email stating no meeting next week if there are no objections. 15:49:24 <dhellmann> ++ 15:49:44 <ttx> ++ 15:49:54 <smcginnis> On my todo list. 15:50:03 <ttx> #topic Open discussion 15:50:07 <ttx> Anything else ? Famous last words ? 15:50:27 <fungi> #link https://sks-keyservers.net/pks/lookup?op=vindex&search=0x4c8b8b5a694f612544b3b4bac52f01a3fbdb9949&fingerprint=on Queens Cycle signing key 15:50:46 <fungi> i'll push up the changes to swap that into production shortly after the ptg (once cycle-trailing releases are completed) 15:50:58 * dims peeks 15:51:18 <fungi> also erin is looking up a/v situation for our ptg helproom now 15:53:29 <smcginnis> I don't have anything else. 15:53:34 <fungi> but don't hold up the #endmeeting for that, i can update everyone in #openstack-release once she gets back to me 15:53:38 <ttx> no big deal if we don't have AV in that room, since we probably won't have slides anyway :) 15:53:50 <ttx> alright then 15:53:56 <ttx> Have a GREAT weekend 15:53:56 <fungi> we can borrow soem cat pics from mordred 15:54:01 <smcginnis> :) 15:54:03 <ttx> #endmeeting