15:01:58 #startmeeting manila 15:01:58 Meeting started Thu Jan 4 15:01:58 2018 UTC and is due to finish in 60 minutes. The chair is bswartz. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:01:59 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:02:01 The meeting name has been set to 'manila' 15:02:03 hey guys 15:02:05 hello o/ 15:02:09 hello 15:02:09 hi 15:02:12 Hi 15:02:39 Welcome back from holidays 15:02:57 Or happy holidays for those of you still celebrating 15:03:33 I don't have any announcements for today 15:03:39 but I did have a topic to discuss 15:03:49 #topic Generic driver attach problems 15:04:10 tbarron: you asked me to take a look at some of the recurring failures in our generic driver tests 15:04:19 on a patch of mine that affects the generic driver 15:04:49 After looking at some job logs, I think there are issues with the nova/cinder attach detach 15:04:58 but I haven't found the smoking gun yet 15:05:16 jungleboyj: any chance you're arround? 15:05:46 What's up? 15:05:48 I haven't looked to see if the new cinder attach detach logic has been flipped on, but that's one possible source of our issues 15:06:34 jungleboyj: is the new attach logic enabled by default in cinder and/or nova yet? 15:06:56 bswartz: No. You have to request the new microversion to use it. 15:07:03 hi 15:07:11 OS_VOLUME_API_VERSION=3.44 15:07:19 Ok 15:07:30 And I'm pretty sure we don't do that anywhere 15:08:00 bswartz: Yeah, you guys shouldn't be using that. 15:08:06 jungleboyj: is there any chance the old cinder v2 API has altered behavior from how it used to be? 15:10:30 in any case, I need to reproduce the failures and track down the cause in an isolated envirornment 15:11:03 because I suspect that the generic driver is broken now, and it's non-voting gate status means we haven't cared 15:11:37 If anyone else has been playing with the latest versions of the generic driver and has seen issues please let me know 15:12:26 okay that's all I had on this topic 15:12:35 bswartz: I don't think anything there should have changed. All the the code changes were going into api/v3 15:12:48 is dustins around to talk about bugs? 15:13:07 he's on vacation till next week 15:13:29 tangentially related to the attach issues, caowei remarked on my HA spec in progress 15:13:32 doing something fun I hope! 15:13:36 https://review.openstack.org/#/c/504987/ 15:13:58 that the nova attach of cinder volumes adds extra challenges 15:14:11 for using the approaches outlined there with the generic driver 15:14:40 fun? if you can call being in New Hampshire (or anywhere on the east coast for that matter) fun at the moment 15:14:52 yeah for a long time the nova-cinder attachment process has been a serious limitation on the generic driver 15:15:28 the most serious problem is the limit of 26 volumes per nova VM 15:15:50 thinking about how to un-attach volumes from a fenced off SVM gives me headaches 15:16:05 yeah there are those issues too 15:17:05 right now though we seem to have a basic problem with out state machine management 15:17:29 manila is trying to detach volumes that cinder believes are always detached 15:17:47 it could be a race condition 15:17:58 s/always/already/ 15:18:20 Can we just 'try-to-detach || true' ? 15:18:54 Yeah I'm sure the fix will be more robust checking before/after the API call 15:18:55 Well, I guess we'd have to check the actual error from cinder ... 15:19:09 I just want to make sure I understand the sequence of events that got us into the bad state 15:19:14 ack 15:20:28 okay 15:20:33 #topic open discussion 15:20:44 We can table the bugs until next week 15:20:58 anything else to be discussed for today? 15:21:04 Just want to call awereness to lots of bug fixing activity whiile we were on vacation. 15:21:13 :) yeah.. 15:21:19 By folks on other continents, so we need reviews. 15:21:31 oh yes, thanks to those fixing bugs 15:21:32 For example, junbo li has been quite busy 15:21:36 I'm behind on reviews 15:21:54 https://review.openstack.org/#/q/owner:junbo85.li%2540gmail.com+status:open 15:21:59 but I want to spend time on reviews so send me a ping if you have anything waiting 15:22:15 I think all of us who came back to lots of email and piled up work are behind. 15:22:17 yeah he did great job 15:22:52 the other thing I'll point out is that we're just 3 weeks from the Queens-3 milestone 15:23:03 feature proposal freeze is next monday 15:23:04 And Jun has been busy as usual :) 15:23:25 :) 15:23:34 so it's time to wrap up any last features targeted for queens 15:24:09 So that deadline means blueprints need to be filed by then, right? 15:24:23 Or approved? 15:24:44 not blueprints -- the code itself 15:24:52 tbarron: both, patches need to be somewhat complete passing jenkins 15:25:10 bswartz: ganso thanks for the clarification :) 15:25:57 the last 2 weeks before feature freeze should just be reviewing/getting things merged (and any criticial bugfixes) 15:26:14 then we move on to more testing/bugfixing and docs fixes until the release 15:26:32 I haven't been following the discussion on the fate of rocky 15:26:38 * vkmc sneaks in late 15:27:06 i.e. whether it will be a 6 month or 12 month cycle 15:27:19 I thought it was going to be 12, but then ttx said that was in doubt 15:27:43 sounds like it will be 6 month ... 15:27:54 nothing's decided.. tl;dr: 12-month cycles aren't going to address the fundamental problem that the proposal was being made for 15:27:54 bswartz: 5 months is the most likely at this point 15:27:57 err 15:27:59 did an official announcement go out and I missed it? 15:28:00 6 months 15:28:01 :D 15:28:13 ok 15:28:13 review in progress : 15:28:13 ttx tries to raise blood pressure ... 15:28:28 https://review.openstack.org/#/c/528772/ 15:28:41 also I'll remind you that PTG is coming up at the end of Feb 15:28:53 Manila will be there 15:29:06 I'll try to close the thread early next week. 15:29:15 so please make travel plans if you plan to attend 15:29:23 i was quoting cdent's TC report.. https://anticdent.org/tc-report-2017-in-review.html#release-cycle-length 15:29:40 I don't have an etherpad yet, but later this month we'll get an agenda for the PTG started 15:30:11 bswartz: don't we have an etherpad? 15:30:19 bswartz: I remember seeing one 15:30:29 ganso: if we do, it was created by someone else 15:31:05 bswartz: maybe I am confusing it with cinder's 15:31:33 we did: https://etherpad.openstack.org/p/manila-rocky-ptg 15:31:45 Doh! 15:31:49 I did create this one 15:31:53 Back before the holidays 15:32:08 bswartz: post-holidays amnesia 15:32:08 * bswartz celebrated too much 15:32:37 Okay well I think that's everything for today 15:32:48 will the manila and cinder sessions overlap? 15:32:48 Thanks all! 15:32:56 I mean ptg 15:33:02 xyang: Minimally 15:33:12 I plan to attend remotely 15:33:37 xyang: we can setup a speaker phone of some kind for remote attendees 15:33:46 great 15:33:51 What I've asked for is time on Tuesday+Friday 15:34:01 I need to check for confirmation that we'll get those times 15:34:08 ok 15:34:31 #endmeeting