16:03:44 #startmeeting cinder 16:03:44 Meeting started Wed Sep 4 16:03:44 2013 UTC and is due to finish in 60 minutes. The chair is jgriffith. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:03:45 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:03:48 The meeting name has been set to 'cinder' 16:03:55 any cinder peoples about? 16:04:01 hi 16:04:03 hello 16:04:04 * caitlin-nexenta waves 16:04:04 I'm here 16:04:06 hello 16:04:07 hi 16:04:10 hi 16:04:11 Nothing on the agenda yet 16:04:16 nope 16:04:23 Hello 16:04:29 I'll just give some updates and open things up 16:04:43 was hoping avishay would be around but apparantly he's off on vacation 16:04:50 #topic reviews 16:04:53 Can I get a quick summary on where we are with the freeze and any exceptions please? 16:04:59 DuncanT-: :) 16:05:05 so on that.... 16:05:12 (I've not been paying enough attention) 16:05:15 everything that's been granted an exception thus far is in process 16:05:34 there's nothing else that I'm planning on accepting especially at this point 16:05:41 a few things were snuck in, but that's ok 16:05:43 hi 16:06:08 what is the status of https://review.openstack.org/41868 ? 16:06:18 https://launchpad.net/cinder/+milestone/havana-3 16:06:33 So take a look at this list 16:06:46 these are the items we should be focused on for reviews today 16:06:52 jgriffith: do you like give some support for landing r/o-volume in nova side? 16:07:00 zhiyan: I can try 16:07:01 okay cool 16:07:19 jgriffith: cool 16:07:26 what moves blueprints from "needs code review" to "implemented"? is that a big red button jgriffith has on his desk or automatic? 16:07:36 * jgriffith currenlty 16:07:40 eharney: ^^ 16:07:51 eharney: seems that the launchpad bot isn't very good lately 16:07:59 eharney: I've been updating status 16:08:20 eharney: it is supposed to be automagically picked up by the bots 16:08:52 so any changes that are not in the 10 needed reviews on that LP page, please ignore for now :) 16:09:19 jgriffith: wanted to check on the FC plugin support..I had addressed pretty much all comments except the latest that came today morning from Navneet 16:09:19 We need to focus on the NetApp and Dell EQL changes 16:09:34 skolathu: I'm a bit concerned about your timing on that 16:10:39 jgriffith: Yeah, I know it got a little late. I had been working on the comments during last few days.. 16:10:50 I am clearing the new comments also now 16:11:10 Hope there won't be any other gating issues 16:11:30 Just a reminder today is the cut off for H3 16:11:46 thus the concern with rushing something like that through 16:11:59 skolathu: it doesn't seem like it's ready IMO 16:12:31 anybody else have thoughts? 16:12:41 hemna: DuncanT- eharney ? 16:13:18 hemna is driving into the office 16:13:20 jgriffith: I think I can clear all comments till now today. But are you suggesting there has to be more reviews? 16:13:20 any h-3 bugs need attention? 16:13:35 It still seems to be getting a lot of feedback to me 16:14:14 eharney: a few, however the priority is features 16:14:24 eharney: we have milestones for bugs still :) 16:14:25 56 comments on the latest review... that's enough to make me wary 16:15:02 DuncanT-: ? 16:15:09 jgriffith: we might as well punt that LVM volume clearing bug off of h-3 at least 16:15:11 DuncanT-: NM 16:15:18 jgriffith: The FC stuff 16:15:26 DuncanT-: yeah, I'm with ya 16:16:06 skolathu: I'm willing to see how it plays out 16:16:14 eharney: sure 16:16:47 anyway... my whole point was: The bp's that need review have to be priority today 16:16:56 there's a lot of complaining lately about review tims 16:16:58 times 16:17:16 I want to remind that submitters are responsible for timely turn around as well 16:17:19 it goes both ways 16:17:36 and it will help your karma considerably to do reviews on other peoples code as well 16:17:43 cinder is not a dumping ground for code/drivers 16:18:01 any questions/concerns? 16:18:22 # topic retype 16:18:27 #topic retype 16:18:34 jgriffith, DuncanT: I think the new comments are pretty straightforward and should be easy..I'll address them shortly and will see how it goes. 16:18:41 so I started working on retype command 16:18:55 but there's some conflicting views on what this should be 16:19:01 will retype require driver changes? 16:19:07 my thought was local retype, with no migration 16:19:30 avishay however submitted a patch that copies that but also makes it do a migration 16:19:35 kmartin: it would 16:19:42 seems to me that it's too late in the game 16:19:56 I agree 16:19:58 since there's disagreement on what it should do etc 16:20:20 and I'm also not sure about migration functionality as it is 16:20:32 did the same thing last release with config and half the drivers were broke out of the gate 16:20:35 * DuncanT- wants it to do everything. With unicorns on top 16:20:53 DuncanT-: quadricorns!!! 16:21:05 kmartin: fair 16:21:10 * DuncanT- starts on a blueprint 16:21:13 alright, I'm going to defer it 16:21:27 jgriffith: +1 16:21:56 #topic other questionable bp's 16:22:01 https://blueprints.launchpad.net/cinder/+spec/volume-acl 16:22:43 Personally I feel this introduces a good deal of un-needed complexity 16:22:54 I'd like to get other people to provide input and feedback 16:23:03 agree 16:23:15 how many people have a good understanding of this work at this point? 16:23:24 * DuncanT- doesn't any more 16:23:29 i know i don't 16:23:41 not me 16:23:43 ok, sounds to me like this one's out 16:24:06 agreed? 16:24:45 TBH I'd much rather wait and do a proper/clean integration via keystone 16:24:53 k 16:25:05 Makes sense to me 16:25:22 jgriffith: sounds good to me 16:25:28 Also makes sense to me. Let keystone/nova/whomever worry about this. 16:25:29 +1 16:25:48 kk 16:26:01 so from my perspective everything else on the list right now is a go 16:26:20 That give us 9 patches as the priority 16:26:43 anything people see missing from that list, or disagreements with what's there? 16:26:57 keep in mind the "add tests in test_db api" can be late if need be 16:27:14 changes to tests are part of feature freeze 16:27:21 s/are/aren't/ 16:28:39 Ok... guess that it then :) 16:28:43 one question 16:28:44 #topic open-discussion 16:28:48 eharney: go for it 16:28:55 what's the schedule/policy/etc around cinderclient changes at this point? 16:29:06 eharney: cinderclient is an open book :) 16:29:14 eharney: it's not tied to milestones 16:29:16 have someone use ceph? 16:29:22 good, because i'm probably going to propose something soon 16:29:31 thanks 16:29:39 eharney: I'll do a tag to pypi later next week and then problem another before final release 16:29:44 eharney: :) 16:29:49 topshare: say what? 16:30:41 so I want to spend today and the rest of the week just focusing on H3 16:30:59 we'll start up on some other things like the driver qual etc first thing next week 16:31:11 I'll get the code up and start talking to folks about some different ideas 16:31:22 anybody have anything else? 16:31:48 * DuncanT- would like to know if anybody considers themselves well versed in policy stuff? 16:31:54 DuncanT-: :) 16:32:09 (because I unserstand even less than I thought I did) 16:32:25 DuncanT-: yeah... sadly I thought I knew more than I do there as well 16:32:31 jgriffith: can you add this to you list: https://review.openstack.org/#/c/44455/ 16:32:48 For everybody's benefit the goal is to figure out how to disable extensions 16:33:11 some way other than just manually enabling all the extensions you want? 16:33:39 guitarzan: even that would be fine via policy file 16:33:51 i'm not sure about the context here (something with extend volume right?) 16:34:02 guitarzan: but I *thought* you could enable/disable specific actions in the volume_actions module 16:34:11 Specifically I want to disable extend and only extend 16:34:12 ah, we disable entire extensions just via cinder.conf 16:34:13 eharney: yes, be able to disable os-extend 16:34:19 I can't find a syntax to do it 16:34:25 DuncanT-: let me look 16:34:33 you may be able to do it via policy.json 16:34:39 Thanks all 16:34:41 set rule is_admin etc 16:34:51 https://bugs.launchpad.net/cinder/+bug/1220752 is the bug 16:34:53 Launchpad bug 1220752 in cinder "Can't disable extend_volume call via policy" [Undecided,New] 16:34:58 eharney: that's something I tried but didn't work out the way I'd hoped :) 16:35:08 eharney: I set a new role "disabled_exts" 16:35:23 eharney: and assigned that to the extension in the policy file 16:35:29 eharney: I may have missed something 16:35:35 jgriffith: some of the work we just did in Nova also tied whether the extension was enabled to whether you were coming in via publicURL API or adminURL 16:35:45 maybe another thing to consider but i'm still kinda fuzzy there on the exact details 16:35:50 nova's policy engine is smarter than cinders 16:35:53 eharney: ahhh... 16:35:54 ah 16:36:01 guitarzan: :( 16:36:21 there was a bug out there somewhere about porting Cinder to use oslo's policy code but i'm not sure where it ended up 16:36:21 I saw some of the syntax the cloud servers guys use and it doesn't work in cinder 16:37:03 eharney: it was a nightmare! 16:37:21 eharney: I started working on it, then somebody said they wanted to take it.. 16:37:21 :) 16:37:33 eharney: I think his name started with hem and ended with a 16:37:36 :) 16:37:36 it looks like you should be able to do policy out the volume api extend call... 16:37:55 Ok, let's take that offline and see if we can get it to work as expected 16:38:02 anything else? 16:38:05 Appreciated 16:38:18 just thanks for everyone reviewing stuff 16:38:25 indeed! 16:38:39 big thank you to EVERYBODY! 16:38:52 3'rd milestone is always hectic 16:39:03 but this one's going well I think with the soft freeze 16:39:20 Ok.... one more day of review focus please :) 16:39:33 remember turn around time on comments/feedback 16:39:59 and don't be crazy about typos in comments and such at this stage if you can avoid it :) 16:40:03 thanks everyone! 16:40:08 #endmeeting cinder