16:00:01 #startmeeting cinder 16:00:02 Meeting started Wed Sep 17 16:00:01 2014 UTC and is due to finish in 60 minutes. The chair is jgriffith. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:03 although I hoped for none, as the agenda was still empty a few hours ago ;) 16:00:04 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:06 The meeting name has been set to 'cinder' 16:00:06 hi 16:00:11 hello 16:00:12 hi 16:00:19 Hey guys1 16:00:22 o/ 16:00:22 hi 16:00:26 hello 16:00:26 hey 16:00:37 hi 16:00:53 hi 16:01:02 Ok, let's get going 16:01:06 #agenda https://wiki.openstack.org/wiki/CinderMeetings 16:01:11 hi 16:01:19 hi 16:01:27 hi 16:01:34 DuncanT: thanks for putting the agenda together 16:01:44 DuncanT: you wanna drive since you did the work :) 16:02:03 Sure 16:02:18 I don't think I can do #topic magic though 16:02:34 FFEs - are we all done with them? looks like it 16:02:43 Just want to check none have been missed 16:02:49 you can probably #chair jgriffith DuncanT or something like that 16:03:03 o/ 16:03:12 bswartz: ? 16:03:32 bswartz: was just trying to be polite and not steal his topics 16:03:42 (there's a meetbot command to set multiple chairs -- can't remember it) 16:04:01 I'll assume nobody has any FFEs pending then? 16:04:04 DuncanT: so yeah, seems like it... IMO I'm done with Feature adds at this point 16:04:52 ...which segues nicely into the next topic. Are we going to -2 the reviews that are up that are new features? 16:04:56 Has the ffe etherpad been updated? 16:04:59 I am :) 16:05:24 timcl: I updated a few, then realised they were all merged and got bored 16:05:44 thx 16:05:47 Ok, -2s it is. 16:06:18 #topic submarine features 16:06:18 o/ 16:06:26 don't do it :) 16:06:29 nuf said? 16:06:45 jgriffith: Can you explain what that means? 16:06:52 :-) 16:06:56 jgriffith: just change the smiley face to a threatening one 16:07:01 then nuf said 16:07:03 Torpedo features? 16:07:14 smcginnis: it means don't try and submit a feature and disguise it as a bug 16:07:28 smcginnis: like "bug = my driver doesn't do consistency groups" 16:07:33 jgriffith: Ah, yeah, sounds bad. 16:07:39 https://review.openstack.org/#/c/119938/ 16:07:52 ^ this is a gray area -- a bug that can only be fixed by adding a new feature 16:08:15 As is NFS Security… 16:08:34 the link I posted was not NFS security, but.... yeah 16:08:41 >_< 16:08:45 YEah 16:08:59 bswartz: not gray IMO 16:09:17 bswartz: that's a bug 16:09:25 okay I agree 16:09:25 pools was the feature 16:09:38 was just looking for clarification 16:09:40 https://review.openstack.org/#/c/121820/ 16:09:46 we consider this a bug 16:09:46 anyway... we're all adults 16:09:49 jgriffith: And NFS Security? We have a bug on that but it is a large set of changes. 16:09:53 surely we can play nicely 16:09:55 but we are adding some information 16:09:57 jgriffith: if we are not required to submit a blueprint for new driver feature any more, it will be even harder to distinguish between them 16:10:00 glenng: that's a bug 16:10:16 I don't want somebody to decide that it's an enhancement and scold me :-) 16:10:17 glenng: but sadly I realize it will likely lead to additional bugs 16:10:20 *likes stepping on bugs* 16:10:35 tbarron: no scolding, just a -1 or -2 16:10:39 jgritffith: Yeah, good chance. 16:10:40 :-) 16:10:47 tbarron: and if somebody is wrong you can inform them of your position 16:11:22 ok... any questions? 16:11:23 kk, the current implementation is pretty broken IMO and it's all orthogonal to 16:11:29 anything outside our own driver 16:11:38 anybody have anything in the queue that might be questionable they want to discuss? 16:11:50 tbarron: don't know what you're referring to? 16:12:01 jgriffith: I have some bugs..dont thnk anybody questioned them 16:12:02 my little bug :) 16:12:08 enuf here ... 16:12:11 tbarron: what is "your little bug" 16:12:15 jgriffith: NFS Security work is a item I would like to touch on. 16:12:19 https://review.openstack.org/#/c/119938/ 16:12:24 DOH 16:12:27 wrong link 16:12:31 https://review.openstack.org/#/c/121820/ 16:12:31 https://review.openstack.org/#/c/121820/ 16:12:37 jinx 16:12:39 jgriffith: this one is already merged. can you help update the blueprint status? https://blueprints.launchpad.net/cinder/+spec/emc-vnx-direct-driver-cg-support 16:13:07 sigh 16:13:49 https://review.openstack.org/#/c/120407/ 16:14:02 https://review.openstack.org/#/c/119311/ 16:14:14 https://review.openstack.org/#/c/119340/ 16:14:25 https://review.openstack.org/#/c/120407/ is clearly a bugfix and fine 16:14:25 geesh netapp! 16:14:27 https://review.openstack.org/#/c/107693/ 16:14:29 alwasy the same 16:14:38 Pick ME!!! 16:14:42 hehe 16:14:53 What is with the parade of review links ? 16:15:13 We will get the reviews and decide how to proceed. 16:15:18 where do these bugs come from after ever feature freeze? something is clearly wrong with testing... 16:15:20 jungleboyj: Reaction to the threat of -2 ;-) 16:15:21 DuncanT: other two plzzz 16:15:45 avishay: bugs from customers 16:16:04 glenng: The NFS security issues are a bug, though I'd like to see it merged soon enough to get some testing before release, given the size of the change 16:16:08 navneet_: well, they're late 16:16:11 glenng: Lots of pending comments 16:16:18 navneet_: and you should at the very least keep on top of them 16:16:31 navneet_: just dumping them in gerrit isn't cool at this point 16:16:35 navneet_: they've all failed 16:16:35 avishay: People don't test until after FF. 16:16:42 navneet_: you could fix them? 16:16:42 DuncanT: Yes sir. I have a question on one item from review. All other items have been refactored. 16:16:42 jgriffith: they are thr since long...I asked guys to reiew them 16:16:48 glenng: :-) 16:17:00 navneet_: they've failed Jenkins 16:17:03 glenng: Awesome, keep up the great work! 16:17:10 avishay: I think the bugs are coming because we spent time on increased test coverage and we found more issues 16:17:11 jgriffith: thats recent auto merge 16:17:28 navneet_: 7 days ago on the one 16:17:31 navneet_: just fix it 16:17:34 jgriffith: I will rebase 16:17:35 navneet_: no need to argue 16:18:34 * avishay proposes a gate test for 100% unit test coverage on all new code + integration tests on new features for Kilo 16:18:46 Clear bugs we can continue fixing... anything questionable can be discussed in IRC 16:18:59 DuncanT: +1 16:19:18 avishay: That'll just mean we get crappy unit tests to increase coverage, and we've enough poor tests as-is 16:19:26 DuncanT: well put 16:19:28 ok.... 16:19:35 #topic Hot Issues 16:19:38 DuncanT: i guess 16:19:41 that reminds me of a question I had: who is in charge of targetting a bug to a milestone, and upto when? I'd love to read a doc/wiki regarding if any.. 16:19:53 Any hot issues that may not have bugs targetted to RC yet? 16:20:04 requirements.txt is ancient 16:20:14 i was looking this morning to see if any updates there are needed... not sure yet 16:20:27 navneet_'s bug wasn't on launchpad milestone page, a possible reason for missing attention 16:20:33 eharney: eharney ? 16:20:40 eharney: new version 5 days ago? 16:20:42 jgriffith: we haven't landed a requirements sync in two months 16:20:58 eharney: https://github.com/openstack/cinder/commits/master/requirements.txt 16:20:59 rushiagr: do we need to put bugs on launchpad milestone? 16:21:14 rushiagr: or somebody incharge? 16:21:29 jgriffith: https://review.openstack.org/#/c/119988/ ? 16:21:31 eharney: Sep 12, 11, 3, Aug 20, 18, 8..... 16:21:33 navneet_: I don't know for sure actually 16:21:37 https://review.openstack.org/#/c/106567/ ? 16:21:58 jgriffith: https://review.openstack.org/#/c/121885/ 16:22:05 eharney: sometimes those get "lost" I think 16:22:23 jgriffith: https://review.openstack.org/#/c/119988 looks valid to me 16:22:25 eharney: I just go off of github 16:22:27 jgriffith: Oh, that is targetted for RC though. 16:22:40 i.e. should we be blocking taskflow 0.4 16:22:56 eharney: let's back up a second please...... 16:23:19 eharney: do you or do you not see the merge from Sep 12? 16:23:45 mornin 16:23:56 jgriffith: yes, i wrote that patch to enable tests to pass for the larger update... 16:24:00 hemna: is here now. We can start! 16:24:15 eharney: so I'm confused by your statement that "we haven't updated in months" 16:24:26 jungleboyj, lol 16:24:28 jgriffith: we haven't done a full sync in months 16:24:39 jgriffith: just a couple of updates w/o examining a lot of them from what i can tell 16:25:21 how much this matters, i'm not sure, but there is a delta there we should probably examine before release time 16:25:33 eharney: fair enough 16:25:47 eharney: honestly those jobs are just auto +2/A as far as I'm concerned 16:26:48 eharney: but regardless... yes, we should be up to date on requirements files before RC1 cuts 16:26:56 * DuncanT really wishes there were reasons articulated for the version bumps, but that is a complaint for another day 16:27:07 eharney: I was just confused by the statement that "it hadn't been updated in months" 16:27:20 DuncanT: well, there is that 16:28:13 anybody have any bugs in mind that aren't just for their 3'rd party driver? 16:28:40 https://review.openstack.org/#/c/119564/ It was already approved, but I had to resolve a merge conflict. 16:28:42 "cinder-client --debug broken" is my only high priority bug 16:28:43 jgriffith, some of our folks here on the horizon team found a filed a few bugs yesterday 16:28:48 https://bugs.launchpad.net/cinder/+bug/1370273 16:28:49 Launchpad bug 1370273 in cinder "volume type-create allows only spaces as type name" [Medium,Triaged] 16:29:23 hemna: To my mind, not a bug 16:29:27 3 bugs related to that type of create call with a " " empty/or space name 16:29:41 hemna: Yeah, seems a little suspicious. 16:29:43 DuncanT, yah that's basically what I told her as well, but they say it affects Horizon 16:29:57 hemna: It's a free text field, user can put what they want in it 16:30:03 hemna: debateable 16:30:06 yup 16:30:10 that's what I said as well 16:30:11 hemna: Then horizon need to fix it, not cinder IMO 16:30:13 that one's hilarious :D 16:30:19 but thought I'd raise it here since you asked. 16:30:22 DuncanT: +1 16:30:50 well the only argument here is, shouldn't cinder be doing the validation, not Horizon? 16:30:59 I don't have strong feelings about it really. 16:31:02 xyang1: I'll review that CG type fix again now 16:31:07 I think Horizon should do the validation 16:31:13 jgriffith, you can mark it invalid if you like. 16:31:16 just my 2 cents 16:31:16 hemna: cinder doesn't have an issue with it :) 16:31:17 hemna: ' ' is a valid name 16:31:43 DuncanT: thanks! Looks that jgriffith just approved it. 16:31:46 it's stupid, but valid - agree 16:31:47 hemna: DuncanT guitarzan so this has implications for breaking behaviors that people currently expect 16:31:48 DuncanT, yah that's what I said as well. 16:32:10 so changing it this late in the game seems bad 16:32:11 hemna: Once you start validating things like displayable characters, unicode becomes a world of hurt 16:32:22 https://bugs.launchpad.net/cinder/+bug/1370280 16:32:24 Launchpad bug 1370280 in cinder "cinder create allows only spaces as display-name" [Low,Triaged] 16:32:28 https://bugs.launchpad.net/cinder/+bug/1370194 16:32:30 Launchpad bug 1370194 in cinder "QOS Spec creation should validate name and key-value pairs" [Medium,In progress] 16:32:31 are the other 2 16:32:38 basically the same thing with different commands 16:32:56 are these people anti spaces? we shouldn't be characterist here 16:32:59 hemna: All invalid bugs IMO 16:33:04 WTF, one bug per iteration? 16:33:11 duplicates and invalid 16:33:15 well they are different commands 16:33:17 at least for now 16:33:21 I'll mark them invalid 16:33:29 just wanted to see what others thought. 16:33:31 guitarzan: white spaces are characters too! 16:33:35 hemna: already did 16:33:36 jungleboyj: exactly! 16:33:44 jgriffith, ok thanks. 16:33:45 :) 16:33:51 hemna: but they are all the same basic issue so they should be dup's IMO 16:34:01 hemna: unless someobdy is getting ranked on number of bugs submitted :( 16:34:14 heh, one of them even has a patch to fix it already 16:34:15 metrics will be the death of us all 16:34:19 the qos bug 16:34:38 jgriffith, nah, just a bug happy horizon dev on our team. 16:34:52 we have a simple outstanding patch to help prevent docs job failures that we should probably land: https://bugs.launchpad.net/cinder/+bug/1368910 https://review.openstack.org/#/c/121224/ 16:34:54 Launchpad bug 1368910 in nova "intersphinx requires network access which sometimes fails" [Medium,In progress] 16:35:26 eharney: I meant to +2/A that one 16:35:41 eharney: Done 16:35:46 :) 16:36:17 hemna: I'm not necessarily object to changing it, just not right now 16:36:25 Shall we move on to a third-party update? We're circling a bit here now I think, can always come back at the end if we ahve time... 16:36:35 hemna: now... if you want to submit a patch that removes things like '/' I'm down with that :) 16:36:37 jgriffith, coolio. thanks for the feedback. I'll relay it. 16:36:42 #third party driver update 16:36:44 :P 16:36:48 bahh 16:36:56 #topic third-party ci updates 16:37:16 Nothing big here, just a ping to see if the status on the wiki page is up to date. 16:37:26 wiki page? 16:38:04 Up to date in regards to Dell. Still working on ours. Feel I am very close. 16:38:11 jgriffith: https://wiki.openstack.org/wiki/Cinder/third-party-ci-status 16:38:34 Seems like the focus right now is on bugs and reviews. I am fine with that. 16:39:11 'Failing tests' is the bit I'd really like people to keep up to date if possible 16:39:26 DuncanT: +1 16:40:04 jungleboyj: thanks, I've updated 16:40:12 * jgriffith system is now reporting 16:40:42 FYI, my setup doesn't report if it fails for "internal" crap 16:40:54 ie; devstack fails to deploy 16:41:01 jgriffith: Cool. 16:41:36 * DuncanT proposes K1 as a new hard cut-off date for getting CI working for anybody who has a driver in J (so new drivers get some slack) 16:41:52 DuncanT: +1 16:42:05 * DuncanT wants to avoid this slipping as a hard requirement indefinitely 16:42:20 we have a problem making the *.gz file browsable 16:42:21 jgriffith: DuncanT What is the preference if tests fail due to internal issues? 16:42:34 smcginnis: Fixing the internal issues :-) 16:42:35 Do we still comment to state that or just die quietly? 16:42:49 DuncanT: You are relaxing in your late release time. ;-) 16:43:19 smcginnis: Ah, reporting that you can't make a run is probably a good thing 16:43:39 smcginnis: Certainly you shouldn't report it as a -1 fail 16:43:48 infra gets angry if you make multiple comments 16:44:01 probably better to retry and complete the test, then comment 16:44:08 Yeah, if you're going to retry, don't comment 16:44:26 If you're giving up, a 0 vote to that effect is probably good 16:44:43 DuncanT: Good, that's what I was thinking. 16:44:52 DuncanT: just curious.... "why"? 16:44:53 * bswartz agrees 16:44:57 Voting -1 because your test rig blew up is the worst situation 16:45:06 DuncanT: agree with that 16:45:21 jgriffith: So that any reviewers waiting for a result knows to stop waiting 16:45:30 DuncanT: so here's a question..... 16:45:49 has ANYBODY ever waited for a review from 3rd party CI system? 16:45:57 * jungleboyj not it 16:45:58 or does anybody think they would? 16:46:11 not me...but checked results 16:46:15 DuncanT: not being contrary, just asking 16:46:26 Certainly not yet. 16:46:28 I did wait for our CI to validate the CG and pool aware patch 16:46:40 jgriffith: for my rbd changes, i would wait for the ceph ci to report 16:46:41 Personally I do my best to ignore the outputs 16:46:41 I check the results -- especially if a particular CI system has proven reliable in the past 16:47:01 just make sure nothing breaks existing functionality 16:47:02 bswartz: hmmm.... that's interesting 16:47:05 * DuncanT has wanted ceph reviews on some patches, easier than chasing for a manual inspection 16:47:20 bswartz: DuncanT ok, good enough for me 16:47:34 +1 non-voting is still valuable information 16:49:30 so for those that are using the infra modules.... 16:49:43 we should cnosider consolidating your changes and feeding them back upstream 16:49:56 should 3rd party CI check for openstack CI vote and recheck if it failed in their environment? 16:50:30 navneet_: in my case I don't even run if OpenStack CI hasn't run and passed 16:50:44 jgriffith: yeah thats a good check 16:50:44 navneet_: but I'm sure some will have concerns witht that approach 16:51:15 jgriffith: it avoids unneccessary run as the coder will reload 16:51:41 you could use the "wait for jenkins +1" as a load tuning knob -- wait if your job queue is long, don't wait if your job queue is short 16:52:07 jgriffith: checking might be useful incase of rerunning internal CI if it fails but openstack CI passed 16:52:24 navneet_: yes 16:52:26 bswartz: +1 16:53:04 Ok... anybody have anything else? 16:53:15 #topic open-discussion 16:53:46 Ok... 16:53:53 thanks everyone 16:53:57 #endmeeting