15:00:45 <bswartz> #startmeeting manila
15:00:46 <openstack> Meeting started Thu Aug 10 15:00:45 2017 UTC and is due to finish in 60 minutes.  The chair is bswartz. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:47 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:49 <openstack> The meeting name has been set to 'manila'
15:00:53 <gouthamr> hello o/
15:00:56 <bswartz> hello all
15:01:00 <dustins> \o
15:01:01 <ganso> hello
15:01:03 <xyang2> hi
15:01:04 <amito-infinidat> hello
15:01:06 <vponomaryov> hello
15:01:08 <markstur> hi
15:01:10 <zhongjun> Hi
15:01:39 <bswartz> tbarron cknight toabctl: courtesy ping
15:01:41 <jungleboyj> @!
15:01:41 <_pewp_> jungleboyj ( ´ ▽ ` )ﾉ
15:01:45 <toabctl> hey
15:01:48 <cknight> Hi
15:02:04 <bswartz> #topic announcements
15:02:14 <bswartz> The RC1 target date is today!
15:02:26 <bswartz> I think we're close to hitting it
15:02:50 <bswartz> we've had some fixes merge, and I booted out some low priority bugs without fixes
15:03:10 <bswartz> in fact RC1 is the only topic on the agenda, so let's get started
15:03:21 <bswartz> #agenda https://wiki.openstack.org/wiki/Manila/Meetings
15:03:26 <bswartz> #topic Bugs / RC1 Status
15:03:35 <bswartz> #link https://launchpad.net/manila/+milestone/pike-rc1
15:04:08 <bswartz> so first let's discuss the pair of quota bugs
15:04:16 <bswartz> #link https://bugs.launchpad.net/manila/+bug/1707378
15:04:17 <openstack> Launchpad bug 1707378 in Manila "Quota usage value error in batch create share" [High,Incomplete] - Assigned to zhongjun (jun-zhongjun)
15:04:21 <bswartz> #link https://bugs.launchpad.net/manila/+bug/1707379
15:04:23 <openstack> Launchpad bug 1707379 in Manila "Quota usage value error in batch delete" [High,Incomplete] - Assigned to zhongjun (jun-zhongjun)
15:04:44 <bswartz> and this proposed fix:
15:04:47 <bswartz> #link https://review.openstack.org/#/c/489501/
15:05:12 <bswartz> it sounds like we still have race conditions in the quota code
15:05:30 <bswartz> tbarron brought up this issue in Austin
15:05:41 <zhongjun> yes, we still have many race conditions in the quota
15:05:49 <bswartz> it appears tbarron is not around this week though
15:06:14 <bswartz> zhongjun: can you describe the approach for fixing this?
15:06:50 <zhongjun> bswartz: We record the wrong usage quota (shares, gigabytes) value
15:06:51 <zhongjun> when we delete one share many times at the same time.
15:06:51 <zhongjun> Because we reserve -1 shares quota before we delete the
15:06:51 <zhongjun> share record in db, then we can continue to get share info
15:06:51 <zhongjun> from db and reserve quota -1 many times.
15:07:14 <zhongjun> bswatz: Changed to remove the quota reserve to the place that after
15:07:14 <zhongjun> the share record has been deleted in db.
15:07:17 <zhongjun> now
15:07:20 <bswartz> is it a race condition or a cleanup error?
15:07:47 <vponomaryov> looks like the latter
15:08:04 <bswartz> the protection of multiple deletes should come from the DB state update from available to deleting
15:08:06 <vponomaryov> we should either revert-back reservation or commit it
15:08:40 <bswartz> why would we change the quota before updating the object state?
15:08:41 <zhongjun> cleanup error for deleting
15:09:36 <zhongjun> bswartz:  Because one share has many share instances
15:10:03 <vponomaryov> anyway, but exists and should be fixed, what exactly should we decide in this meeting?
15:10:24 <vponomaryov> rc1 or next release ?
15:10:33 <zhongjun> vponomaryov: We have two ways to fix it
15:10:35 <vponomaryov> s/but/bug/
15:10:39 <bswartz> vponomaryov: given that we're at the RC target date we have to decide whether this bug stays in Pike or gets punted to Queens
15:11:18 <bswartz> if we keep these bugs targeted at pike we need to fix them before we can release RC1
15:11:36 <bswartz> and I'm not sure there's a low risk fix here for both issues
15:12:06 <bswartz> given that quotas are a longstanding problem area, pike would be no worse than ocata if we deferred this bug
15:12:40 <bswartz> but I hate just kicking the can down the road and zhongjun has done some work here to try to fix the issue
15:13:13 <bswartz> I guess the question is, can we merge the current fix as is, or is more work needed?
15:13:46 <vponomaryov> bswartz, need to review it more attentoively first
15:13:52 <bswartz> we don't have to decide in this meeting -- we can handle the issue through code review
15:13:58 <zhongjun> bswartz:  need more review for it
15:14:01 <bswartz> but here we have a chance to discuss the patch
15:14:31 <gouthamr> bswartz zhongjun: can this be backported to stable/pike? it doesn't seem like a ship stopper to me..
15:14:38 <zhongjun> bswartz: I tested it before by myself, but only one guy take a look at it
15:14:45 <bswartz> however if the code reviews don't turn out positive this afternoon we'll have no choice but to punt it
15:15:14 <bswartz> anything else about these bugs before we move on?
15:15:28 <zhongjun> gouthamr: I could be back ported to stable/pike if we don't do many change in code
15:15:29 <vponomaryov> yeah, how many hours do we have before afternoon?
15:15:32 <bswartz> gouthamr: that's an option but it has downsides
15:15:57 <bswartz> vponomaryov:  the moment we have resolution of the 3 remaining bugs I will push a tag
15:16:12 <bswartz> the gate infra has been extra cranky this week so I'd rather not wait
15:16:30 <bswartz> but we have at least 3 hours to review it
15:17:02 <bswartz> I have it downloaded in my dev environment and I'm testing it
15:17:12 <bswartz> but I don't have a multibackend setup
15:17:50 <bswartz> alright next bug
15:17:53 <zhongjun> bswartz: It just need multiple manila-api services
15:18:15 <bswartz> zhongjun: is it possible to do that w/ devstack?
15:18:42 <bswartz> I run with m-api under apache
15:18:52 <vponomaryov> bswartz: requires either manual change of screen configs or manila-devstack plugin
15:19:13 <bswartz> I could probably hack something up
15:19:15 <zhongjun> bswartz: I don't think so
15:19:16 <bswartz> but not in 3 hours
15:19:22 <vponomaryov> bswartz: on single node you will need to change ports for each subsequent instance of API service
15:19:26 <bswartz> okay moving on
15:19:46 <bswartz> #link https://bugs.launchpad.net/manila/+bug/1659023
15:19:47 <openstack> Launchpad bug 1659023 in Manila "Consistent Snapshots are broken in the NetApp cDOT driver" [Medium,In progress] - Assigned to Ben Swartzlander (bswartz)
15:19:55 <bswartz> #link https://review.openstack.org/491877
15:20:01 <gouthamr> ^ looking at it
15:20:30 <bswartz> been wanting to fix this since late ocata -- when we killed the CGs code
15:20:55 <bswartz> thanks to vponomaryov for getting the share-groups stuff wrapped up in pike
15:21:19 <bswartz> this just needs reviews today
15:21:33 <bswartz> if anyone finds problems I'm happy to push more patchsets today
15:21:47 <bswartz> but if we find a serious issue we might need to punt this one too
15:22:04 * bswartz hopes we don't find a serious issue
15:22:59 <vponomaryov> bswartz: better to say "hope we don't HAVE serious issue" ))
15:23:07 <bswartz> true
15:23:17 <gouthamr> i can review this.. the chnge looks sane in theory.. will try running the tests that exist for consistent_snapshot_support
15:23:34 <gouthamr> if you already haven't, bswartz
15:24:27 <bswartz> gouthamr: I did manual testing -- I wasn't able to get the whitebox tests to run thanks to netapp's gerrit rejecting my SSH key
15:24:39 <gouthamr> oh, that's a solvable problem
15:24:40 <gouthamr> :)
15:25:17 <bswartz> okay that's it for my agenda
15:25:20 <bswartz> #topic open discussion
15:25:25 <bswartz> anything else for today?
15:25:31 <zhongjun> link:https://review.openstack.org/#/c/492358/
15:26:03 <zhongjun> We missing a little parameter in manilaclient
15:26:08 <bswartz> doh
15:26:20 <bswartz> it will be easy to merge that fix in queens
15:26:33 <bswartz> how bad is it to not have it in pike?
15:26:52 <vponomaryov> not bad at all
15:26:54 <gouthamr> just bad CLI user experience, i think..
15:26:57 <bswartz> oh wait
15:27:00 <bswartz> I think I ran into this bug
15:27:07 <gouthamr> create a share-group-type and then add extra-specs later
15:27:07 <bswartz> and I assumed it was something I was doing wrong
15:27:11 <zhongjun> We can not create a share group with group specs
15:27:19 <bswartz> yeah that's exactly what I do gouthamr
15:27:19 <zhongjun> in cli
15:27:32 <bswartz> okay this can easily wait until queens
15:27:38 <bswartz> thanks zhongjun
15:27:46 <gouthamr> we can merge it though, can't we?
15:27:55 <gouthamr> just not tag a release right away?
15:27:58 <bswartz> IMO, yes
15:28:01 <zhongjun> gouthamr: yes, we still have another way to do it
15:28:08 <bswartz> stable/pike branched a while back for the client
15:28:30 <gouthamr> ah yes..
15:28:34 <bswartz> 10 days ago
15:28:47 <bswartz> so it's frozen now
15:28:57 <bswartz> but master should be open for queens-related fixes
15:29:21 <bswartz> okay just a reminder -- we have 3 weeks until the Pike release
15:29:29 <bswartz> there's still a lot of docs issues that need addressing
15:29:46 <bswartz> please use the next weeks to update any docs related to changes in pike
15:29:56 <bswartz> and we have the docs migration to deal with too
15:30:35 <bswartz> Manila PTG will be the week after the Denver event
15:30:54 <bswartz> which is 4.5 weeks away I think
15:31:27 <bswartz> alright we're done for today
15:31:31 <bswartz> thanks for reviewing these last 2 changes
15:31:36 <bswartz> #endmeeting