17:01:03 <dansmith> #startmeeting nova_cells
17:01:03 <openstack> Meeting ended Wed Oct 11 17:01:00 2017 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)
17:01:04 <openstack> Minutes:        http://eavesdrop.openstack.org/meetings/interopwg/2017/interopwg.2017-10-11-16.00.html
17:01:05 <openstack> Minutes (text): http://eavesdrop.openstack.org/meetings/interopwg/2017/interopwg.2017-10-11-16.00.txt
17:01:06 <openstack> Log:            http://eavesdrop.openstack.org/meetings/interopwg/2017/interopwg.2017-10-11-16.00.log.html
17:01:07 <openstack> Meeting started Wed Oct 11 17:01:03 2017 UTC and is due to finish in 60 minutes.  The chair is dansmith. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:01:08 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:01:10 <openstack> The meeting name has been set to 'nova_cells'
17:01:25 <melwitt> o/
17:01:26 <tssurya_> 0/
17:01:31 <mriedem> o/
17:01:35 <dansmith> wow tssurya_ has a tall head
17:01:43 <melwitt> haha
17:01:46 <tssurya_> hehe :D
17:01:48 <dansmith> #topic bugs
17:02:07 <dansmith> mriedem: are you still debugging that one from a few minutes ago or are you good now?
17:02:23 <mriedem> i marked it incomplete
17:02:27 <mriedem> and put my notes in there
17:02:31 <dansmith> right on
17:02:42 <melwitt> I found a couple of bugs while working on improving our usage of CellDatabases
17:02:51 <melwitt> https://bugs.launchpad.net/nova/+bug/1721670
17:02:53 <openstack> Launchpad bug 1721670 in OpenStack Compute (nova) "Build notification in conductor fails to send due to InstanceNotFound" [Undecided,Fix released] - Assigned to melanie witt (melwitt)
17:02:57 <dansmith> yep, and patches are up for them all yes?
17:03:18 <melwitt> https://bugs.launchpad.net/nova/+bug/1722404
17:03:19 <openstack> Launchpad bug 1722404 in OpenStack Compute (nova) "Database transactions can fail with "TypeError: Can't upgrade a READER transaction to a WRITER mid-transaction" because of scatter_gather_cells" [Undecided,In progress] - Assigned to melanie witt (melwitt)
17:03:31 <melwitt> yep patches are up. first one is merged so I proposed a backport for pike
17:03:33 <dansmith> and backports?
17:03:36 <dansmith> cool
17:04:05 <dansmith> so there was another bug filed last week, I think after this meeting, about instance list
17:04:08 <melwitt> I requested nominations for pike series on both bugs
17:04:28 <dansmith> which has since been fixed
17:04:50 <dansmith> it was a fairly obscure detail I think, but otherwise haven't heard other complaints about our instance listing, so tentatively hopeful that that stuff is good
17:05:07 <dansmith> so anything else here? any open/unsolved bugs?
17:05:28 <mriedem> not from me
17:05:35 <dansmith> cool
17:05:40 <dansmith> #topic open reviews
17:05:48 <dansmith> tssurya_: has a patch for nova-manage
17:05:50 <tssurya_> I have my first ever open stack patch here for the enhancement Wishlist bug . Reviews are most welcome (and sorry for large number of duplicate patch sets, was having some trouble regarding patch merge conflicts) : https://review.openstack.org/#/c/510844/
17:05:55 <dansmith> I saw some traffic on it, but haven't circled back to look yet
17:06:14 <dansmith> tssurya_: cool
17:06:49 <dansmith> I have some remaining instance list patches up for review: https://review.openstack.org/#/c/508595
17:07:07 <dansmith> they squash buildrequests into the instance list stuff so we stop having stupid sorting with those
17:07:17 <dansmith> and also a patch that makes us only query cells we have instances in,
17:07:40 <dansmith> which will help performance for people that segregate users by class into cells for organization or locality
17:07:51 <dansmith> the buildrequest one is definitely something I'd like to land soon
17:08:00 <dansmith> the smarter cells patch is kinda optional, but I think it's worth doing
17:08:16 <mriedem> segregate users by class into cells?
17:08:26 <mriedem> so, tenant isolation host aggregates?
17:08:36 <dansmith> if you take the edge case,
17:08:52 <dansmith> you might have users in a remote office be booting things in the cell that is in their office
17:09:02 <dansmith> or if people use cells for segregating new/old hardware,
17:09:15 <dansmith> you might have performance users only booting new instances in the new hardware cell
17:09:23 <dansmith> but yeah, basically,
17:09:39 <dansmith> if your aggregate mapping makes you only boot things in one cell, no reason to make instance list always query other cells
17:09:39 <mriedem> i was just trying to figure out how you make the connection,
17:09:46 <mriedem> since it'd have to be via the host mappings implicitly
17:10:01 <dansmith> no, organization through one of the available methods, like aggregates,
17:10:09 <dansmith> but just making the cells lister thing not list unnecessary cells
17:10:19 <dansmith> based on what cells you have instancemappings for
17:10:38 <mriedem> that's this one right https://review.openstack.org/#/c/509002/ ?
17:10:38 <dansmith> if we fix the instance archive thing deleting instance mappings,
17:10:49 <dansmith> you'd also not list against cell0 if you have no error instances right now
17:10:57 <dansmith> mriedem: https://review.openstack.org/#/c/508595/7
17:11:00 <dansmith> er, no
17:11:06 <dansmith> the one at the end of that one
17:11:24 <dansmith> mriedem: 002 is just a query method to make the next patch quick
17:11:43 <mriedem> yeah, have you run this through the danperftestmachine9000?
17:12:02 <mriedem> to see that the up front grouping doesn't affect a normal list too much?
17:12:05 <dansmith> no, because it will only affect things with more cells
17:12:28 <dansmith> I can't tell a difference between 500/500 cell0/cell1 and 1000 in cell1 reliably, so I'm not going to be able to see this either I think
17:12:57 <dansmith> I can do it if you want, but it's not high on the prio list I don't think
17:14:28 <dansmith> anyway, other open reviews.. melwitt has those fixes and corresponding fixture tweaks
17:14:31 <dansmith> what else?
17:14:49 <melwitt> yeah, the fixture patch is https://review.openstack.org/#/c/508432 and I would appreciate review on it
17:16:01 <dansmith> okay nothing else I guess, so moving on
17:16:04 <dansmith> #topic open discussion
17:16:07 <mriedem> crap mel's notification fix reminds me i was going to collapse the 2 for loops in schedule_and_build_instances
17:16:24 <melwitt> ohh, right
17:16:30 <dansmith> anything else here?
17:16:54 <mriedem> i have another patch to do the instance.tag stuff at server create better too,
17:16:58 <mriedem> but it has to be cleaned up
17:17:05 <dansmith> okay
17:17:07 <mriedem> not directly cells related, but it came up when adding that stuff in schedule_and_build_instances
17:17:16 <dansmith> yeah
17:18:10 <dansmith> okay if nothing else, I propose we get back to work
17:18:34 <mriedem> yup
17:18:39 <melwitt> yis
17:18:42 <tssurya_> yes
17:18:46 <dansmith> #endmeeting