22:04:38 <devananda> #startmeeting db
22:04:39 <openstack> Meeting started Thu Jan 24 22:04:38 2013 UTC.  The chair is devananda. Information about MeetBot at http://wiki.debian.org/MeetBot.
22:04:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
22:04:42 <openstack> The meeting name has been set to 'db'
22:05:15 <devananda> I don't have an agenda or anything :)
22:05:44 <devananda> but some really cool patches have landed recently, so I thought might be good to talk about the next steps
22:05:52 <devananda> or see what everyone else is doing
22:06:00 <devananda> #topic open discussion
22:07:14 <devananda> so, support for unique indexes landed yesterday
22:07:38 <devananda> #link https://review.openstack.org/16940
22:07:48 <devananda> and it looks like shadow tables // archiving is about to land
22:08:01 <devananda> #link https://review.openstack.org/18493
22:08:06 <jog0> devananda: awesome
22:08:14 <jog0> so now we can start using unique indexes?
22:08:17 <devananda> yep
22:08:21 <jog0> rather start enabling them
22:08:30 <devananda> Boris is planning to create one review / migration per table
22:08:40 <jog0> sweet
22:08:41 <devananda> to add actual unique constraints, and test the migration fwd/bkwd
22:09:25 <jog0> devananda: as there are some big changes to DB landing for Grizzly, perhaps it would be nice to send an email out to openstack-dev saying what has changed, and what is coming
22:09:42 <jog0> to let people know that the DB is getting better
22:09:52 <jog0> and how many cool things have landed
22:09:57 <devananda> probably a good idea :)
22:10:18 <devananda> there's also the no-db-compute stuff which seems to be slowing down a lot
22:10:38 <devananda> though i haven't followed it closely
22:10:54 <jog0> not sure how far they are on that one, but looks likes its made a lot of progress
22:11:02 <jog0> russellb: ^
22:11:42 <jog0> so lets review the big cahnges that have landed so far
22:11:57 <jog0> and what is still scheduled
22:12:12 <jog0> so we have the basics for a email
22:12:18 <devananda> k
22:12:36 <lifeless> russellb: does reviewday do a topo sort (e.g. all reviews in a stack get promoted to the highest importance of the reviews under it?
22:13:00 <russellb> lifeless: no idea on that
22:13:03 <russellb> dprince wrote it
22:13:07 <zykes-> question: plans to support stuff else then mysql ?
22:13:09 <russellb> on no-db-compute, still some work to do
22:13:24 <russellb> I have a review up that hacks nova-compute to log a traceback every time db.anything gets called
22:13:24 <devananda> zykes-: afaik postgres is already supported
22:13:38 <zykes-> oh
22:13:45 <russellb> still looks like we can do it by grizzly-3, but have to work at it
22:14:03 <krtaylor> I added postgres CI tests
22:14:20 <devananda> russellb: would be great to have a status page that listed the db calls still coming from n-cpu :)
22:14:22 <russellb> postgres is used on smokestack as well
22:14:39 <russellb> devananda: heh, yeah ... it's basically the nova-compute log from the latest test run on that review right now
22:14:57 <devananda> heh
22:15:25 <russellb> for reference ... https://review.openstack.org/#/c/20052/ ... http://logs.openstack.org/20052/7/check/gate-tempest-devstack-vm/27998/logs/screen-n-cpu.txt.gz
22:16:06 <devananda> jog0: the meeting wiki seems to be the best reference right now for ongoing DB work at a high level: http://wiki.openstack.org/Meetings/DBTeamMeeting
22:16:09 <russellb> 105 entries in that log file (many of them the same thing)
22:17:07 <jog0> devananda: yeah,  most of those are in progress AFAIK
22:17:11 <jog0> if not all
22:17:23 <devananda> yea
22:17:55 <jog0> although I did file a bug on non-blocking-db
22:18:01 <jog0> with comstud
22:19:03 <devananda> going to jot down my understanding of the different project status' here: https://etherpad.openstack.org/db-status
22:19:21 <devananda> and use that as a basis for an email
22:19:23 <jog0> sounds good
22:19:50 <sdague> russellb: so postgresql in tempest gate is currently bust, you know if dprince is running all the tests or just smoke?
22:20:16 <russellb> sdague: i'm not actually sure ...
22:21:05 <sdague> http://logs.openstack.org/periodic/periodic-tempest-devstack-postgres-vm-full/ if anyone wants to help sort out the fails, wanted to get that actually gating by g-3 to keep us sane
22:21:33 <dripton> sdague: I'll take a look at postgres failures.
22:21:44 <russellb> sdague: if you go to smokestack.openstack.org, scroll down to a nova patch, you should be able to dig into the logs for a given run that was using postgres
22:21:51 <devananda> dripton: hey there :)
22:22:13 <sdague> dripton: I think I got the whitebox one addressed in review, but there is a floating ips one, they are usually type mismatch issues
22:22:15 <russellb> he may have tempest disabled, not sure
22:22:24 <sdague> russellb: ok
22:22:43 <sdague> sorry, just figured I'd stick that out there as I saw the pg statement
22:22:48 <russellb> sdague: i'm not actually seeing tempest in the log file right now
22:23:14 <devananda> dripton: I +1'd your shadow-table patch. is there a patch for copying deleted rows to them?
22:23:34 <dripton> devananda: not yet; I was waiting for the first part to go in first.  But I think it'll be easy.
22:24:33 <devananda> dripton: great. if it's easy, i think it would be good to have the copy-deleted-rows patch up to get feedback on it
22:24:50 <devananda> the particulars of that were a bit of a hot topic at the last summit, IIRC
22:25:20 <dripton> Will do. I wanted to make sure the shadow tables were okay with people first.  If they are then it's straightforward copying.
22:25:27 <jog0> devananda: dripton +1.  It would be nice to  be able to try out the shadow tables with copying deleted rows before merging the migration
22:25:41 <devananda> ++
22:26:03 <dripton> ok, I'll try to get the copying patch up soon.   (If it's as easy as I think, maybe tomorrow.)
22:26:40 <jog0> devananda: lets add impact to the ehterpad
22:32:26 <jog0> devananda: looks good
22:32:38 <devananda> awesome, thx :)
22:32:42 <jog0> there are definitely some nice changes in there
22:32:53 <jog0> lots of exciting stuff for Grizzly
22:34:25 <dripton> Does that etherpad auto-save?  The one I use most has a save button.
22:34:45 <jog0> AFAIK yes
22:34:55 <devananda> dripton: yes. click the history button to see
22:35:42 <devananda> any other topics folks want to bring up?
22:35:59 <jog0> devananda: yeah
22:36:13 <jog0> haven't finished collecting/parsing the logs yet
22:36:39 <jog0> but think I found an example where mysql reports it takes 0.003156 second and nova.db.api  takes 0.08 seconds
22:37:11 <jog0> but I will collect those logs and pastebin them when I double check
22:37:59 <devananda> sounds good. an occasional skew of 0.05s doesn't surprise me too much in a cloud env
22:38:03 <devananda> unless it's consistent
22:38:39 <jog0> devananda: well I have very verbose logs this time so maybe we can make something of it
22:38:45 <devananda> :)
22:39:36 <jog0> devananda: this is related to finding out why some REST API commands take over a second
22:40:17 <devananda> right
22:41:42 <devananda> if there's nothing else, shall we end meeting a bit early?
22:41:47 <dripton> +1
22:41:56 <jog0> +1
22:42:04 <devananda> great. thanks all!
22:42:07 <dripton> bye
22:42:08 <devananda> #endmeeting