21:01:09 <russellb> #startmeeting nova
21:01:09 <openstack> Meeting started Thu Jan 30 21:01:09 2014 UTC and is due to finish in 60 minutes.  The chair is russellb. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:01:10 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:01:13 <openstack> The meeting name has been set to 'nova'
21:01:16 <russellb> hello, everyone!
21:01:24 <melwitt> hi
21:01:26 <n0ano> o/
21:01:27 <dripton> hi
21:01:31 <alaski> hi
21:01:42 <jog0> o/
21:01:47 <russellb> #topic general
21:01:47 <jaybuff> o/
21:01:51 <cyeoh> hi (will be away intermittently and have to leave early)
21:01:56 <sdague> o/
21:01:57 <russellb> icehouse schedule:
21:01:59 <russellb> #link https://wiki.openstack.org/wiki/Icehouse_Release_Schedule
21:02:03 <russellb> icehouse winding down on us
21:02:13 <russellb> icehouse-3 deadlines for nova:
21:02:15 <russellb> #link http://lists.openstack.org/pipermail/openstack-dev/2014-January/025675.html
21:02:23 <russellb> less than 1 week before blueprints need to be approved
21:02:39 <dansmith> o/
21:02:50 <beagles> hi
21:02:58 <russellb> less than 3 weeks until code for blueprints must be up for review
21:03:03 <sandywalsh> o/
21:03:04 <russellb> any schedule questions?
21:03:28 <russellb> #topic sub-teams
21:03:34 <russellb> let's dive into some sub-group efforts
21:03:44 * n0ano gantt
21:03:47 <russellb> n0ano: go for it
21:03:54 <n0ano> few things
21:04:36 <n0ano> no db scheduler - boris is making progress but the performance studies hit bottleneck unrelated to the scheduler, his team will have to solve them before then can really work on the scheduler part
21:04:54 <jog0> n0ano: what type of bottlenecks?
21:05:16 <jog0> and do we have bugs filed for 'em
21:05:21 <n0ano> he didn't provide details, they are still analyzing, I don't think they know where it is yet
21:05:36 <russellb> OK, so we don't expect that for icehouse-3 then?
21:05:39 <russellb> should we defer it?
21:05:55 <n0ano> seems unlikely but I'd let boris make that kind of call
21:06:10 <russellb> boris-42: ^^^
21:06:14 <russellb> ok
21:06:33 <russellb> saw some gantt updates this week
21:06:33 <jog0> +1 for defer, it would be great to have a few months running something like that before making a stable release
21:06:57 <n0ano> code forklift to gantt - passing the unit tests, getting the patch reviewed, next step is to get it to pass the tempest integration tests
21:07:00 <russellb> jog0: yeah, end of release cycle isn't the best time to land critically invasive bits
21:07:06 * melwitt novaclient (short report)
21:07:21 <lifeless> o/
21:08:01 <russellb> n0ano: OK, get your devstack patch in?
21:08:26 <n0ano> russellb, I just updated it this morning, I'm hopefull that's it's mergeable now
21:09:02 <russellb> n0ano: OK, well keep up the work on it ... hopefully we can demonstrate it running by juno
21:09:02 <n0ano> review of https://review.openstack.org/#/c/67666/ is always good
21:09:29 <n0ano> I think that's a doable goal (with a lot of work still needed)
21:09:31 <russellb> sdague: devstack for gantt ^
21:09:41 <sdague> russellb: ok, will look
21:09:48 <russellb> k thanks!
21:09:55 <russellb> n0ano: any other topics?
21:10:06 <n0ano> of course, big meeting this week :-)
21:10:08 <russellb> or anything you need?
21:10:32 <n0ano> some discussion on policy based scheduler, BP at https://review.openstack.org/#/c/61386/
21:10:50 <n0ano> looks interesting but still needs work.
21:11:12 * mriedem joins late
21:11:13 <russellb> and needs to be split up
21:11:25 <n0ano> finally, instance groups changes are still under review, they are working on revising to make the patches palatable
21:11:36 <russellb> n0ano: i've been wondering about those ... would love to see that make it
21:11:48 <russellb> at the last summit, the objection was that the design got way out of control (too complicated)
21:11:53 <jog0> n0ano: is that for v2 and v3 or just v3?
21:11:59 <russellb> and we wanted to see it return to much more simple approach that it had before
21:12:12 <n0ano> jog0, they're still trying for both v2 & v3
21:12:26 <n0ano> they want to split the code into 3 parts, v2, v3 and doc
21:12:34 <jog0> thoughts on just doing v3?
21:12:40 <n0ano> that will help the review process but it's still complicated
21:12:59 <n0ano> jog0, didn't go into that so don't know.
21:13:04 <jog0> adding more v2 API as we try to get v3 out the door seems like a distraction
21:13:11 <jog0> not against it per se though
21:13:31 <russellb> not huge amount of extra code
21:13:42 <russellb> definitely v3 though, v2 a bonus
21:14:06 <n0ano> I'd push for do v3 first and then add v2 later but that's debo's call
21:14:41 <russellb> OK, hope we can get that in then
21:15:01 <russellb> melwitt: you're up
21:15:19 <melwitt> the biweekly novaclient report 2014/1/30:
21:15:19 <melwitt> open bugs, 141 !fix released, 84 !fix released and !fix committed
21:15:19 <melwitt> 26 new bugs, 0 high bugs
21:15:19 <melwitt> 12 patches up, 3 are WIP, reviews are all active
21:15:36 <russellb> a few new ones, but not bad
21:16:00 <russellb> anything you think needs special attention?
21:16:14 <russellb> probably about time we had a novaclient release ... haven't done one in a while
21:16:21 <russellb> wonder when the time is right for that
21:16:24 <cyeoh> russellb: I was just about to ask for that
21:16:25 <russellb> now?  after icehouse-3?
21:16:35 <russellb> it's cheap and easy to release really
21:16:36 <cyeoh> we can now boot and do pretty much everything against the V3 API
21:16:44 <mriedem> garyk has vm diagnostics stuff going into the client
21:16:48 <cyeoh> so a new release would allow us to start doing scenario testing
21:16:52 <russellb> OK
21:16:53 <mriedem> so i'd vote after his patches land in nova/client for that blueprint
21:16:58 <melwitt> there's this review seems straightforward, adding the client side support for rdp console https://review.openstack.org/#/c/44250/
21:17:12 <russellb> i told them i'd review the nova side for the rdp part
21:17:28 <russellb> so it sounds like we'll need one just after feature freeze
21:17:37 <russellb> but i could do one now too if that's valuable
21:17:46 <russellb> and then another for whatever lands in the next month as we go to freeze
21:17:55 <cyeoh> one now would be useful so we can start working on the tempest v3 api scenario tests
21:18:15 <russellb> cyeoh: OK, i'll put that on my todo (may be tomorrow)
21:18:21 <cyeoh> russellb: thanks!
21:18:28 <russellb> np
21:18:39 <russellb> anything else on novaclient?  or other sub-team stuff?
21:19:04 <melwitt> no, that's it. I didn't notice the nova side of the rdp hadn't been merged yet, my bad
21:19:17 <mriedem> i had a client related question
21:19:22 <russellb> melwitt: np
21:19:26 <russellb> mriedem: go for it
21:19:30 <mriedem> this bug https://bugs.launchpad.net/nova/+bug/1256119
21:19:37 <mriedem> adds flavor extra spec validation
21:19:41 <mriedem> in horizon, client and nova api
21:19:48 <mriedem> is it normal to do that triple maintenance?
21:19:53 <mriedem> i thought nova api validation would be enough
21:19:58 <mriedem> let the clients handle the 400
21:20:12 <russellb> well, definitely in the API
21:20:18 <russellb> the client?  meh, probably not IMO
21:20:25 <mriedem> ok, that's what i said in the reviews
21:20:30 <mriedem> but wasn't sure
21:20:31 <russellb> horizon, they may have good reason
21:20:46 <mriedem> yeah, front-side validation for perf or something
21:21:00 <mriedem> ok, that's all
21:21:11 <russellb> performance, or just a better UI with better feedback or something
21:21:12 <cyeoh> its pretty common in novaclient to do a bit of duplicate validation.
21:21:39 <mriedem> cyeoh: ok, that's what i wasn't sure of, i didn't like duplicating the regex all over the place
21:21:39 <russellb> same could be said in the CLI
21:22:16 <cyeoh> mriedem: yea its bad when it gets out of sync. I think people probably did it because they want very specific error messages (for the CLI)
21:22:35 <mriedem> right
21:22:46 <mriedem> ok, i got enough, don't mean to hold the meeting up
21:22:50 <russellb> all good :)
21:22:52 <russellb> #topic bugs
21:23:17 <russellb> we've actually made some good progress in the last week on nova gate bugs
21:23:31 <russellb> still plenty to chase though overall ... http://status.openstack.org/elastic-recheck/
21:23:50 <dansmith> progress why? because I was out and didn't make any new gate bugs?
21:23:54 <russellb> this page was added recently, very helpful to see the stuff we haven't categorized yet: http://status.openstack.org/elastic-recheck/data/uncategorized.html
21:23:59 <russellb> dansmith: mainly, yes
21:24:06 <dansmith> russellb: tell perry I need more vacation then
21:24:40 <russellb> last week we talked about a bug day
21:24:43 <mriedem> the untagged bug list has been picked over a bit in the last week too
21:24:50 <russellb> john g isn't around, he was going to organize
21:24:53 <russellb> i'll have to ping him tomorrow
21:24:56 <lifeless> sorry I've been failing at driving a bug team
21:25:09 <russellb> lifeless: it's OK, you have your hands full
21:25:17 <lifeless> russellb: I rather suspect I do
21:25:21 <russellb> bug team management position open :)
21:25:24 <mriedem> we have over 90 untagged bugs on sunday, down to 48 now
21:25:29 <russellb> lifeless: tripleo is a huge mission :)
21:25:38 <mriedem> a lot of those are check failures that are dupes or no longer showing up in logstash
21:25:40 <russellb> mriedem: thanks for the help :)
21:25:54 <mriedem> np, point is it's not as bad as it probably looks
21:26:02 <mriedem> when you sift through the clutter
21:26:08 * dansmith nominates mriedem as bug lackey^Wwrangler
21:26:15 <mriedem> fudge
21:26:17 <cyeoh> +1
21:26:20 <russellb> mriedem: you'd be great!
21:26:35 <mriedem> i consider it a hobby o-)
21:26:40 <russellb> organize a bug day now and then ... get a few people to meet weekly to do triage together
21:26:41 <mriedem> while tox runs
21:26:43 <russellb> misery loves company?
21:26:59 <mriedem> will think about it
21:27:01 <russellb> :)
21:27:09 <russellb> alright, moving on for now
21:27:11 <russellb> #topic blueprints
21:27:22 <russellb> #link https://launchpad.net/nova/+milestone/icehouse-3
21:27:27 <russellb> blueprints must be approved by tuesday
21:27:39 <russellb> we have to start trimming this down
21:27:57 <russellb> if there are particular ones you feel are especially important, we should talk about the priority
21:28:03 <russellb> i've given up on the nova-core sponsor thing for icehouse
21:28:12 <russellb> and would like to just hand pick a set that we feel are the most important to review and merge
21:28:58 <russellb> if you have stuff on this list, i'd appreciate taking a look and making sure the status is accurate
21:29:06 <russellb> and if you know it won't make it, go ahead and defer to "next"
21:29:45 <russellb> any blueprints folks want to talk about specifically?
21:29:53 <alaski> I'd like to propose https://blueprints.launchpad.net/nova/+spec/instance-tasks-api as important in order to get into v3, despite my neglect up to this point
21:30:11 <jog0> russellb: I would like to see https://blueprints.launchpad.net/nova/+spec/support-libvirt-1x make it in
21:30:28 <cyeoh> alaski: yea I'd really like to see that get in
21:30:29 <russellb> alaski: still drafting though ... needs design ACK before approved :-/
21:30:41 <russellb> but yeah, that probably has to make it, or v3 doesn't make it (per list discussion)
21:31:03 <russellb> jog0: sounds good, feel free to bump to medium
21:31:17 <lifeless> I filed a new one last week
21:31:20 * lifeless looks it up
21:31:22 <alaski> russellb: I'll have a design and code by tomorrow, which can at least get it into discussion
21:31:25 <mriedem> jog0: we need to link that bp to the bug blocking newer libvirt
21:31:27 * mriedem looks
21:31:29 <russellb> alaski: great!
21:31:37 <jog0> mriedem: go for it
21:31:42 <lifeless> russellb: https://blueprints.launchpad.net/nova/+spec/disable-file-injection-by-default
21:31:45 <russellb> alaski: and let's plan to dive into it more deeply in person, maybe we can push progress quickly that way
21:31:58 <russellb> lifeless: didn't target to icehouse-3
21:32:03 <cyeoh> russellb: I'd like to get the v3 api related blueprints to get bumped up a bit. to make sure we have a consistent interface
21:32:18 <russellb> lifeless: needs target and assignee, and that puts it in the review queue
21:32:21 <lifeless> russellb: oh, sorry. The targeting stuff always gets me
21:32:39 <russellb> np, there's just way too much noise in there, so i'm only looking at the targeted stuff
21:32:42 <lifeless> russellb: done
21:32:54 <alaski> russellb: sounds good
21:33:15 <russellb> cyeoh: i think that's fine.  it needs to be prioritized to have a chance to make icehouse
21:33:26 <russellb> cyeoh: send me a list of blueprints that need to be bumped
21:33:49 <cyeoh> russellb: thx, will do
21:34:18 <russellb> speaking of v3, XML is almost ripped out
21:34:32 <russellb> a good bit of that merged in the last 24 hours
21:35:04 <cyeoh> excellent :-) That makes new stuff going in significantly easier to review too
21:35:30 <sdague> yay for XML v3 being gone :)
21:35:47 <russellb> sdague: thanks for pushing that
21:35:52 <sdague> np
21:36:34 <russellb> #topic open discussion
21:36:37 <russellb> anything else for today?
21:37:34 <dansmith> *crickets*
21:37:37 <cyeoh> so just one thing on the V3 API
21:37:38 <jog0> russellb: we need to be careful watching for new gate bugs
21:37:49 <jog0> so we can keep the gate queue moving
21:38:02 <russellb> jog0: for sure ... could use some sort of notification when stuff crops up
21:38:05 <russellb> hard to just always study the data
21:38:16 <russellb> i'm hoping my focus can start shifting back to nova now
21:38:21 <jog0> russellb: tracking uncalssified failures
21:38:27 <devananda> I just bumped deprecate-baremetal-driver to "Beta available" -- there's a review up and Ironic folks have it working. More polish on the way soon.
21:38:35 <russellb> devananda: cool
21:38:39 <jog0> so just watch http://status.openstack.org/elastic-recheck/
21:38:50 <lifeless> devananda: \o/
21:39:04 <mriedem> jog0: if i have to be bug boy you have to be gate man
21:39:12 <melwitt> another V3 API thing, I was wondering if I can put up a patch to add back the security groups extension functionality that was removed assuming nova-network deprecation
21:39:16 <russellb> mriedem: totally
21:39:25 <jog0> haha
21:39:27 <lifeless> melwitt: :(
21:39:39 <russellb> jog0 is the nova gate czar
21:39:44 <lifeless> maybe someone should write a neutron backend that calls nova-network RPCs
21:39:52 <lifeless> then we don't need the public API to be messed up
21:40:00 <russellb> lifeless: ha
21:40:05 <russellb> hadn't considered that ...
21:40:13 <russellb> somewhat evil
21:40:26 <cyeoh> melwitt: if nova-network really is going to stick around for a long time then there is a lot of nova-network stuff we have to put back in
21:40:46 <russellb> sadly yes
21:40:55 <russellb> at this point, i have no confidence in any timeline for nova-network going away
21:41:07 <melwitt> right.. I had removed it is why I ask, forgot to mention that heh
21:41:16 <cyeoh> melwitt: but i'm also a bit wary of the benefit of releasing with incomplete nova-network (random bits missing) rather than having none at al
21:41:20 <russellb> that was the right thing to do based on previous planning
21:41:26 <russellb> plan has changed now :-/
21:41:36 <sdague> cyeoh: what's the minimum yuo think we need?
21:41:52 <sdague> if we had a good minimum, we could do a hard press to get it in
21:42:15 <cyeoh> sdague: yea that's true. I guess the fixed/floating stuff has to go in
21:42:29 <cyeoh> stuff like cloudpipe is probably low priority
21:42:36 <russellb> not sure cloudpipe worked
21:42:40 <russellb> so meh
21:42:41 <cyeoh> heh
21:42:45 <sdague> yeh
21:42:47 <melwitt> :)
21:42:51 <cyeoh> security groups as melwitt  mentions
21:43:10 <cyeoh> and its much easier if we live with some parts of the API only working with nova-network (but obviously is ugly)
21:43:15 <russellb> network creation stuff
21:43:40 <cyeoh> yea the networks extension has to go back in
21:43:46 <cyeoh> and need to reverify the server boot path
21:44:10 <lifeless> russellb: how big is the neutron gap really? - there is HA for l3 already
21:44:19 <lifeless> russellb: I meant to reply to your email, just haven't done so yet
21:44:25 <cyeoh> also need novaclient changes (this part is where we'd really be pushing the deadlines I suspect)
21:44:40 <russellb> lifeless: feature gap is part of it, the quality/testing/reliability/scale gap is also part of it
21:44:55 <russellb> lifeless: and my impression from neutron folks is that significant pieces just need to be rewritten
21:44:55 <beagles> lifeless: quality/perf is a huge part
21:45:01 <lifeless> russellb: fwiw we have /no/ nova-network test scenarios in tripleo today
21:45:25 <melwitt> cyeoh: novaclient changes to put things back? or has some v3 api stuff not been added yet still?
21:45:28 <russellb> like, for example, with the open source plugins, you can't run more than one API server, or more than one API worker
21:45:42 <russellb> so you go beyond proof of concept, and your API worker will fall the heck over
21:45:47 <sdague> lifeless: as a data point, devananda switched ironic testing to nova-network yesterday so he could merge code again, as the neutron kernel bug tickle meant he couldn't merge code otherwise
21:45:49 <cyeoh> melwitt: I avoided porting V2 code to V3 for novaclient where it was nova-network specific
21:45:50 <russellb> it's just unusable
21:45:57 <melwitt> cyeoh: goti t
21:46:09 <melwitt> *got it
21:46:28 <lifeless> sdague: interesting, but ironic can't boot VM's with nova-network
21:46:30 <devananda> yep. we added tempest tests to ironic's gate pipe, and with neutron enabled (even though we aren't using it yet) we couldn't merge anything
21:46:37 <sdague> lifeless: yep
21:46:40 <devananda> lifeless: i know
21:46:45 <lifeless> sdague: so I'm not sure what it means :>
21:46:53 <devananda> lifeless: but tempest can't pass with neutron enabled right now
21:47:08 <lifeless> devananda: ack
21:47:09 <russellb> lifeless: yeah it's a tough situation ... we've all been planning based on neutron being ready for prime time long ago
21:47:10 <devananda> so we won't be able to test that functionality of ironic
21:47:11 <melwitt> cyeoh: I'd like to help on that side if you need it. I haven't gotten a chance to dig into novaclient much yet though I want to
21:47:26 <lifeless> devananda: btw
21:47:27 <devananda> *to have tempest test ...
21:47:35 <lifeless> devananda: the tripleo test cluster is usable by you now
21:47:41 <devananda> lifeless: sweet!
21:47:42 <lifeless> devananda: just define jobs
21:48:03 <cyeoh> melwitt: that'd be great - although we can't really merge until the nova patches merge, they can be done speculatively I think (the api should in most cases not change much from V2 - just general cleanups)
21:48:30 <lifeless> devananda: for instance - a 'up to seed usable' test http://logs.openstack.org/23/70023/1/check/gate-tripleo-deploy/d49f20d/console.html
21:49:01 <melwitt> cyeoh: true. I also want to help with the nova patches since I'm pretty familiar. so I'll keep an eye out for the etherpad tracking that work to see if I can pitch in
21:49:44 <devananda> lifeless: so we'll probably want to make seed able to use nova-ironic next
21:50:07 <devananda> lifeless: so taht we dont need to bring up undercloud every time we want to test ironic.
21:50:20 <cyeoh> melwitt: I'll try to get an etherpad list of stuff we need to do for nova-network out later today
21:50:22 <lifeless> devananda: thats a good idea; and add a parallel job for that
21:50:36 <melwitt> cyeoh: awesome, thanks.
21:50:37 <devananda> lifeless: we should move to #tripleo to continue this :)
21:50:42 <lifeless> devananda: triggered on anything tripleo + devstack-gte + *ironic* changes
21:50:46 <russellb> heh
21:50:55 <russellb> well thanks everyone for coming, i'll close the meeting and we can move back to project channels
21:50:59 <russellb> #endmeeting