17:00:01 <dtantsur> #startmeeting ironic
17:00:01 <openstack> Meeting started Mon Aug 21 17:00:01 2017 UTC and is due to finish in 60 minutes.  The chair is dtantsur. Information about MeetBot at http://wiki.debian.org/MeetBot.
17:00:02 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
17:00:04 <openstack> The meeting name has been set to 'ironic'
17:00:10 <nicodemos> o/
17:00:13 <vdrok> o/
17:00:14 <etingof> o/
17:00:14 <rloo> o/
17:00:24 <kaifeng> o/
17:00:36 <dtantsur> hi all! thanks for coming and welcome to the ironic meeting :)
17:00:38 <rpioso> o/
17:00:42 <crushil> o/
17:00:43 <stendulker> o/
17:01:01 <dtantsur> our agenda predictably resides on the same page:
17:01:03 <dtantsur> #link https://wiki.openstack.org/wiki/Meetings/Ironic
17:01:37 <rama_y> o/
17:01:40 <dtantsur> #topic Announcements / Reminder
17:02:01 <dtantsur> #info ironic 9.0.1 released with 9.1.0 to follow soon
17:02:27 <dtantsur> for those who was not following our release struggles: we have to quickly release 9.0.0, then 9.0.1 just to unbreak all the CI jobs
17:02:38 <dtantsur> may be worth a separate info item:
17:02:51 <dtantsur> #info ironic 9.1.0 is expected to be our final release for pike (not 9.0.0)
17:03:02 <pas-ha> o/
17:03:03 <Nisha_Agarwal> o/
17:03:20 <rloo> a big shout-out/thank you to TheJulia, sambetts and vdrok for helping out last week with the release
17:03:31 <dtantsur> yes, thanks a lot (and you rloo too)!
17:03:39 <fellypefca> o/
17:03:40 <vdrok> yeah rloo too :)
17:03:51 <rloo> :) that's what happens when the ptl goes on vacation, we don't play!
17:04:06 <Nisha_Agarwal> yes Thanks to all :)
17:04:07 <sauloaislan> o/
17:04:19 <dtantsur> then you'll be glad to know that I'm not going on vacation any soon (if at all this year) :D
17:04:22 <rloo> which reminds me. Congrats to dtantsur for being PTL again.
17:04:35 <dtantsur> thanks all for putting your trust in me :)
17:04:48 <Nisha_Agarwal> Congrats dtantsur
17:04:51 <Nisha_Agarwal> :)
17:05:03 <stendulker> Congrats dtantsur :)
17:05:05 <etingof> dtantsur++
17:05:06 <vdrok> \o/
17:05:22 <dtantsur> :)
17:05:25 <pas-ha> \o/
17:05:43 <rpioso> Thank you, all, for getting the Dell iDRAC h/w type merged :)
17:05:50 <dtantsur> you're welcome
17:06:00 <dtantsur> now, I think, we have hardware types for all supported classic drivers
17:06:12 <rpioso> dtantsur, TheJulia: Thank you for running.
17:06:16 <rloo> yup and yay!
17:06:21 <dtantsur> yeah, thanks TheJulia (if you're here now)
17:06:27 <rloo> that was for hw types.
17:06:29 <xavierr> o/
17:06:36 <rloo> yay for TheJulia for running!
17:06:51 <dtantsur> #info final releases for all other ironic projects are done
17:06:58 * rloo hopes TheJulia doesn't get blinded by the eclipse...
17:07:09 <dtantsur> rloo++
17:07:25 <dtantsur> now, please watch out for the incoming bugs
17:07:36 <dtantsur> this is the time when people start putting Pike into their testing
17:07:49 <dtantsur> #info PTG planning: https://etherpad.openstack.org/p/ironic-queens-ptg
17:08:12 <dtantsur> this is not the first time I post this link, but PTG is coming :) so please add your ideas, and review the others'
17:08:38 <rloo> dtantsur: when are we going to discuss/plan the PTG? next meeting or one after that?
17:08:41 <dtantsur> mmm, and the Sydney schedule was published, but I don't have the link handy
17:08:48 <dtantsur> rloo: lemme check the calendar
17:09:06 <dtantsur> oh, yeah, next two meetings we should spend some time on it
17:09:10 <rloo> dtantsur: Sept 4 is a holiday in US & Canada, so that leaves next Mon.
17:09:19 <rloo> dtantsur: unless we do it outside our weekly meeting
17:09:20 <dtantsur> we can start today, the agenda is light
17:09:34 <dtantsur> then continue until this glorious event actually happens :)
17:09:43 <Nisha_Agarwal> dtantsur, ++
17:10:00 <dtantsur> anything else?
17:10:08 <rloo> dtantsur: i'm not sure i want to start today. should we ask folks to update the etherpad, and we'll start next monday?
17:10:29 <dtantsur> rloo: we have plenty of stuff in the etherpad, to be honest. at least a quick review won't hurt already
17:10:52 <rloo> dtantsur: a quick review is fine. just don't want to knock any out w/o giving folks a chance
17:11:04 <dtantsur> yep, let's leave it for the next meeting
17:11:14 <dtantsur> it = knocking stuff out
17:11:22 <dtantsur> moving on?
17:11:30 <rloo> +1
17:11:38 <dtantsur> #topic Review subteam status reports (capped at ten minutes)
17:11:47 <dtantsur> #link https://etherpad.openstack.org/p/IronicWhiteBoard line 133
17:11:59 <dtantsur> I don't expect a lot of updates this time, given that we've just released
17:12:34 <dtantsur> we've successfully moved our CI to using resource classes, starting with Pike
17:13:41 <rloo> dtantsur: what about https://review.openstack.org/#/c/443628/, wrt CI & resource classes?
17:14:07 <rloo> dtantsur: we want that test too?
17:14:22 <dtantsur> rloo: this is really nice to have, but it tests the technical details
17:14:28 <vdrok> this is a basic test to show the resources reporting is correct.
17:14:37 <dtantsur> our current integration tests verify that the whole procedure actually works
17:14:48 <rloo> ah, thx for clarifying
17:15:02 <dtantsur> I think we need to find a way to stop even setting properties in the CI
17:15:13 <dtantsur> that way we'll prove that we're actually using resource classes
17:16:26 <pas-ha> will those be still set by nova in instance_info? cuz we need e.g. size of root partition for partition images
17:16:34 <pas-ha> and that comes from flavor
17:17:02 <dtantsur> pas-ha: I think so; these fields are not removed from flavors any soon
17:17:08 <dtantsur> (well, dunno about soon, but not in Pike)
17:17:11 <pas-ha> ok
17:17:31 <sambetts> surely we still need some properties like disk_gb to be able to confirm that an image will fit onto a baremetal??
17:17:59 <dtantsur> interesting question. I'd say it can become optional now
17:18:29 <sambetts> what would happen if I try to put a 10gb image onto a 5gb disk then? we'd have to wait until deployment tries it and fails?
17:18:49 <dtantsur> sambetts: yes; but how realistic is this situation outside of the CI?
17:19:02 <dtantsur> as opposed to having to use inspection (often in-band)?
17:20:10 <sambetts> unsure what you mean, is Ironic now making the assumtuion that all images regardless of size will fit onto any baremetal ?
17:20:19 <wanyen> dtantsur: what if user use oout-of-band inspection?  Will it work?
17:20:37 <rpioso> +1 re: oob concern
17:20:47 <dtantsur> sambetts: I'm just saying that in 99% production cases the images will be much smaller than the disks
17:20:57 <pas-ha> dtantsur: +
17:21:08 <dtantsur> wanyen, rpioso, well, the inspection will work, but it won't set a resource class (not right now, at least)
17:21:21 * dtantsur senses a good topic for the PTG
17:21:29 <rpioso> dtantsur: +1
17:22:23 <wanyen> so Ironic already switch to rsc class scheduling in pike?  I thought rsc class scheduling will take effect in Queens
17:22:38 <rpioso> First class status for oob inspection is desired.
17:22:41 <sambetts> +1 I think that doesn't stop us from providing an early failure if someone happens to have a huge disk image they've copied from an existing baremetal so its not small, and then try to deploy it
17:23:04 <sambetts> I think we need to cover it at the PTG
17:23:05 <dtantsur> wanyen: it's not ironic that switches, it's nova. otherwise, it's possible to use them in pike, and they will be the only thing in queens
17:24:19 <vdrok> sambetts: dtantsur wrt image size, I think we can just introduce a new property for that that is not going to be picked up by nova?
17:24:27 <dtantsur> okay, we can discuss it later, please put your thoughts to the etherpad
17:24:40 <dtantsur> vdrok: we can reuse the same, nova just no longer will use it for scheduling
17:24:51 <dtantsur> anything else re statuses?
17:24:56 <vdrok> ah, ok
17:25:28 <dtantsur> moving on?
17:25:36 <rloo> +1 moving on
17:25:51 <dtantsur> #topic Deciding on priorities for the coming week
17:26:02 <dtantsur> ahemm... finish the release, review specs? :)
17:26:36 <wanyen> dtantsur, can you consider sum update for pike?
17:27:26 <dtantsur> wanyen: unfortunately, we're way past deadlines for inclusion in pike. I already included the iLO inspection update, despite it being somewhat too late..
17:28:37 <dtantsur> any ideas for the weekly priorities in addition to what I mentioned above?
17:29:16 <dtantsur> I would maybe add pas-ha's clients refactoring patches (for Queens, not Pike)
17:29:47 <rloo> dtantsur: i think we should have/add a section for candidate patches for release. unless you want us to add those links under 1?
17:30:01 <dtantsur> rloo: yep, I'd prefer them to be added there
17:30:20 <rloo> dtantsur: i am thinking doc-related. only. i hope.
17:30:24 <rloo> ok that works for me, thx.
17:30:38 <dtantsur> so, what about clients refactoring?
17:30:51 <vdrok> I'm good with that
17:30:53 <rloo> what client refactoring?
17:31:07 <dtantsur> I mean https://review.openstack.org/#/q/topic:bug/1699547
17:31:26 <rloo> dtantsur: oh, yeah. would be good.
17:31:48 <pas-ha> there's an interesting idea by vdrok in a neutron one
17:32:13 <pas-ha> but I'd like to get more opinions before diving into metaclasses... :)
17:32:23 * dtantsur -2 immediately :D
17:32:45 <dtantsur> I will check it
17:32:59 <pas-ha> thanks
17:33:25 <dtantsur> also note that in most cases a metaclass can be replaced by overriding __new__
17:33:40 <dtantsur> in this case even __init__, but I'll take a closer look
17:33:47 <dtantsur> are we ok with the priorities list?
17:34:08 <rloo> +1
17:34:14 * etingof would challenge dtantsur's opinion on metaclasses ;)
17:34:20 <vdrok> +1
17:34:37 <dtantsur> etingof: I did too many crazy things with sqlalchemy, now I have kind of an allergy :)
17:34:54 <dtantsur> #topic Open discussion
17:35:12 <dtantsur> wanna get a quick look at the PTG proposals?
17:36:02 <dtantsur> I'm open to other topics too, e.g. we can discuss properties/capabilities
17:38:00 <kaifeng> dtantsur: wrt node fault, any plan with it?
17:38:18 <dtantsur> kaifeng: I'm not working on it, if that's your question
17:38:32 <dtantsur> but this is something we should try finishing in Queens IMO
17:39:33 <dtantsur> pas-ha: will you be able to do some benchmarking of the ansible driver before the PTG?
17:39:57 <kaifeng> dtantsur: i mean, if it will be in the priority list, cuz the spec don't have enough reviews
17:40:18 <rloo> dtantsur: wrt pike goals, is there some xproject session at PTG that someone ought to attend?
17:40:30 <pas-ha> dtantsur: honestly not sure :( though we have some old results with probably not that big concurrency per node
17:40:34 <dtantsur> kaifeng: well, it's because we were trying hard to finish the Pike release. we're nearly done, so please expect more reviews starting this week
17:40:59 <dtantsur> rloo: see line 108 and below
17:41:12 <dtantsur> there are at least several other sessions of potential interest
17:41:23 <rloo> dtantsur: none of them are pike goals?
17:41:30 <dtantsur> oh, pike goals, sorry
17:41:31 <rloo> dtantsur: have the pike goals been picked even?
17:41:47 <dtantsur> you mean, queens goals? I think a few of them are already planned
17:41:57 <rloo> dtantsur: oops, yeah, queens.
17:42:09 <dtantsur> #link https://governance.openstack.org/tc/goals/queens/index.html
17:42:11 <rloo> mind your ps and qs...
17:42:20 <dtantsur> the policy one will have a session/workroom
17:42:31 <dtantsur> and I think there will be some status update on the QA track wrt the plugins
17:42:44 <dtantsur> I'm not aware of a session dedicated to getting more goals
17:42:48 <rloo> dtantsur: the qa stuff is what john signed up to do, right?
17:42:55 <dtantsur> yep :)
17:43:04 <rloo> dtantsur: maybe we should have a volunteer for the policy stuff
17:43:19 <dtantsur> yeah, we totally should. it does not look too hard
17:43:29 <dtantsur> also keep in mind that we probably should finish the pike goals too
17:43:33 <pas-ha> isnt that already done?
17:43:37 <rloo> jlvillal: please attend ptg session for 'split tempest plugins into separate repos/projects'
17:43:41 <pas-ha> i mean policy in code
17:43:49 <vdrok> yeah policy stuff seems like what we have already
17:44:05 <rloo> pas-ha: i think it might be but honestly, i haven't paid attention. someone should just take ownership/verify/
17:44:20 <pas-ha> https://github.com/openstack/ironic/blob/master/ironic/common/policy.py
17:44:54 <rloo> pas-ha: yes, so someone read the description for that goal and verify that ^^ satisfies it.
17:45:06 <rloo> c'mon someone, this is low-hanging-fruit! volunteer?
17:45:15 <pas-ha> '.. and document..' might be another strory, not sure how up-to-date is it for this
17:45:24 <pas-ha> I could
17:45:32 <dtantsur> thanks pas-ha :)
17:45:38 <rloo> and we have a winner! thx pas-ha! :D
17:45:49 <dtantsur> I'm putting a topic to the PTG to do a health check of the goals
17:46:33 <rloo> pas-ha, as part of that, you have to add policies to all ironic projects! ;)
17:46:51 <pas-ha> only those that have API
17:47:09 <pas-ha> anr policy.json
17:47:11 <pas-ha> and
17:47:15 <dtantsur> Nisha_Agarwal: did you have a chance to talk to nova folks about the capabilities? (your topic on line 75)
17:47:18 <pas-ha> which leaves inspector
17:47:42 <rloo> pas-ha: :)
17:47:42 <pas-ha> that much I'l handle :)
17:47:49 <Nisha_Agarwal> dtantsur, No. as we were still completing on pike, i couldnt do that
17:48:28 <dtantsur> Nisha_Agarwal: it would be good to have at least an initial discussion, since you're not going to the PTG to talk to them f2f
17:48:28 <Nisha_Agarwal> i intend to do a small POC from nova to ironic this week for quantitative capabilities and then initiate discussion with them
17:48:35 <dtantsur> sounds good
17:48:56 <dtantsur> okay, the list of topics looks good to me for now
17:48:59 <Nisha_Agarwal> dtantsur, yes. Thanks
17:49:01 <dtantsur> anything else to discuss?
17:49:40 * dtantsur wonders if we need any ironic-inspector related topics
17:50:22 <Nisha_Agarwal> dtantsur, i would like to add a low hanging fruit topic for inspector/ironic (Boot Management for Inspector)
17:50:37 <sambetts_> The split of discovery and inspection was something we discussed before
17:50:44 <dtantsur> Nisha_Agarwal: what exactly would you like to discuss? also note that somebody should be there to lead the topic..
17:50:58 <sambetts_> Neutron/inspector integration etc
17:50:59 <Nisha_Agarwal> dtantsur, :( yeah and i am not there
17:51:01 <dtantsur> sambetts_: please propose if you feel like discussing
17:51:25 <dtantsur> Nisha_Agarwal: I'll try to talk about it, it may be controversial
17:51:27 <Nisha_Agarwal> dtantsur, so we will have that discussion over RFE and IRC
17:51:29 <rloo> hmm, any network-related things to discuss at ptg?
17:51:40 <Nisha_Agarwal> dtantsur, ok thanks
17:51:51 <wanyen> I am interested in what needs to be done in oob inspection to work with rsc class scheduling
17:52:07 <kaifeng> +1 to neutron/inspector integration
17:52:36 <sambetts_> Dtantsur: Will do, although  etherpad on my phone is terrible so I'll have to add it later
17:53:02 <rpioso> +1 understanding oob inspection's interaction with resource class scheduling
17:53:06 <sambetts_> rloo: I can give a routed network update
17:53:18 <dtantsur> wanyen: I've proposed a topic for that, see line 99
17:53:21 <rloo> sambetts_: i'll add that to etherpad :)
17:53:48 <vdrok> I plan to get into network events, but this was discussed, just spec review left I guess
17:54:05 <wanyen> dtantsur, tx
17:54:18 <dtantsur> vdrok: yeah, I think we should Just Do It
17:54:24 <sambetts_> I think neutron events is important for queens
17:54:25 <pas-ha> I was actually having a crazy idea for inspector... will add  to the etherpad...
17:54:33 <pas-ha> sambetts_: ++
17:55:03 * dtantsur likes crazy ideas :)
17:55:09 <dtantsur> yes, please add stuff to the etherpad
17:55:14 <rpioso> dtantsur: Thank you, re: future of inspection
17:56:16 <rloo> 4-minute bell
17:56:47 <dtantsur> anything else or should we stop here?
17:57:03 <rloo> +1 stop. folks can add to the etherpad afterwards
17:57:20 <dtantsur> okay, thanks everyone!
17:57:32 <vdrok> thanks!
17:57:35 <dtantsur> #endmeeting