17:00:00 <jroll> #startmeeting ironic
17:00:01 <openstack> Meeting started Mon Jan 23 17:00:00 2017 UTC and is due to finish in 60 minutes.  The chair is jroll. 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'
17:00:04 <jlvillal> o/
17:00:05 <xek> o/
17:00:07 <vdrok> o/
17:00:10 <zhenguo> o/
17:00:13 <joanna> o/
17:00:13 <rloo> o/
17:00:13 <jroll> #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting
17:00:14 <dtantsur> o/
17:00:16 <jroll> agenda is there ^^
17:00:40 <stendulker> o/
17:00:43 <soliosg> o/
17:00:57 <jroll> alllllright let's do this
17:01:06 <jroll> #topic announcements and reminders
17:01:11 <jroll> crunch time!
17:01:15 <mjturek> o/
17:01:17 <jroll> #info client freeze this week
17:01:23 <bfournie> o/
17:01:26 <jroll> #info soft feature freeze next week
17:01:26 <krtaylor> o/
17:01:31 <hshiina> o/
17:01:31 <mgoddard> o/
17:01:34 <aslezil> o/
17:01:42 <jroll> #info goal for final release is in two weeks
17:02:12 <jroll> #info don't forget to add in things you want to discuss at the pike ptg:
17:02:14 <jroll> #link https://etherpad.openstack.org/p/ironic-pike-ptg
17:02:20 <xavierr> o/
17:02:34 <jroll> #info ptl nomination week, if you want to be PTL for pike please nominate yourself
17:02:55 <yuriyz|2> o/
17:02:57 <jroll> #info we got portgroups and attach/detach stuff landed in nova
17:03:04 <jroll> nice work everyone on that one, it's been a ride \o/
17:03:06 <jroll> I think that's it, anything I missed?
17:03:37 <rloo> yay, congrats on getting those features, nova patches landed!
17:03:45 <rpioso> o/
17:03:52 <rloo> (and I think we worked well with nova folks)
17:03:56 <jroll> ++
17:04:18 <rloo> jroll: wrt client freeze this week -- which day?
17:04:34 <jroll> rloo: thursday is the absolute deadline, I'd prefer to release wednesday
17:04:44 <jroll> that goes for most openstack-y release deadlines
17:04:55 <rloo> inspector client & ironic client, right? ok, will aim for wed.
17:05:09 <jroll> yep
17:05:16 <jlvillal> nice! :)
17:05:26 <jroll> I need to clarify IPA with doug, too, but I think we did that as a service last time
17:05:54 <rloo> jroll: one more question about client. i saw some mention of thinking about deprecating ironic cli in future. if so, we should make sure we have full osc support.
17:06:15 <jroll> rloo: yeah, that would be part of the plan
17:06:45 <jroll> anything else?
17:06:48 <rloo> jroll: ok, dtantsur brought up last week that we're missing at least two things in osc. i'll try to go through this week.
17:07:36 <jroll> rloo: okay, I don't think we need to rush that though
17:07:47 * jroll moves on
17:07:50 <jroll> #topic subteam status reports
17:08:03 <jroll> #link https://etherpad.openstack.org/p/IronicWhiteBoard
17:08:09 <jroll> starts at line 66
17:08:13 * jroll updates a couple things quickly
17:09:26 <rloo> vdrok, sambetts: wrt portgroups. are patches waiting for reviews, or still need to be created? I don't see any doc patches.
17:09:58 <vdrok> rloo: yeah, no docs yet. I think they will not be that big. will get to that this week
17:10:13 <rloo> dtantsur, wrt CI refactoring, there is more to do wrt refactoring, right? not just those two patches that got merged?
17:10:19 <rloo> vdrok: thx
17:10:24 <dtantsur> rloo, much more, I think
17:10:31 <dtantsur> we still haven't consolidated our jobs
17:10:37 <rloo> dtantsur: ok
17:10:38 <jroll> yeah, we never got into the CI refactors this cycle :(
17:10:42 <dtantsur> Lucas is pushing forward UEFI testing, which is extremely important IMO
17:11:02 <rloo> dtantsur: and there's still more to do wrt UEFI testing?
17:11:12 <dtantsur> rloo, we don't have a UEFI job yet
17:11:22 <rloo> dtantsur: ok. am going to note that down.
17:11:45 * jroll made some notes on nova + resource classes in the cross-project section, fyi
17:12:03 <jroll> soft power/reboot is looking good, nice!
17:12:25 <rloo> jroll: wrt rolling upgrades, did we decide to punt it to Pike? i'll update whiteboard with that.
17:12:42 <jroll> rloo: don't think we made an official decision, but that's how I'm leaning
17:13:25 <zhenguo> hi, wrt node tags, would you plan to land it before the final release?
17:14:06 <jroll> zhenguo: that would be cool, if we have time, maybe punting rolling upgrades makes some room for that
17:14:11 <rloo> zhenguo: node tags has an API / client component?
17:14:21 <jroll> if not, I would like to get it in very early in pike
17:14:24 <zhenguo> rloo: yes, API
17:15:25 <zhenguo> jroll: ok, and I will be on vacation soon
17:15:38 <zhenguo> so maybe early in pike is ok
17:15:48 <rloo> zhenguo: when will you be away?
17:16:03 <zhenguo> rloo: it's Spring Festival!
17:16:07 <jroll> zhenguo: lucky you! :)
17:16:14 <zhenguo> jroll: thanks
17:16:31 <rloo> zhenguo: enjoy :)
17:16:46 <jroll> zhenguo: I'll see if we can get a review this week, then, might be enough
17:16:48 <zhenguo> rloo: thanks, I will be out of office in 2 days
17:17:03 <zhenguo> jroll: thanks
17:17:12 <jroll> zhenguo: np, thanks for the update :)
17:17:33 <rloo> jroll: when's the nova deadline for features? do you remember? this week?
17:17:42 <jroll> rloo: yep
17:18:03 <rloo> jroll: ok, so we're shooting for soft power off & soft reboot?
17:18:16 <jroll> so I think our priorities look roughly the same as last week, though I removed rolling upgrades until we decide if we're punting or not
17:18:22 <jroll> rloo: yeah, one has a +A already'
17:18:51 <rloo> jroll: ok, i was looking at status, nmi won't make it in until pike.
17:19:18 <rloo> jroll: wrt resource classes. besides doc + testing, shouldn't need any code changes in ironic/ocata?
17:19:41 <jroll> rloo: agree re: nmi
17:19:45 <jroll> rloo: correct
17:19:53 <jroll> it's just metadata in ironic
17:20:01 <rloo> jroll: thx for confirming (both) :)
17:20:04 <jroll> :)
17:20:26 <jroll> is there anythign we should change/add in the priorities? I'm curious how much more BFV we want to do this cycle
17:20:43 <dtantsur> "nmi won't make it in until pike" why?
17:20:45 <jroll> and if we should get node tags or rescue mode up there in case folks run out of things to review :)
17:20:45 <rloo> jroll: i was just going to mention bfv. i think it is good to review but not a priority.
17:20:59 <rloo> dtantsur: need client release and nova patch. i think.
17:21:13 <jroll> if client merged, I can release that today
17:21:18 <rloo> dtantsur: nova freeze is this week
17:21:18 <mat128> o/
17:21:27 <dtantsur> yeah, nova patch is probably impossible, but our part can be finished
17:21:34 <jroll> idk how the nova thing is looking, but I feel like matt wanted to punt it to pike
17:21:49 <jroll> oh, we landed the client patch, yeah
17:21:55 <jroll> so we're done then right? :)
17:21:57 <rloo> dtantsur: i have no opinion either way on nmi. you guys decide.
17:22:18 <vdrok> ironic side looks pretty close yeah
17:22:31 <jroll> ah, osc
17:22:36 <rloo> according to status, looks like one osc patch
17:22:47 <TheJulia> jroll: I have some time tonight to work on BFM, but my week is largely travel + meetings this week, so I won't get much done on it
17:22:49 <dtantsur> yep, osc patch and api-ref.
17:22:54 <TheJulia> err, BFV
17:23:01 <jroll> TheJulia: okay, I'll deprioritize
17:23:11 <jroll> should we be focusing on node tags or rescue mode or both, then?
17:23:26 <dtantsur> node tags - maybe, yeah. it's been a while
17:23:30 <dtantsur> dunno how rescue looks now
17:23:37 <dtantsur> again, we need a nova part, right?
17:23:50 <jroll> yep
17:23:52 <rloo> jroll: node tags has 2 patches outstanding and i suspect they are simple (but i haven't looked at them). i don't know about rescue -- anyone know?
17:24:07 <jroll> rescue is mostly done other than IPA side
17:24:19 <jroll> "done" meaning code is up and has had many iterations
17:24:19 <rloo> i think if there haven't been much reviews on rescue, i would punt to pike
17:24:24 <JayF> It's our goal to have it working in devstack this week
17:24:28 <JayF> rescue
17:24:35 <JayF> all the ironic side stuff is pretty straightforward and done
17:24:41 <jroll> keep in mind that driver comp / rescue / node tags all have API version bumps, so be concious of merge conflicts please :)
17:24:43 <JayF> it's not been getting reviewed because it constantly needs rebasing
17:24:46 * rloo recalls past releases where we tried to get features in at the last minute, and shudders
17:25:14 <rloo> i think if we get driver composition done, we're lucky. am worried about pushing for more stuff this late.
17:25:14 <dtantsur> JayF, is there a nova patch for it?
17:25:26 <jroll> dtantsur: there is
17:25:30 <jroll> this is all on the whiteboard btw
17:25:36 <JayF> it's not likely to land in nova in pike
17:25:43 <jroll> in pike?
17:25:45 <jroll> why?
17:25:47 <JayF> er I mean
17:25:49 <JayF> until pike
17:25:50 <mat128> before pike?
17:25:52 <jroll> lol, yeah
17:25:57 <JayF> but if we punt it in ironic to pike, I'm concerned it'd miss nova in pike as well
17:26:09 <JayF> it seems to work best if we have a full ironic release of a feature before trying to get it into nova
17:26:20 <rloo> JayF: nah, not if it gets into ironic pike early.
17:26:21 <jroll> this is true
17:26:33 <rloo> JayF: rescue has an API component. client release is this week.
17:26:34 <jroll> idk, let's say if people have time, take a look and see how it feels
17:26:35 <dtantsur> too many client changes to land before Wed...
17:26:42 <rloo> dtantsur: ++
17:26:47 <JayF> rloo: that's a really good point, about client changes needing to land. dangit :(
17:27:07 <rloo> JayF: we are landing soft reboot, power off in both nova & ironic in ocata.
17:27:08 <jroll> I'm skeptical we'd finish rescue in a ~week anyway
17:27:20 <rloo> JayF: it can be done :)
17:27:25 <JayF> Yeah, that's fair. Can we make it something we work to land first thing in pike cycle?
17:27:34 <jroll> JayF: ++
17:27:38 <rloo> JayF: along with bfv...
17:27:44 <jroll> and rolling upgrades
17:27:45 <rloo> JayF: and rolling upgrades...
17:27:47 * jroll gets scared
17:27:48 <dtantsur> it's up to a new PTL to make sure the priorities are set early and get worked on ;)
17:27:49 <jroll> :D
17:27:58 <rloo> and that ^^. heh.
17:28:00 <jroll> :P
17:28:19 * jroll thinks we're all pretty much on the same page and expects no surprises with pike priorities
17:28:21 <rloo> although jroll can attest to how easy it is as ptl to have folks focus on priorities...
17:28:29 <jroll> haaaaaaa.
17:28:45 <TheJulia> heh
17:28:46 <jroll> alright, anything else for this bit of the meeting?
17:28:51 <rloo> so yeah, set priorities early. the rest... ??!!
17:29:05 <dtantsur> wait, aren't we going to make Ironic a proper CMDB finally?
17:29:07 <TheJulia> rloo: somewhere there is Profit.....
17:29:13 <jroll> /ban dtantsur
17:29:16 * dtantsur runs away very quickly
17:29:17 <TheJulia> dtantsur: z cycle maybe?
17:29:23 <rloo> dtantsur: add to ptg etherpad :D
17:29:35 * jroll still has a git repo named 'alanis' - a collection of all things ironic
17:29:43 <mat128|afk> lmao
17:29:57 <TheJulia> On a serious note, it might actually be worth discussing at the ptg
17:30:08 <jroll> alright, let's move on if there's nothing else
17:30:18 <jroll> TheJulia: quite possible
17:30:25 <jroll> #topic RFE review
17:30:39 <jroll> so, Nisha_Agarwal asks me about 3 times a week to look at this RFE
17:30:44 <jroll> #link https://bugs.launchpad.net/ironic/+bug/1656863
17:30:44 <openstack> Launchpad bug 1656863 in Ironic "RFE : Enhance pxe_ilo to do inband inspection also" [Undecided,In progress] - Assigned to Nisha Agarwal (agarwalnisha1980)
17:31:02 <jroll> so I'd like to see if we can approve it here, so that I'm done staring at it
17:31:10 <vdrok> :D
17:31:16 <jroll> it seems fine to me, although I'm skeptical if we need that config
17:31:31 <jroll> what do folks think?
17:31:41 <JayF> I mean, ilo driver has a pattern of making everything configurable
17:31:47 <JayF> I don't think it's too bad to offer that config
17:32:10 <JayF> That rfe seems very straightforward though
17:32:14 <TheJulia> It seems logical to approve given the context
17:32:16 <jroll> yeah, that's why I wanted to talk about it, I'm just not sure at all if we should have the config
17:32:29 <jroll> e.g. is it normal to only do OOB, or is it normal to do both, or?
17:32:54 <dtantsur> there is a spec for it fwiw
17:32:55 <jroll> I think I'm fine with it as-is but objections welcome
17:33:08 <jroll> dtantsur: this one breaks pxe_ilo out of that spec, since it doesn't depends on a boot driver
17:33:17 <jroll> boot driver in inspector*
17:33:23 <jroll> I told nisha that would be fine
17:33:46 <dtantsur> aha
17:33:58 <dtantsur> well, we already have (I think) oneview doing it
17:34:04 <dtantsur> should be fine
17:34:18 <jroll> ok cool, sounds like 4 +1s including myself
17:34:25 * jroll approves
17:34:38 <jroll> thanks y'all
17:34:41 <jroll> #topic open discussion
17:34:45 <jroll> anyone have a thing?
17:34:52 <joanna> I have :)
17:34:55 <rloo> why restrict to only doing OOB? (but i haven't been paying attention so ignore me)
17:35:15 <jroll> rloo: one reason would be to avoid waiting for the agent to boot if not needed
17:35:23 <JayF> rloo: That's not explained in the RFE, but I assumed it was so deployers could maintain existing behavior (OOB-only) as desired
17:35:26 <jroll> (inband just gives extra inspection details, that one might not need)
17:35:28 <jroll> also that
17:35:28 <stendulker> stendulker: that saves a agent boot
17:35:32 <jroll> joanna: go ahead
17:35:35 <joanna> remember the discussion before the Holidays about status codes and Retry-After?
17:35:43 <joanna> I created a spec for that: https://review.openstack.org/#/c/414194/
17:35:45 <TheJulia> I think the restriction is more a product detail than anything else
17:36:01 <stendulker> yes to maintain backward compatibility as well
17:36:06 <jroll> joanna: nice, dtantsur and lucas will be interested ^^
17:36:22 <rloo> it just seems odd to have an oob_inspection_only config. vs a config that allows you to specify oob and/or inband.
17:36:24 <jroll> probably sambetts too
17:36:34 <joanna> and there are opinions that status codes shouldn't be changed not to introduce such a big incompatibility between API versions
17:36:39 <dtantsur> I will get to the specs when we finish landing Ocata features, I guess ;)
17:36:45 <jroll> dtantsur: +1000
17:37:09 * jroll thinks we should only prioritize specs and almost-done-features for the first few weeks of pike
17:37:11 <joanna> and I would be really grateful if we could agree on how this change should look like :)
17:37:18 <TheJulia> rloo: Thinking about it, once driver comp is in, it doesn't _really_ make sense to have that config because there could just be an interface that goes "why not both?"
17:37:27 <joanna> I understand it's not a priority now, I just don't want this topic to die :)
17:37:36 <rloo> TheJulia: yeah, i was wondering how it fit with driver comp too
17:37:40 <jroll> joanna: ++
17:38:05 <stendulker> rloo: makes sense. Will work on that suggestion
17:38:07 <joanna> jroll: thanks! :)
17:38:11 * rloo votes for a PTL that promises to go through all the specs. ha ha.
17:38:26 <rloo> stendulker: thx for looking into it
17:38:42 * jroll will never promise that, not enough brainpower in the world
17:38:44 <TheJulia> rloo: All the specs! \o/
17:38:53 <jroll> stendulker: should I wait to approve that yet, then?
17:38:55 <TheJulia> joanna: Adding it to my queue
17:39:37 <jroll> ok
17:39:39 <joanna> TheJulia: awesome, thanks
17:39:40 <jroll> anything else?
17:40:32 <vdrok> jroll: rloo not sure how driver compostion would affect that tho, you still have a single inspect_interface column. To be able to disable in-band, you need a config option, or a separate interface, which would duplicate most of the stuff?
17:41:02 <TheJulia> vdrok: or a separate inspection interface that is aware of multiple options and that enumerates through them
17:41:09 <jroll> vdrok: yeah, dunno
17:41:24 <JayF> vdrok: you'd have like, 3 interfaces supported in ilo hardware type: IB only, OOB only, IB+OOB
17:41:26 <rloo> vdrok: and/or general configs, not specific to ilo.
17:41:38 <JayF> vdrok: as opposed to yet another ilo config option
17:41:48 <rloo> oh, i like JayF's suggestion
17:41:49 <JayF> jroll: ^ in light of this discussion, I'm not as +1 on that ilo rfe again
17:41:58 <jroll> lol
17:42:11 <TheJulia> JayF: +1 to that suggestion
17:42:12 <jroll> yeah stendulker said he wanted to investigate more, too
17:42:19 <jroll> can y'all comment on the rfe then?
17:42:34 * jroll moves back to unapproved
17:43:02 <vdrok> JayF: IB is already there, just inspector, it seems to be the matter of doing IB+OOB and OOB separately or together
17:43:12 <JayF> vdrok: exactly
17:43:18 <jroll> yep
17:43:24 <jroll> let's take this conversation to the RFE
17:43:35 <jroll> and then everyone gets a free 15 minutes
17:43:35 <TheJulia> ++
17:43:46 <vdrok> yup, thanks everyone!
17:43:47 <rloo> ah, the sound of crickets...
17:43:47 <TheJulia> \o/
17:43:51 <jroll> thanks folks
17:43:53 <jroll> #endmeeting