19:01:17 #startmeeting ironic 19:01:19 Meeting started Mon Nov 24 19:01:17 2014 UTC and is due to finish in 60 minutes. The chair is devananda. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:20 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:01:22 The meeting name has been set to 'ironic' 19:01:26 morning all whos here for the Ironic meeting 19:01:28 o/ 19:01:34 o/ 19:01:34 \o/ 19:01:35 o/ 19:01:36 hi all! 19:01:36 o/ 19:01:38 o/ 19:01:41 \o/ 19:01:55 hi 19:01:56 \o 19:01:57 o/ 19:02:01 (in this meeting this time) 19:02:01 o/ 19:02:02 o/ 19:02:07 jroll: :) 19:02:08 o/ 19:02:45 as usual, the agenda is posted here: 19:02:47 #link https://wiki.openstack.org/wiki/Meetings/Ironic 19:03:00 to note, the format of the meeting has changed 19:03:16 #topic Announcements 19:03:19 status information posted here: 19:03:22 #link https://etherpad.openstack.org/p/IronicWhiteBoard 19:03:38 #undo 19:03:38 Removing item from minutes: 19:03:48 er, sorry - meant to undo my topic change 19:03:49 oh jumpped the gun 19:03:49 #undo 19:03:49 Removing item from minutes: 19:04:08 as NobodyCam helpfully linked, the subteam status reports were prepared ahead of time 19:04:11 and posted to the etherpad 19:04:17 #link https://etherpad.openstack.org/p/IronicWhiteBoard 19:04:38 Those are to be sent out to the ML to be persisted (in archives), correct? 19:04:51 yes 19:04:55 we'll go over those after the announcements, and I'll email a summary afterwards 19:05:00 JayF: after the meeting 19:05:01 perfect 19:05:02 that's the largest change to the meeting format 19:05:10 ok -- moving on now, for real :) 19:05:12 #topic announcements 19:05:36 I’d like to call attention to old specs to be abandoned after this meeting 19:05:39 #link https://etherpad.openstack.org/p/ironic-juno-specs-to-be-removed 19:05:45 this is Thanksgiving week in the US, so folks who recognize that holiday (or jsut want time off work) may be less available 19:05:50 (unless updated to kilo cycle) 19:06:18 I'll be offline Thursday, but probably here Friday 19:06:48 #topic subteam status review 19:06:52 I'll be off thurs and mostly kinda around fri 19:07:07 * devananda goes and looks at the etherpad for any last-minute changes he hasn't seen 19:07:30 thank you all for the up front status updates... I like this format much better :) 19:07:44 JayF: woot! good stuff on the agent tests 19:07:58 JayF: need to discuss that bug at all today? 19:08:02 * JayF points mostly at jroll 19:08:10 * jroll hides 19:08:15 devananda: if anything, wanna discuss making that job voting for IPA 19:08:24 devananda: even if we want to wait on Ironic due to the recheck bug 19:08:27 devananda: maybe we should let adam_g discuss that bug, I know he's been diving into it 19:08:51 if that bug is close, let's wait and just make it voting on both 19:08:58 jroll: yes I would like adam_g 's input there 19:09:04 if we're still unsure, let's set to voting on just IPA for now 19:09:05 yeah--ive been trying to get as much debug about the it as i can 19:09:09 * jroll 2c 19:09:17 but its still a bit of a mystery 19:09:39 #link http://bugs.launchpad.net/bugs/1393099 19:09:43 Launchpad bug 1393099 in openstack-ci "test_baremetal_server_ops fails with timeout waiting for provision_state to change" [Undecided,Confirmed] 19:09:56 once in a while, the VM node just fails to reach back to the hosts TFTP server, with a connection timed out error 19:10:08 JayF, jroll: even after chatting with clarkb last week, i'm still unclear on whether or not making that job voting on IPA is going to introduce it into the serialized gate 19:10:10 ive had the test running in two places in a tight loop hoping to catch it but haven't 19:10:27 devananda: I don't know enough about the "serialized gate" to understand that 19:10:31 but, one way to find out is just to do it, see, and if we were wrong (and it did) then back it out right away 19:10:55 think of jobs as being transitive for generating the list of projects that are cogated 19:10:56 devananda: making that job voting on *IPA* won't, making that job voting on *Ironic* will 19:11:01 devananda: was what I thought clarkb was saying 19:11:09 adam_g: it fails to connect at all, yes? not that it fails to get the whole image there? 19:11:16 so if nova runs jobs A and B, ironic runs jobs B and C, and IPA runs C then they are all in the same gate pipeline 19:11:17 (wondering if using ipxe would help at all) 19:11:22 jroll, right 19:11:49 clarkb: and if IPA runs jobs C and D -- does D have the ability to affect Nova? 19:12:00 devananda: if there is a transitive path to nova yes 19:12:08 right 19:12:09 eg D to C to B 19:12:12 so that's my concern 19:12:29 instability in this test, even if it only runs on changes to IPA, will transitively affect changes in Nova 19:12:32 clarkb: ironic runs agent_ssh-nv, IPA runs agent_ssh-src-nv, would that be a valid path or considered two different jobs? (I think -src is just a parameter to agent-ssh) 19:12:33 not because it runs against them 19:12:52 its name based so those are distinct 19:12:57 but because it would have the potential to reset the job queue for a Nova change 19:13:06 ok, so we should be fine then, I would think 19:13:16 it's not a direct path at that point 19:13:54 eg, if the zuul job queue is (IPA patch) -> (Nova patch) and that test failure causes the IPA patch to get kicked out, it resets the jobs on the Nova patch too 19:14:06 clarkb: is that accurate? 19:14:20 devananda: yes 19:14:31 yay! 19:14:39 So that won't be true in the case of IPA/Ironic though 19:14:46 because as jroll pointed out, the jobs have different names 19:15:04 and slightly different characteristics which prevent that from happening (IPA job builds from src, and is named as such, Ironic job pulls from tarballs.o.o) 19:15:37 ah, ok. 19:15:49 if there is no job overlap between ironic and IPA at all -- then yes, it's fine 19:15:54 but that is not my understanding 19:16:22 there are /some/ jobs that run on both ironic and ipa changes and /some/ that run on only one. thus, there is still a transitive dependency 19:16:32 (we are over our 10 minute cap) 19:16:34 I don't believe that's true ^ 19:16:35 https://review.openstack.org/#/c/131632/ 19:16:37 https://review.openstack.org/#/c/132746/ 19:16:41 NobodyCam: thanks for the time check 19:16:41 I don't see any name overlap 19:16:49 jroll: evidence++ 19:16:56 *exact* name overlap, that is 19:17:36 let's come back to that after the meeting 19:17:43 I want to check the current jobs, not just the proposed ones 19:17:47 ok 19:18:11 I don't see lucas online today? 19:18:23 is anyone else here who wants to lead the boot / deploy interface split discussion? 19:18:27 he is off, or at least said he maybe off today 19:19:05 #topic split the boot and deploy interfaces 19:19:49 is there a spec for that? I looked but didn't see anything (obvious) 19:19:53 do we even want to talk about this without lucas? 19:20:05 he's done the most research into it 19:20:19 +1 to skip then 19:20:21 I don't see any spec for it yet -- just checked again 19:20:47 #info deferring this discussion since the lead is absent, and there doesn't appear to be a spec 19:21:00 ya I was just looking for that 19:21:04 #topic vote on the bear's name 19:21:12 oh hi 19:21:17 another topic started by lucas, but I think NobodyCam wants to lead :) 19:21:29 I added a list of the votes to the bottom of the whiteboard 19:21:39 that cam thru the ML 19:21:47 s/cam/came/ 19:22:12 not sure how the best way to vote on the name 19:22:36 NobodyCam: can you give a little background for folks that might have missed the email? 19:22:45 I am not sure that the meeting is a good place to vote. Not everyone can attend this meeting. 19:22:56 sure 19:23:07 although what's a meeting for, if you can't vote there... 19:23:20 I can start a vote here 19:23:30 with the names as options. everyone gets one vote 19:23:35 lucas created a mascot for Ironic and we have been adding potitional name to the whiteboard 19:23:50 Lucas started a ML thread about selecting a name 19:24:28 the results from the ML thread are on the whiteboard around line 179 19:25:05 let's not spend a lot of time on this today ... 19:25:18 I'm fine with voting by mail 19:25:26 (says the guy who hasn't voted on that thread yet) 19:25:41 I am good with that :) 19:26:05 #info If you want to vote on the mascot's name, please reply to the mailing list thread 19:26:06 devananda: not voting on the thread could mean you are worrying about more important things than bear names ... like bare metal 19:26:14 I think lucas wanted to pick the top 5 of the suggestions, and have folks vote on those. but he isn't here, so I say let's let him deal with it ;) 19:26:14 just a quick posting of the current top picks: 19:26:16 JayF: possibly ... 19:26:23 Bearthoven 5 19:26:24 Blaze Bearly 4 19:26:24 Ursus 2 19:26:24 Pixie 2 19:26:24 Pixie Boots 2 19:26:26 Ironica 2 19:26:38 capping this to one more minute 19:26:41 (sorry for large paste) 19:26:50 * NobodyCam is done 19:27:03 ok, moving on now :) 19:27:10 #topic IRC meeting times 19:27:23 this came up at the summit 19:27:32 which is another place that, unfortunately, not everyone could be 19:28:15 at the last meeting, we discussed some time slot options 19:28:23 and I followed up with an email to the list with two proposals 19:28:52 I've tallied the replies here: http://paste.openstack.org/show/jjharvUG2JGpUVoyZq9F/ 19:29:09 I believe I liked #2 but cann't lay my hands on hte archive like 19:29:21 (Unfortunately people started voting on the mailing list, when it was really a call to veto, IIRC) 19:29:37 devananda: I did not +1 option #1 19:29:42 devananda: I only +'d option #2 19:30:01 JayF: oh - apologies. I must have misread your reply :( 19:30:08 mrda: I did actually ask folks to reply to the thread with a vote 19:30:28 mrda: since the point is to not expect everyone to make the current meeting time, it seemed contradictory to expect folks to show up here and vote :) 19:30:31 devananda: no vote meant "fine with either", correct? 19:30:35 If you don't mind the quote: 19:30:37 aiui; it was vote if you could / would attend both times, don't vote if you could attend one in either proposal 19:30:38 Please respond with a "-1" vote to an option if that option would cause you to miss ALL meetings, or a "+1" vote if you can magically attend ALL the meetings. If you can attend, without significant disruption, at least one of the time slots in a proposal, please do not vote either for or against it 19:30:45 and minus-vote if you couldn't attend any of them 19:30:50 correct 19:31:06 I received no "-1" votes at all 19:31:21 looks like #2 is the winner, then? 19:31:22 which means that both options can accomodate everyone attending at least one half the meetings 19:31:32 and more + votes on option #2 19:32:01 if anyone has objections, please raise them now 19:32:06 will any required room changes be posted to the aganda wiki? 19:32:30 otherwise I'll go ahead with arranging a meeting time change based on Option #2, update the appropraite wiki's and calendars, and send out an announcement later today 19:32:36 NobodyCam: yes 19:32:41 :) 19:32:43 for the 'record', what is option 2 and yay, a decision was made! 19:32:55 #info Option #2: alternate between Monday 1700 UTC && Tuesday 0500 UTC 19:33:00 Actually, my preference is to leave things how they are, but I am happy with group consensus :) 19:33:10 1700: and 0500 alternating weeks 19:34:13 devananda: when do we start the new meeting time? 19:34:41 devananda: from next week? 19:34:53 naohirot: I will announce all that on the mailing list after this meeting 19:35:03 once I've updated the calendars 19:35:09 devananda: ok 19:35:38 that's it for our agenda! phew! 19:35:40 #topic open discussion 19:36:04 I like this new format :) 19:36:19 http://animals.howstuffworks.com/mammals/zebra-stripes.htm 19:36:24 I have a small devstack change up which should help with diagnosing IPA test failures, if anyone wants to have a look: https://review.openstack.org/#/c/136867 :) 19:36:24 I'd like to bring up the midcycle meeting timing 19:36:25 lol 19:36:48 (didn't add it to the agenda last week, since I was still checking with folks internally) 19:37:13 thats what open discussion is for 19:37:15 devananda: where will the midcycle meeting held? 19:37:33 I have a solid offer from HP to host it in Grenoble 19:37:50 devananda: and do you have firn dates? 19:37:55 s/firn/firm/ 19:38:11 either the week of Jan 27 - 30, or the week of Feb 3 - 6 19:38:23 both are possible for the host 19:38:24 we are going to attempt to setup video / audio for those folks who can not make it in person? 19:38:52 I'm open to either afaik 19:38:54 NobodyCam: that's nice :-) 19:39:00 (not sure yet if I can attend in person) 19:39:08 NobodyCam: I'd like us to make a much stronger effort for that than we have in the past, but I think we need to know in advance who will be attending remotely 19:39:12 * JayF would not attend either set of dates, but would love to attend emotely 19:39:19 *remotely 19:39:39 I like emotely, use all the emotes :D \o/ :P 19:39:43 I will attend remotely if it can be arranged 19:39:44 worth a section on the whiteboard? 19:39:51 yes 19:39:53 * mrda probably couldn't make the early set of dates 19:40:05 we should probably mail the list and get a feel for who could/would come 19:40:13 * devananda arranges an informal poll 19:40:17 http://doodle.com/4dkpqxr235m35rbt <<< poll! 19:40:24 (does that link work?) 19:40:34 yup :) 19:40:38 great 19:41:23 also, fwiw, I chose those dates because they coincide with kilo-2 milestone 19:41:50 quickly added to whiteboard 19:41:52 much later in the cycle than that (like, too close to feature freeze) seems like it would be more challenging 19:41:58 to get any code done 19:42:37 kilo-2 milestone is week of Feb 5 19:42:51 yup 19:43:32 those dates also coincide with FOSDEM 19:43:58 ok, i'll email the ML with that info 19:44:05 and those dates happen to be 'mid cycle' :-) 19:44:41 any other topics? 19:44:42 coincidence! :) 19:45:14 devananda: do/will we have dates wrt specs? 19:45:23 rloo: as in spec freeze, etc? 19:45:27 yup 19:45:30 rloo: https://wiki.openstack.org/wiki/Kilo_Release_Schedule 19:45:31 if we have time I'd love to hear what the thoughts are about the new state machine spce 19:45:52 devananda: thx. the 'feature*' dates then. 19:46:06 I've been doing a lot of spec reviews last week (like, hardly anything else) 19:46:22 and jsut have to say, there are a lot of really good specs up there! 19:46:35 yes! 19:46:53 NobodyCam: I think we're waiting for victor to post a new revision? 19:47:07 ...so we might just have a lot of really good code to review this cycle :P 19:47:29 should have one soonish according to his reply to my just before meeting ping :) 19:49:03 well, if there's nothing else to discuss here, let's wrap up early :) 19:49:09 thanks everyone! 19:49:16 w00t ' 19:49:22 thanks devananda 19:49:26 great meeting 19:49:30 #endmeeting