19:01:47 #startmeeting ironic 19:01:48 Meeting started Mon Jul 21 19:01:47 2014 UTC and is due to finish in 60 minutes. The chair is lucasagomes. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:01:49 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:01:50 o/ 19:01:50 o/ 19:01:52 The meeting name has been set to 'ironic' 19:01:58 hi all 19:01:59 \o 19:02:07 as usual, our agenda is available up here: 19:02:11 #link https://wiki.openstack.org/wiki/Meetings/Ironic 19:02:12 wanyen 19:02:17 #topic announcements 19:02:32 so first, I'm chairing the meeting today because devananda and NobodyCam is at the TripleO sprint this week 19:02:43 yay lucasagomes! 19:02:51 woohoo 19:02:52 and... next week (July 28th) we will have the Ironic sprint in Beaverton \o/!! 19:02:57 \o/ 19:02:59 #link https://wiki.openstack.org/wiki/Sprints/BeavertonJunoSprint 19:03:02 more woohoo 19:03:12 wohoo*2 19:03:18 There's a small pre-meetup pizza outing on Sunday 27th, if you're going to be there and is interested in going please put ur name here: 19:03:22 #link https://etherpad.openstack.org/p/juno-ironic-sprint 19:03:38 if you're coming, please also take a time to think about our goals for the midcycle and edit it on the etherpad above ^^^ 19:03:54 one more thing on the announcements 19:03:56 * dtantsur is not coming ;( 19:04:04 we won't be having this meeting next week due to Ironic midcycle meetup!! 19:04:09 #info no weekly meeting on July 28, due to Ironic midcycle meetup 19:04:15 that's all the announcements, I think! 19:04:37 #topic release cycle progress report 19:04:55 specs... thanks for everybody which are reviewing them 19:05:04 we have a lot of specs proposed and we managed to merge a good part of it 19:05:14 But still I think we need more core people in the ironic-specs groups 19:05:23 if you're core reviewer in Ironic and want to help with the specs, please lemme know 19:05:27 +10^10 19:05:29 !!! 19:05:29 dtantsur: Error: "!!" is not a valid command. 19:05:34 hah 19:05:35 I could help 19:05:42 russell also returns tomorrow 19:05:46 I would be less optimistic about the current state: we're in trouble with specs :( 19:05:55 thanks guys! 19:06:01 yet on specs 19:06:10 we are now approaching our spec propsal freeze, this Thursday (July 24th) is the last day we have to propose a spec, more info see: 19:06:15 #link http://lists.openstack.org/pipermail/openstack-dev/2014-June/039031.html 19:06:15 (you don't need to be a core-revewer to be a core-spec-whatever) 19:06:28 lucasagomes: I will try to help review more specs 19:06:30 dtantsur: OTOH, we can only land so much code this cycle, so... we can't approve everything 19:06:36 rloo, yeah, that's true russel is there 19:06:46 jroll, we didn't approve everything we're going to land actually 19:06:50 Per suggestions from last week’s chat room discussion, we have refactored iLO virtual media deploy driver into iLO Virtual Media iSCSI Deploy Driver https://review.openstack.org/#/c/97744/ and iLO-IPA Deploy Driver https://review.openstack.org/#/c/97744/. We would like to target these 2 specs for Juno. Please review these specs as soon as you can. Thanks! 19:06:52 so anyone that wants to help with the specs please do! 19:06:58 dtantsur: such as? 19:07:14 wanyen, right, feel free to bring that up again at the open discussion 19:07:19 jroll, IPA, discovery, something-else-proposed-by-you, not everything from iLO 19:07:40 #info spec proposal freeze on July 24! 19:08:01 dtantsur: true. let's do that this week :) 19:08:09 dtantsur, right we will talk more about specs on the discussion topic, we have an entry for that 19:08:15 jroll, would be great :) 19:08:18 ack 19:08:22 about the ironic nova driver, last week in the nova weekly meeting this topic was raised and there's still some things to iron out with the nova guys about whether the driver can be merged in nova or not 19:08:23 lucasagomes: that date is a freeze for new proposals, or is the date where no more specs will be approved for juno? 19:08:33 rloo, for new proposal 19:08:35 rloo, new ones 19:08:43 thx :-) 19:08:52 after that date, specs will be -2'd and reopened in the K cycle (afaiui) 19:09:02 tl;dr the spec for the driver was merged but the spec to the upgrade path from nova baremetal to ironic wasn't, BUT, it has been proposed for a spec-freeze exception 19:09:10 devananda, summarized the discussion on the ML, please check it out: 19:09:13 lucasagomes: the only thing stopping the nova ironic driver merging is reviews really, and most of that is in our court 19:09:16 #link http://lists.openstack.org/pipermail/openstack-dev/2014-July/040627.html 19:09:36 mrda, there was a problem with the migration path from nova bm to ironic 19:09:43 the spec for the migration didn't get merged yet 19:09:47 mrda: please elaborate? we can only +1 those reviews? 19:10:00 but it has been proposed for a spec-freeze exception... and we hope that it's going to land on J 19:10:03 yes, that's one of many hurdles :) 19:10:09 rloo, yup we can 19:10:14 yeah 19:10:27 check the ML link above, deva summarized it nicely 19:10:39 lucasagomes, mrda, so you're saying we should, cuz the more +1s the more the core-reviewers will notice? 19:11:15 rloo, well... I bet it helps 19:11:19 it's unclear why us +1'ing the nova reviews would help... since we've already landed the code, we're probably ok with the state 19:11:33 yeah that's true 19:11:42 hang on, I can explain 19:11:49 (can't type quick enough) 19:11:50 mrda, go for it :) 19:12:19 So the proposed patches are https://review.openstack.org/#/c/{103164,5,7} 19:12:49 when a nova reviewer adds comments we take them and propose a ironic reviewer with the fixes for that regview comment 19:12:56 #link https://review.openstack.org/#/c/103164 19:13:01 we need ironic reviewers to sign off and approve those changes 19:13:05 #link https://review.openstack.org/#/c/103165 19:13:06 #link https://review.openstack.org/#/c/103167 19:13:21 and then we propose the new changes back onto the original nova reviews 19:14:03 so https://review.openstack.org/#/c/107316/ and https://review.openstack.org/#/c/102695/ and https://review.openstack.org/#/c/107882 all need ironic eyeballs and +2's 19:14:06 mrda, ah right, i saw some patches like that in Ironic, so we should priorize them I assume 19:14:13 mrda: is it obvious which patches are ironic-patches-addressing-nova-comments 19:14:29 the current state is listed on the whiteboard (https://etherpad.openstack.org/p/IronicWhiteBoard) 19:14:40 lucasagomes: yes please 19:14:56 so deva was doing this, I'm now doing it for him to allow him to do other things 19:15:06 I'm the patch monkey :) 19:15:07 #info cores to prioritize the reviews to the nova driver 19:15:21 alright... ta much mrda for the info 19:15:21 anteaya: hi; at some point you were trying to unify the names used by third party testing systems; the changes you suggested have not taken place. right? just wanted to double check. thanks. 19:15:26 let's move on? 19:15:36 does anyone have questions on this? you can ping me in channel 19:15:43 ^ 19:15:49 #topic subteam status reports 19:16:05 adam_g, hi! care to summarize the tempest work from last week? 19:16:19 Shrews, ^ (I know ur at the tripleO midcycle) 19:16:20 (sorry for asking my question at the wrong place) 19:17:23 alright they might be afk... let's move on 19:17:35 dtantsur, hi there! anything regarding bug status and fedora support? 19:17:52 lucasagomes, nothing on Fedora, to the bugs: 19:18:01 Open: 124. 9 new, 41 in progress, 0 critical, 16 high and 8 incomplete 19:18:10 (stats are generated using my new script, so not diff this time) 19:18:28 #info Bug stats: 124 Open, 9 new, 41 in progress, 0 critical, 16 high and 8 incomplete 19:18:55 9 new bugs - I have problems triaging some of them, they require deeper knowledge of Nova 19:18:59 dtantsur, thanks! the script is hosted on ur github account? u wanna share it so ppl can take a look? 19:19:00 will be glad for any help 19:19:07 moment, trying to find :) 19:19:21 https://github.com/Divius/ironic-bug-dashboard 19:19:32 can output stats and new bugs :) 19:19:34 #link https://github.com/Divius/ironic-bug-dashboard 19:19:39 more features may come on request 19:19:44 dtantsur, ta much! 19:19:47 actually it may even turn into web dashboard 19:20:00 that will be great 19:20:01 that's all for me, will appreciate help with triaging 19:20:19 jroll, JayF hi there! any updates on IPA? 19:20:29 h! 19:20:31 hi! 19:21:11 so as always, our todo list is here: https://etherpad.openstack.org/p/ipa-todos 19:21:22 #link https://etherpad.openstack.org/p/ipa-todos 19:21:40 we have one +2 on the spec, would love one more https://review.openstack.org/#/c/98506/ 19:21:54 o/ 19:22:17 JoshNang's swift temp url patch landed... new swift client should be released very soon 19:22:30 which unblocks https://review.openstack.org/#/c/81391 19:22:40 which is the last dep for the main agent patch, https://review.openstack.org/#/c/101020 19:22:52 nice 19:23:12 I also put up a WIP patch for devstack... it's not ready and not working 100%, but I'm close. I think it's a config issue https://review.openstack.org/#/c/108457/ 19:23:19 i'm told a new client should be out shortly, then i'll put in a patch to bump the global reqs file 19:23:28 jroll: https://review.openstack.org/#/c/73920/ is what we're waiting on for a swiftclient release. up to jenkins now 19:23:33 and to go with that, WIP etherpad for testing IPA with devstack https://etherpad.openstack.org/p/ipa-devstack 19:23:36 notmyname: yep :) 19:23:39 I will try to review the spec very very soon, part of the midcycle will be focused on trying to land IPA, I'd like to try it out as well 19:23:46 ah great! devstack will be handy for that 19:23:51 indeed :) 19:24:11 I plan to get that working this week :) 19:24:29 +100000 19:24:45 I think that's it from me... one thing that was brought up this morning though 19:24:46 Sounds like a good plan ;) 19:24:53 good stuff! 19:25:05 jroll, JoshNang notmyname thank u 19:25:12 auth for IPA - touchy subject :| 19:25:28 yup... I know u guys don't wanna pass tokens around 19:25:29 we're of the opinion that keystone admin tokens should not be passed to the agent 19:25:38 and instead operators should secure the network 19:25:54 idk if we should talk about this now or maybe just next week 19:26:08 well let's bring it to the open discussion ? 19:26:15 sounds good 19:26:15 people may throw some nice ideas 19:26:17 ack 19:26:29 romcheg: howdy! any updates on the nova bm db > ironic db migration? 19:26:43 lucasagomes: there are some news as usual :) 19:27:03 But this time it's bad news 19:27:08 :( 19:27:21 Nova team blocked my patch because the spec is not yet ready 19:27:45 romcheg1: link? 19:27:49 They suggested to ask for for a SFE 19:27:52 yeah, but it has been proposed for a spec-freeze exception, right? 19:27:58 right 19:28:32 But it's complicated with Grenade testing 19:28:41 #link https://review.openstack.org/#/c/95025/ 19:28:45 mrda, ^ link for the spec 19:28:53 Thanks lucasagomes 19:29:02 #info I still need some volunteers for helping me with Grenade testing 19:29:08 romcheg1: and for the blocked review too please? 19:29:14 romcheg1, yeah I heard grenade is not well documented and not that robust yet 19:29:20 I thought, from reading the list, grenade testing was impossible due to bm not working in devstack? 19:29:23 * mrda will follow up with some Nova people 19:29:29 lucasagomes: s/not well/not/g 19:29:34 heh 19:29:47 JayF, yea... I think it's been discussed in the ML right now 19:29:53 http://lists.openstack.org/pipermail/openstack-dev/2014-July/040627.html 19:29:59 mrda: https://review.openstack.org/#/c/91165/ 19:30:00 romcheg1, lol ok 19:30:09 mrda: wrong! 19:30:38 mrda: wrong link. Lucas gave you the right one 19:30:45 ta 19:30:55 romcheg1, anything else from u? 19:31:27 lucasagomes: I wrote some scripts to Grenade that run my migration scripts to turn Ironic to Nova BM 19:31:32 No 19:31:44 The opposite, sorry :) 19:31:50 Nova BM to Ironic 19:31:56 nice... yeah nova bm to ironic 19:32:01 are the scripts available somewhere? 19:32:25 No, they are not, because I did not figure out all the dependencies 19:32:48 Probably I will just patch the scripts to cherry-pick migration scripts until they are merged 19:33:10 Cross-project dependencies make thinks harder 19:33:29 indeed it is 19:33:46 romcheg1: even getting them up on your personal github would be good 19:33:57 right, there's also that mailist that I pointed before, you may want to put there ur findings 19:34:01 mrda: posting is not a problem 19:34:37 the problem is that those scripts rely on the other scripts that are not merged yet to another project 19:34:45 ok 19:35:00 I will figure out smth fast I think 19:35:10 ack, or even an etherpad with instructions 19:35:11 romcheg1: just wip it until the dependency is merged 19:35:18 etherpad++ 19:35:36 so people willing to help you can get up to speed 19:35:45 lucasagomes, jroll, I will update the etherpad 19:35:52 romcheg1, ta much! 19:36:07 my pleasure :) 19:36:13 alright, anything else on grenade? upgrade path? 19:36:25 no, that's all 19:36:32 ok 19:36:36 thanks 19:36:50 GheRivero, hi there! any updates on Oslo? 19:37:36 ok GheRivero is at the TripleO mid-cycle and might be afk 19:37:55 so let's go for the discussion topics 19:38:00 # topic Spec reviews bottleneck 19:38:08 dtantsur, you added that one... wanna start? 19:38:17 #topic Spec reviews bottleneck 19:38:32 my point was mostly stated: we have to little eyes on specs 19:38:45 that's especially bad with specs raised by one of us 19:39:02 yeah :( 19:39:12 currently there is ~3 active specs core :) 19:39:24 russell is back from two weeks off tomorrow 19:39:33 jroll, great news! 19:39:39 \o/ 19:39:42 should be extremely helpful 19:39:54 I'd still have at least 7 cores 19:39:58 but I agree, more spec cores ftw 19:40:03 right, so the current spec-core team is 19:40:05 #link https://review.openstack.org/#/admin/groups/352,members 19:40:18 we're under pressure with a lot of prioritized specs under review and just as many non-prioritized 19:40:19 if you wanna be part of it, let us know :) 19:40:28 do we follow the same process for adding them, or just do it? 19:40:34 * jroll nominates JayF 19:40:54 * JayF hears the rumble of a bus coming 19:40:57 stats here: http://stackalytics.com/report/contribution/ironic-specs/30 19:41:01 jroll, well I wouldn't mind adding a active member of the community... I think devananda can answer that better tho 19:41:08 lucasagomes: right 19:41:22 so yeah, I'm +1 to JayF 19:41:27 lucasagomes: is that something you can discuss at midcycle next week? 19:41:35 we can ask devananda, when he's available 19:41:41 rloo, definitely ! we should 19:41:47 rloo, I would say we need it earlier... 19:42:24 dtantsur: like yesterday? :-) Chris has done a lot of reviews -- someone twist his arm. 19:42:29 rloo, mind adding the topic to https://etherpad.openstack.org/p/juno-ironic-sprint ? 19:42:43 right, the spec cutoff is thursday 19:43:01 yeah :/ but still we will have to continue reviewing the specs that are in the queue 19:43:11 judging by stats and my experience, I would nominate NobodyCam, JayF and rloo ;) 19:43:15 I'll make a point to review specs one way or the other 19:43:27 Cool, thanks! 19:43:29 lucasagomes: done. 19:43:34 rloo, thanks 19:43:57 alright anything else on the specs? or should we move to the open discussion? 19:43:58 dtantsur: I'll think about it. 19:44:15 moment 19:44:30 I would ask people knowing about iLO to follow iLO-related specs 19:44:42 it's no good, if lucasagomes and I will be approving them :) 19:44:50 lol 19:44:50 that's it from me 19:44:52 hah indeed 19:45:40 for ilO specs, we can still get input form your folks about overall design. 19:45:48 s/form/from 19:45:53 ^^ 19:46:09 wanyen, sure, we will provide it 19:46:21 yeah, we can help with that :) 19:46:30 we need reviewers for UEFI boot. We uplaoded a new rev to address Deva's comment but siince then we haven't got new review. 19:46:36 but it's always helpful to get feedback from somebody knowing a thing or two about iLO :) 19:47:01 wanyen, that's why I say, we have problems with specs 19:47:16 there are so few reviewers knows iLO so it will be review bottleneck 19:47:51 Nobodycam is in Raleigh this week. I can relay anything you need relayed 19:47:52 wanyen, there's only 3 active specs core now. I'm gonna schedule more reviews tomorrow 19:47:59 well we don't need to try everything, we can look at it, see if the code is good and vote based on it 19:48:37 Per suggestions from last week’s chat room discussion, we have refactored iLO virtual media deploy driver into iLO Virtual Media iSCSI Deploy Driver https://review.openstack.org/#/c/97744/ and iLO-IPA Deploy Driver https://review.openstack.org/#/c/97744/. We would like to target these 2 specs for Juno. Please review these specs as soon as you can. Thanks! 19:49:10 wanyen, +1, thanks! 19:49:23 ok 10 min beep... let's go for the open discussion or we won't have much time 19:49:28 #topic open discussion 19:49:32 we got +2 from Deva for our original iLO deploy driver spec but lifeless suggested that we refactored it so please review them 19:49:49 so... auth for IPA. we think passing tokens around is bad, and deployers should secure the network. client certs are a thing, could help (aweeks has been poking at this). thoughts? 19:49:51 so jroll, about the authentication... is keystone trusts something that we should have in Ironic to help with that? 19:50:08 I don't know what keystone trusts are... but I'm betting they require v3 api? 19:50:16 we don't have manpower to land keystone trusts, as to me... 19:50:20 yeah afaiui it does require v3 19:50:24 lucasagomes: I have a different, possibly more radical suggestion 19:50:31 yeah, that's a non-starter for us right now :| 19:50:34 lucasagomes: why does Ironic need to remain a fully-open admin api? 19:50:54 what do you mean by "fully-open"? 19:50:57 If we setup users with proper roles in keystone, and Ironic honored it, we could pass a token to the Agent that enabled it to do *only what it needed to do* 19:51:08 whereas right now if Ironic validates a token at all, it's pretty close to wide-open 19:51:12 JayF: I don't think it does. we just need to do some work on policies 19:51:29 JayF: https://github.com/openstack/oslo-incubator/blob/master/openstack/common/policy.py is a thing and used today in ironic 19:51:29 So lets make a user that by-policy and only do what IPA needs to do 19:51:30 yeah, it was a design decision at the beggining but now I start to see that we def need some policies implemented 19:51:47 JayF: in conjunction with https://github.com/openstack/ironic/blob/master/etc/ironic/policy.json 19:51:47 do we have time for it in J? 19:51:48 and you reduce risk massively vs shipping a 'full admin' token 19:52:02 dtantsur: maybe? doubtful 19:52:06 just a reminder: 3 days before specs deadline 19:52:11 dtantsur, hmm that's hard to say really 19:52:15 I'd suggest this: we land agent spec as-is without auth 19:52:25 and that we look at ways to implement auth as an addon afterwards 19:52:26 without auth *on the heartbeat urls* 19:52:31 yes 19:52:33 JayF ++ 19:52:41 jroll, JayF +1 19:52:44 yeah baby steps 19:52:48 the auth is a broad problem 19:52:50 see also https://review.openstack.org/#/c/101020/10/ironic/api/config.py 19:52:52 :) 19:52:53 the PXE approach is far from perfect 19:53:03 I mean, I have hundreds of machines running IPA, just authenticated by the network being secret 19:54:15 I like this plan 19:54:19 thanks y'all 19:54:25 Before time ran out, wanted to ask: are there folks here, who also want any kind of (in-band) hardware discovery landed in J? 19:54:35 just figuring out, who can help us 19:55:20 because we're trying to ship at least simple PoC discovery, that works 19:55:34 I think it's a great feature to have, but I think that there are higher priorities 19:55:41 #link https://review.openstack.org/107344 19:55:51 this is the spec that dtantsur is working on ^ 19:56:12 right now, I can online an entire rack with one bash script 19:56:23 which also can do multiple racks 19:56:37 so it honestly wouldn't save *me* much time... but that's just me 19:56:51 that's a priority for us, so I'll be anyway busy with it... 19:57:03 right 19:57:07 lucasagomes: what the priority for https://review.openstack.org/#/c/100951 Discover at node-create ad node-update 19:57:12 I just think there are more important features to land 19:57:14 yeah I think this is the feature number #1 that the tuskar folks wants implemented in Ironic 19:57:15 uefi etc 19:57:42 but yeah the priority is different for each of us 19:58:22 Nisha, thanks! yeah that's another way to do it 19:58:35 Nisha, I would like to see it landed, though it's not a high priority for me 19:58:36 and we want to have something similar for the drac driver 19:58:43 I think Nisha's spec is meant to be on top of dtantsur's 19:58:50 jroll, no longer 19:58:53 ohhh 19:58:56 it was before, now it's changed 19:59:02 * jroll shuts up :) 19:59:03 things have changed a lot due to our previous talk 19:59:19 that's my problem with the spec procedure: we have real talk mostly on IRC 19:59:27 heh 19:59:29 still more-or-less silent in comments :) 19:59:33 one minute or so left. 19:59:42 20 seconds :P 19:59:43 alright folks, I don't think we have more time 19:59:47 so 19:59:48 to the channel! 19:59:49 thanks 19:59:50 thanks everyone, great meeting 19:59:55 as usual 19:59:56 thanks for running it lucasagomes :) 19:59:59 and I hope to see some of you guys next week!! 20:00:06 don't forget to #endmeeting :P 20:00:08 #endmeeting