15:00:44 #startmeeting ironic 15:00:45 Meeting started Mon Nov 18 15:00:44 2019 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:46 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:49 The meeting name has been set to 'ironic' 15:00:53 etingof: me toooooooo 15:00:55 o/ 15:00:55 morning TheJulia (again) o/ 15:00:57 o/ 15:01:00 etingof: and I think I need to be on a boat in like an hour 15:01:01 o/ 15:01:03 o/ 15:01:05 \o 15:01:09 o/ 15:01:16 Good morning everyone! 15:01:35 o/ 15:01:37 Our agenda this week has a few items, and those can be found on the wiki. 15:01:39 * arne_wiebalck will need to leave half-way through the meeting 15:01:41 #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting 15:01:43 o/ 15:01:52 bfournie: o/ 15:01:57 #topic Announcements / Reminder 15:02:08 o/ 15:02:12 I have to items to announce 15:02:39 o/ 15:02:45 The first is that I sent out a summary of the ironic related sessions and the PTG to the mailing list. 15:02:47 #link http://lists.openstack.org/pipermail/openstack-discuss/2019-November/010789.html 15:02:56 TheJulia++ 15:02:59 The second, I'm now in San Diego for Kubecon 15:03:03 Thanks a lot TheJulia ! 15:03:12 TheJulia: the summary is really cool :) 15:03:15 So it may take me a little while to reply this week. 15:03:20 Thanks everyone! 15:03:20 +1, thanks 15:03:36 dtantsur: did we have any action items from last week? 15:03:56 looks like there were none 15:04:09 o/ 15:04:12 \o 15:04:15 TheJulia: the naval yard in San Diego is awesome, if you have time :) 15:04:30 TheJulia: not that I can think of 15:04:52 Normally I would say we proceed to looking at subteam status reports, but I've proposed a new priority list for this cycle 15:05:16 #link https://review.opendev.org/#/c/694704/ 15:05:16 patch 694704 - ironic-specs - Ussuri project priorities - 2 patch sets 15:06:21 Any objection to going directly to priorities for the week? 15:06:59 none from me 15:07:00 * TheJulia suspects she could find crickets 15:07:29 Okay then! 15:07:43 #topic Priorities for the week 15:08:04 #link https://etherpad.openstack.org/p/IronicWhiteBoard 15:08:06 Line 171 15:09:06 mgoddard: you seem to be going back and forth on https://review.opendev.org/#/c/672252/, is a -1 required to signal revision required? 15:09:07 patch 672252 - ironic-specs - Add support for node reconfigure - 5 patch sets 15:09:07 * iurygregory is late o/ 15:09:11 o/ iurygregory 15:09:46 TheJulia: yeah, I'll add one 15:10:52 While I'm deleting/updating entires there, does anyone have patches to propose? 15:11:57 mine are there 15:12:11 arne_wiebalck: I'm going to move xXraphXx's IPA patches down to the bottom of the list since they haven't been rebased/updated yet :( 15:12:15 I'm working on the patches to drop the py27 jobs going to push during the week but no hurry =) 15:12:19 tzumainn: are you planning on the remaining ownership patches this week? 15:12:34 I could take the owned allocations, it's a fair chunk of work 15:12:35 dtantsur, hi! I'm planning on working on the one for ports, followed by allocations 15:12:36 TheJulia: sure ... it's still on my list :) 15:12:44 mine parches are there as well 15:12:45 tzumainn: ah, lemme know if you want to do allocations yourself 15:12:52 dtantsur, I wouldn't mind taking a shot! 15:12:58 unless you are somehow supremely bored 15:12:59 wonderful! I can help with reviews/guidance. 15:13:04 okay, sounds good! 15:13:04 I'm never bored :) 15:13:09 hahaha, I figured :) 15:13:17 (I am sometimes bored, but from procrastination, not from lack of tasks) 15:14:01 awesome https://review.opendev.org/#/c/689551/ merged 15:14:02 patch 689551 - ironic - Allow node owners to administer nodes (MERGED) - 7 patch sets 15:14:12 I feel like there should be a release soon! 15:14:29 that can be arranged 15:14:40 which project(s)? 15:14:42 lets wait, I need to work something out with the stable team first :\ 15:14:53 ok 15:14:56 I ran out of time last week :( 15:15:56 Well if there is nothing else to add to the list, I think it looks fairly good for this week and we can proceed to Discussion 15:15:58 tzumainn: JFYI another thing I'm pondering (well, not only me, it's been a recurring topic) is a single-endpoint deployment API (rather than a combination of node update + VIF + node provision state). 15:16:15 I can try putting together a spec for that this week 15:16:21 TheJulia: ^^ 15:16:47 dtantsur, ah, that'd definitely be of interest! 15:17:01 I'd personally prefer to see us focus on some performance issues in our API, but... if there is interest 15:17:35 TheJulia: which exactly? I think I may have some time (wow) this cycle. 15:17:52 but I also think that the ownership work is semi-useful without this new deployment API 15:17:55 TheJulia: Re: Baremetal SIG, line 297, I believe you've found the OSF POCs. Right? 15:18:11 because we don't want non-privileged users to have node update rights 15:19:02 dtantsur: this is true, if we can somehow improve the performance to get a list of nodes and then get nodes, it would be a good thing... HOW exactly, needs to be determined, but I've had a few people grumble about how long it takes to list out machines in large deployments :\ 15:19:36 I'm not sure there is any magic to squeeze there other than telling people to use 'fields' extensively 15:20:03 Maybe we just need to be mindful 15:20:08 rpioso: danny@openstack.org and wes@openstack.org 15:20:10 but, well, I can put it on my TODO list. maybe I can figure out something using x000 of fake nodes 15:20:25 TheJulia: ty 15:20:58 dtantsur: yeah, one item in talking with nova at the ptg was "hey, if we can load 1000 nodes into ironic, and then see how long it takes for those to get into resource tracker, we can begin to somewhat measure this 15:21:28 trivial to do if somebody tells me how to reliably define "gets into resource tracker" 15:21:38 Wow my lag is getting really bad from irccloud 15:21:39 #topic Discussion 15:22:07 dtantsur: well, placement actually 15:22:17 TheJulia: dtantsur: I think the nova team decided that it would be much more sensible to use conductor groups. 15:22:26 Two topics for this week 15:22:29 To address scaling issues. 15:22:34 arne_wiebalck: I've not followed that thread, and that does not surprise me 15:22:55 TheJulia: We will give that a try in our deployment. 15:22:55 I can likely take an action item to try and describe what that looks like, at least the way it was intended to be used 15:22:56 yeah, I've seen it. also our effort will be limited when it comes to fixing nova (as opposed to our API) 15:23:21 TheJulia: I sense another reference architecture! (which we'll be too busy to finish, sigh) 15:23:22 The RT code seems to be somewhat complicated. 15:23:33 "somewhat" :) 15:23:58 :D 15:24:00 And there are worries to break things for the non-ironic use case (which is arguably the main use case). 15:24:07 jroll: ;) 15:24:09 dtantsur: re "gets into resource tracker", timing this method is probably the best starting point: https://opendev.org/openstack/nova/src/branch/master/nova/compute/manager.py#L9185 15:24:15 dtantsur: well, I would agree with a reference architecture but the overall code review process applyign to docs kind of kills that from working because eveyrone is differently opinionated 15:24:18 anyway, discussion items 15:24:23 that's the periodic task 15:24:32 oh sorry, I thought this was open discussion :) 15:24:55 yeah, let's get back to it later :) 15:25:15 #info There is an open question of if we would like to have a mid-cycle. We've been offered space in Central Europe for two days. Feb 25-26 maybe? 15:25:40 Europe \o/ 15:26:00 I don't expect everyone to have a "yes" answer, but I think it would be a good thing to organize since the next official gathering with be in Vancouver in June 15:26:26 Please consult with your management, look at prices, and maybe we can revisit the question next week? 15:26:30 * iurygregory CERN? =P 15:26:32 I'm tentatively yes (assuming budget approval) 15:26:42 iurygregory: yes 15:26:45 TheJulia: do you have a rough idea which Central Europe? prices can be VERY different 15:27:07 CERN would be awesome and very hard budget-wise at the same time 15:27:16 GVA 15:27:20 must be a geographical center of Europe... 15:27:53 oh boy, I need to find a way to justify this (I won't be able to, sigh) 15:28:11 * TheJulia is sensing ?maybe? 15:28:19 * dtantsur would probably go to GVA even on his own budget 15:28:27 jroll, we are all in the same boat =) 15:28:39 I did look, it is only an 8 hour train ride from CDG 15:28:42 * iurygregory would do the same thing XD 15:28:43 iurygregory: well, he also needs an intercontinental flight 15:28:43 I would definitely go no matter what :D 15:28:46 iurygregory: not all, I'm sure arne can justify it :D 15:28:47 what about virtual mid-cycle? 15:29:12 virtual mid-cycles are cheap, we can have both 15:29:32 ^ ++ 15:29:36 We're happy to host a mid-cycle here at CERN. Most important thing for now would be to settle on dates. 15:29:50 As I'd need to book the room. 15:30:02 (and organize tours ;-) 15:30:18 end of february seems ok, it's in the middle between now and the next summit 15:30:21 (to understand how our software is used ;)) 15:30:22 but we have the dates, we just don't know about how many can make it 15:30:33 Anyway, we should kind of move on to the next topic 15:30:50 Any FOSS conferences around that time? 15:31:11 FOSDEM in early Feb 15:31:28 Somewhere somebody has to have a central conference calendar 15:31:30 Anyway 15:31:42 Next discussion topic 15:32:06 dtantsur: Thx 15:32:20 It has been raised to our attention that the nova virt driver in ironic lacks documentation. 15:32:22 #link https://bugs.launchpad.net/nova/+bug/1852446 15:32:23 Launchpad bug 1852446 in OpenStack Compute (nova) "Hypervisors in nova - no subpage details for ironic" [Undecided,New] 15:33:06 While surely some might just link into ironic's documentation, is anyone interested in maybe taking a stab at adding some documentation to nova? 15:33:47 maybe we should link it to this page until we do write some? https://docs.openstack.org/ironic/latest/install/configure-compute.html 15:34:09 Seems a good first step 15:34:19 I'm interested, just would like to know if this has to be done this week (next week I have time to start working on this) 15:34:26 We don't want to duplicate everything, the audience is different there too 15:34:35 ++ 15:34:36 also it depends on how much details we want there of course 15:34:47 iurygregory: does not need to be done this week 15:35:26 and maybe I will bother people with questions XD 15:35:26 iurygregory: you going to take a shot at putting something together? 15:35:31 Sounds good :) 15:35:36 TheJulia, sure =) 15:36:58 Moving on! 15:37:01 #topic Baremetal SIG 15:37:22 First item, the white paper! 15:38:06 arne_wiebalck: The bare metal logo program participants submitted case studies topic came up in a discusison with foundation folks last week. 15:38:53 TheJulia: Can we use them as case study input? 15:39:08 arne_wiebalck: They are going to iterate through participants and likely send introduction emails for the SIG that you should be getting with a reminder that they should submit their content to the whitepaper for case study input 15:39:29 TheJulia: Nice! 15:39:32 So at least a forward step there! 15:39:43 Remember, anyone can help work on the whitepaper! 15:40:01 We'll need some "glue text" with the larger picture. 15:40:03 arne_wiebalck: I guess that kind of covers the first two items :) 15:40:04 TheJulia: \o/ 15:40:14 TheJulia: It does, thanks! 15:40:22 arne_wiebalck: I'm already happy to help with glue text 15:40:33 TheJulia: Awesome, thanks! 15:40:56 are there any tasks requiring average English skills? :) 15:41:15 dtantsur: always! :) 15:41:31 For the ops meetup, one of the things we said at the PTG was to show some presence at these events ... anyone planning to attend? 15:41:32 One other item for the SIG that Arne and I both feel would be good, is that at least one ironic team contributor coul dmake it to the Operators Meetup in London on January 7th-8th 15:41:49 London.. likely nope 15:42:14 none of those needs a visa can make it 15:42:19 * those who need 15:42:21 it's a couple hours train from Paris :P 15:42:25 forbidden lands 15:42:27 One issue I had with these meetings in the past was that the agenda was built very late. 15:42:34 * dtantsur appoints rpittau 15:42:38 lol 15:42:56 So, it is somewhat hard to know what to expect (and justify in time). 15:42:59 I might be able to attend 15:43:01 for the record, I enjoyed the ops meetup in Berlin. a nice opportunity to hear what smaller operators are thinking 15:43:41 dtantsur: I was in Milan. While it is nice to meet other ops, it is good when project cores are around. 15:43:48 I might be also able to attend :) 15:43:55 Also, they tend to be separate industry verticals. The NYC meetup I attended a few months back were mostly financials and their only ask was "help them figure out a migration path from..." 15:44:04 cobbler I think 15:44:23 oh gosh 15:44:49 ouch =X 15:44:50 * TheJulia senses rpittau is backing away from the computer now 15:44:51 :) 15:44:58 kind of :D 15:45:06 migration from cobbler came up in Berlin too 15:45:18 No RFE's listed to review, so lets go to Open Discussion 15:45:26 #topic Open Discussion 15:45:39 * arne_wiebalck has to leave, sorry o/ 15:45:41 dtantsur: interesting... perhaps a blog post or something is... required 15:45:44 o/ arne_wiebalck 15:45:46 bye arne_wiebalck o/ 15:45:59 arne_wiebalck: \o 15:46:08 official cobbler-ironic migration guide? 15:46:22 * TheJulia backs away from the computer a little 15:46:32 maybe? 15:46:45 We have so many things that need work though :( 15:47:49 maybe simply because they are not in the help wanted :P 15:49:18 Hi TheJulia dtantsur rpittau please add this patch in your review list: https://review.opendev.org/#/c/671038/ 15:49:18 patch 671038 - ironic - DRAC: Drives conversion from raid to jbod - 21 patch sets 15:49:50 it's already on vendor priorities 15:50:05 kaifeng_: it is a possibility, or enough of the major problems have been solved 15:50:45 kaifeng_: also help wanted is a whole nightmare at the openstack level 15:51:45 i am feeling we have many information scattered around etherpads but lacking a full list 15:52:40 I was wondering if we should convert the etherpad with the CI failures to a more persistent doc 15:53:03 for ironic, it feels more like we don't ask for help early enough, or that sometimes we ask and people volunteer, and then those people get pulled downstream for weeks or months :( 15:53:07 I mean the issues resolved etherpad 15:53:12 Which is not their fault :( 15:53:40 rpittau: perhaps? 15:54:16 Some of those failures have occassionally appeared for operators, so it would be visibility I guess 15:54:27 and google indexable 15:54:32 maybe a wiki section ? 15:54:32 well 15:54:37 $favoritesearchengineindexable 15:54:56 Lots of people in the grass roots linux community seem to be trying to figure out ways to de-google 15:55:41 Wiki might work 15:56:35 is there any requirement for the white paper cases? 15:58:19 kaifeng_: no requirement other than a willingness to share perspective 15:59:42 Anyway, our meeting time is almost over 15:59:44 Thanks everyone 15:59:47 thank you! 15:59:48 Dmitry Tantsur proposed openstack/ironic-python-agent-builder master: Generate checksums for DIB images https://review.opendev.org/694800 15:59:50 ty 16:00:28 thanks! 16:00:44 tx 16:01:13 #endmeeting