15:00:21 #startmeeting ironic 15:00:22 Meeting started Mon May 3 15:00:21 2021 UTC and is due to finish in 60 minutes. The chair is rpittau. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:23 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:25 The meeting name has been set to 'ironic' 15:00:30 o/ 15:00:50 o/ 15:00:53 o/ 15:00:54 o/ 15:00:59 o/ 15:01:06 Hello everybody and welcome to yet another ironic weekly meeting! 15:01:18 I'm honored to be your host today :) 15:01:23 o/ 15:01:42 you can find our agenda at the usual place 15:01:53 #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting 15:02:21 Let's start with some announcements 15:02:27 #topic Announcements / Reminders 15:02:44 #info ironicbaremetal.org now has blogs! 15:02:44 #link https://ironicbaremetal.org/ 15:03:00 thanks TheJulia for working on that 15:03:10 and also... 15:03:11 \o 15:03:11 Always happy to help :) 15:03:13 #info To aid the effort to add more content to the website TheJulia created and etherpad to collect links and content 15:03:13 #link https://etherpad.opendev.org/p/ironic-media 15:03:20 yay! thx TheJulia!! 15:03:45 any idea, link, thought, drop that there :) 15:04:13 ++ 15:04:35 on a more sad note 15:04:40 #info mgoddard is stepping down as release liaison. Any volunteers should inquire during Open Discussion. 15:04:40 thanks mgoddard for your contribution :) 15:05:09 If iurygregory would like to do this, I'd propose him. He's been doing really good job recently. 15:05:15 and thank you mgoddard :) 15:05:24 ++ 15:05:52 yup =) I would be happy to do 15:06:57 any volunteers let's defer to Open Discussion 15:07:00 and ofc I have some issues with irccloud :/ 15:07:12 anyhow 15:07:17 me too actually. Backing DB appears to be down 15:07:22 yeah 15:07:27 :( 15:07:34 Does anyone have anything else to announce or remind us of? 15:07:52 I'm not expecting to be available next week and I'll be away on Friday 15:08:08 I'll likely be away next week on Thu (holiday) and Fri 15:08:24 I'm also very limitedly available this week because of KubeCon EU 15:08:39 same for Kubecon and I'll be out Thu-Fri next week 15:08:56 I will also be off Thu/Fri next week 15:10:29 seems no more announcements, we can move on 15:10:36 #info Review action items from previous meeting 15:10:36 I wanted to mention an RFE for review. Is now a good time? 15:10:59 Oh there's a section for that, bit early. Sorry! 15:11:04 MahnoorAsghar: later we have an RFE review section, thanks :) 15:11:13 Perfect! 15:11:36 it seems we didn't have any action items from last meeting 15:11:52 so let's pass to the next part 15:12:02 #topic Review Subteam status reports 15:12:52 I think we're still defining priorities for Xena, so no status yet 15:12:58 I was about to say something similar 15:13:32 we do have a patch to review with the Xena themes 15:14:10 please have a look at it when you have a minute :) https://review.opendev.org/c/openstack/ironic-specs/+/784143 15:14:40 alright, moving forward 15:14:48 #topic Deciding on priorities for the coming week 15:14:54 #link https://review.opendev.org/q/status:open+hashtag:ironic-week-prio 15:15:11 we don't have a lot of priorities in the list 15:15:37 I can't really think of anything else that really needs to be added right now 15:15:54 https://review.opendev.org/q/topic:%22iscsi-gate%22+(status:open%20OR%20status:merged) 15:16:01 I'm going to add one as well https://review.opendev.org/c/openstack/ironic/+/788734 15:16:04 oh yes, that is likely a good set of things to add 15:16:09 and the growing list of https://review.opendev.org/q/topic:%22no-iscsi%22+(status:open%20OR%20status:merged) 15:16:49 tagging 15:16:53 feel free to add ironic-week-prio tag to all of the patches there not mergeed 15:16:56 thanks TheJulia :) 15:16:59 burn-in is also ready to get more reviews I think: https://review.opendev.org/q/topic:%22story_2007523%22 15:17:03 I'd also appreciate https://review.opendev.org/c/openstack/ironic/+/788718 15:17:09 right, thanks arne_wiebalck 15:18:20 it looks like we have a good maount of priorities for the week now! :) 15:18:24 heh 15:18:44 \o/ 15:19:09 onward! 15:19:21 #topic Discussion 15:19:39 I have a topic for discussion, more a yes/no question 15:19:48 got a reply on ipa-builder stable/branches from the ML http://lists.openstack.org/pipermail/openstack-discuss/2021-April/022032.html, should we move on then? 15:21:05 hasn't Herve also replied ? 15:21:13 he did 15:21:15 don't see this on the page 15:21:17 yeah, the reply does not appear there ? 15:21:41 oh 15:21:45 click next message =) 15:21:46 it cut by month 15:21:48 http://lists.openstack.org/pipermail/openstack-discuss/2021-May/022170.html 15:22:29 I think we should just go ahead and do it 15:22:46 lazy consensus through a lack of objections 15:22:59 sounds good to me 15:22:59 any more opinions ? 15:23:22 We should probably say it is unlikely we break anyone's setup. 15:23:31 ++ 15:23:33 I think this is Herve's question. 15:23:44 I think arne_wiebalck should say that :D 15:23:44 arne_wiebalck: I agree, I had in mind to reply to Herve anyway 15:23:53 It is more about, being able to fix issues 15:25:16 he probably thinks we split the repos since someone wanted to use only IPAB without the IPA 15:25:48 anyway, I am all for moving forward as suggested by rpittau 15:26:22 ok, seems we're good 15:26:38 #topic Baremetal SIG 15:26:59 we're going to revisit some proposals from last meeting 15:27:18 Yeah, I would like to settle on the agenda for the next meetings. 15:27:24 I kept the list since last week because I think we failed to record owners. 15:27:30 looks like we have topics for the next 2 meetings 15:27:52 I've been thinking a scaling ironic blog post might be best, to at least start out and then maybe the sig session could be more as a Q&A 15:27:52 We have dtantsur for IPAB and TheJulia for Bifrost. 15:28:17 dtantsur: June and TheJulia: July if that is ok? 15:28:24 should be good 15:28:43 we still have a slot for next week 15:28:53 any volunteers, suggestions? 15:29:29 (otherwise I can do a "Why Ironic at CERN?" filler) 15:29:48 seems like a nice filler :) 15:30:25 TheJulia: we have done a scaling post some time ago: https://techblog.web.cern.ch/techblog/post/conductor-groups/ 15:31:04 anyone else who would like the slot? 15:31:05 I was thinking more a high level concepts/modeling 15:31:23 sure, I was thinking as further input 15:32:01 Ok, I think that is it then for the SIG this week :) 15:32:11 cool! 15:32:22 then let's move on 15:32:29 #topic RFE review 15:32:43 @MahnoorAsghar I think you had something ? 15:32:55 Yes 15:32:55 arne_wiebalck: hmm, so maybe we have an opportunity for a scaling 100 series "class" and then like s scaling 400 series class 15:33:15 TheJulia: :-D 15:33:34 I wanted to ask for reviews on this RFE: https://storyboard.openstack.org/#!/story/2008866 15:33:46 TheJulia: I think this is a good idea, actually! 15:34:58 MahnoorAsghar: I wonder if that'd be a better fit as a node vendor passthru... we don't typically put hardware information at the top level of nodes 15:35:19 I'm also kind of thinking vendor-passthrough 15:35:24 MahnoorAsghar: so that's specific for Dell ? 15:35:28 Alright 15:35:52 We were thinking of making it extendable for other vendors 15:35:53 Bonus: vendor passthru will /probably/ be easier to implement than a real API endpoint 15:35:54 It always puzzled me how our consumers determine the bare metal hardware RAID controller and physical disk IDs. 15:35:59 I guess my couple concerns are 1) this data only seems to be the oob data. In-band data is more detailed and may have additional things based upon inspection, and the inspection dat should be visible. 15:36:25 MahnoorAsghar: sometimes we implement a feature for the first vendor as passthru, and then as more vendors implement it an API emerges and they converge :) 15:36:27 2) other vendors adopting this is a whole question since some just don't care or really *need* to care about some of the fields. 15:36:38 we don't always have to start at the point of a vendor-agnostic convergent api 15:37:00 I'm really hesitant to say "lets do this as a full api" without another vendor being behind it and committing to the same 15:37:12 yeah, that's why I asked 15:37:37 TheJulia: same 15:37:41 although the idea is good 15:38:03 on a related note, someone on the meetup last week mentioned remote erasure via iDRAC for some RAID controllers. would be cool to support it. 15:38:05 JayF, TheJulia: I expect it would be a non-starter to Metal3 if the solution is not vendor agnostic. 15:38:26 rpioso: I don't think this claim is uniformly true 15:38:37 although it will surely be harder to provide API for 15:39:52 anyway, I agree with TheJulia. if no other vendors back that, it will be de facto vendor passthru 15:39:58 dtantur: vendor-passthru makes our software harder to use and I thought we had been trying to get away from that. 15:40:11 when possible 15:40:13 dtantsur: ^^^ 15:40:23 this is not proved possible yet 15:40:33 I mean, we need a viable path to having this implemented for a lot of hardware 15:40:34 can we at least fetch RAID devices via pure Redfish? 15:40:39 only a dell implementation of an API is not really good :( 15:40:40 And some things are always going to be vendor passthrough or vendor specific, and that is okay 15:40:57 dtantsur: Yep, agree when possible. Let's see if it is :-) 15:41:40 Dmitry Tantsur proposed openstack/ironic master: Update refarch with information about image_download_source https://review.opendev.org/c/openstack/ironic/+/789377 15:42:01 let's maybe leave some notes on the RFE itself ? 15:42:02 a possible approach is to fetch it during inspection and put in properties (?) 15:42:16 rpittau: sounds good :D 15:42:20 dtantsur: Yes, the IDs are available via Redfish. The Redfish raid interface works :-) 15:42:24 dtantsur: I was thinking the same 15:42:31 dtantsur: That could work as well 15:42:33 dtantsur: I think that's an OK approach, but if we were going to do that, we should probably think about it for more than jsut disks? IDK :/ 15:42:54 Hell, you could even exclude inspector -- just have IPA list disks on boot and store that 15:42:57 ala clean / deploy steps 15:43:11 (We never did implement an API for "boot a ramdisk and tell me possible clean steaps", did we?) 15:43:29 I have no idea how easy is to list controllers/disks in the iDRAC notion from inside the machine 15:43:31 JayF: that could also work 15:43:43 this "Disk.Bay.0:Enclosure.Internal.0-1:RAID.Intergrated.1-1" stuff is not something we receive now IIRC 15:44:26 In addition to those IPA suggestions, a purely out-of-band approach would be useful. No ramdisk boot would be required. 15:44:45 dtantsur: How about local host Redfish or WS-Man? 15:45:02 dtantsur: It is not available in the inspection, but is available in the ironic conductor logs 15:45:33 rpioso: Redfish host interface is something we should eventually research. but credentials are a huge issue. 15:46:31 dtantsur: Understood. IPMI? I'm unfamiliar with its prospective knowledge/reporting of those IDs. 15:47:04 IPMI would be a hard no, I think 15:47:11 even if it's possible, it'd be very vendor/config dependant 15:47:14 yeah 15:47:24 Again, a fully OOB approach would be useful. 15:47:39 Sounds like Redfish at some point in the future. 15:47:50 I feel like we should soon seriously rethink our notion of inspection. 15:47:56 dtantsur++++ 15:48:11 This is basically going down the path of "make node inspection info available over API" 15:48:14 which is not a bad idea 15:48:21 yep 15:48:37 but I'm loathe to include all that scope in MahnoorAsghar's idea, because disks, and how they interact with root hints, is an actual UX problem now 15:48:40 and probably get rid of in-band and OOB inspection separation (which we've already started fwiw) 15:48:42 I'm going to add a triaged tag to the RFE, but please remember to add your notes on the discussion there :) 15:49:07 any non-trivial API addition requires a spec, I think 15:49:23 JayF: This is not directly related to root device hints. Rather, just RAID configuration. 15:49:30 hmm 15:50:26 we have 10 minutes left, maybe continue the discussion after the meeting? 15:50:27 rpittau: Shall I add the notes from this discussion there? 15:50:37 MahnoorAsghar: sure! 15:50:54 rpittau: lets continue it after the meeting 15:51:02 ok 15:51:16 let's move on then 15:51:21 #topic Open Discussion 15:52:03 going back to the "release liaison" topic, do we have other volunteers? 15:53:22 it doesn't look so 15:53:26 It is not really that much work and if you do any release work at all, it is really just being an alternate point of conteact 15:53:34 and you get the authority to ship $stuff ;) 15:53:50 * dtantsur assumed rpittau was a liaison, but apparently no 15:53:51 oh the +1 PTL =O 15:54:08 On another note, we have even more blogs now on https://ironicbaremetal.org/blog/ 15:54:17 edits can be made through github :) 15:54:24 WOOT 15:54:27 :D 15:54:27 mmm I'm not, and I think I will volunteer 15:54:57 #link 15:55:02 whoops 15:55:12 #link https://ironicbaremetal.org/blog/ 15:55:28 anything we need to do to volunteer? like we need to send a patch to a repo or something? 15:55:40 dtantsur, TheJulia, please consider me also for release liaison 15:55:50 * dtantsur +1 15:56:02 +1 15:56:05 rpittau, iurygregory, another responsibility is to react to notifications from the release team 15:56:16 both emails and gerrit (you'll be added to any reviews touching ironic deliverables) 15:56:29 ack =) 15:56:32 iurygregory: who the current liason is for projects is tracked on the release wiki 15:56:36 sounds good :) 15:56:46 mm, no, not wiki 15:56:56 I thought it was in the wiki, is it now a repo? 15:57:15 https://opendev.org/openstack/releases/src/branch/master/data/release_liaisons.yaml 15:57:19 (It has been ages since I've updated it) 15:57:24 okay, then I guess it moved \o/ 15:57:31 nice! 15:58:12 alright 3 minutes left, do we have other topics to discuss ? 15:58:29 #topic Who is going to run the next meeting? 15:58:34 \o/ and now blog links work correctly 15:58:46 so the PTL needs to send a patch or should just me and/or rpittau send ? =) 15:58:46 Any volunteers? 15:58:52 I can run =) 15:59:04 #info iurygregory will run next meeting! Thanks! :) 15:59:07 iurygregory: rpittau: either of you can submit a patch and the a PTL can +1 it 15:59:09 TheJulia: you can pick other links from my blog, especially tutorials alike 15:59:33 dtantsur: ack 15:59:37 TheJulia, ack 15:59:54 alright, we're at the top of the hour, let's continue the discussions out of the meeting :) 16:00:05 thanks everyone! 16:00:08 #endmeeting