15:00:18 #startmeeting ironic 15:00:19 Meeting started Mon Jun 8 15:00:18 2020 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:21 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:21 o/ 15:00:23 The meeting name has been set to 'ironic' 15:00:23 o/ 15:00:25 o/ 15:00:25 o/ 15:00:29 1 point goes to iurygregory for being ready to start the meeting :) 15:00:31 o/ 15:00:31 \o 15:00:32 o/ 15:00:34 o/ 15:00:35 iurygregory++ :D 15:00:42 o/ 15:00:43 =) 15:00:45 Good morning, time for our ironic meeting! 15:01:00 Our agenda is on the wiki, you can find it at the next link 15:01:02 #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting 15:01:12 o/ 15:01:16 #topic Announcements / Reminders 15:01:36 I have two items on the agenda for annoucements/reminders 15:02:01 The first is the priorities change set in ironic-specs. I'd like to get this wrapped up this week, so reviews would be super helpful. 15:02:09 #link https://review.opendev.org/#/c/720100/ 15:02:10 patch 720100 - ironic-specs - WIP - Victoria Cycle Priorit(y|ies) - 6 patch sets 15:02:16 The other is PTG notes, which can be found at... 15:02:23 #link https://meetpad.opendev.org/etherpad/p/ironic-ptg 15:02:35 Does anyone have anything else to announce or remind us of? 15:02:41 o/ 15:02:45 \o 15:03:04 we need to do some math 15:03:09 and figure out when we're doing the next release 15:03:24 dtantsur: on the priorities document 15:03:26 already 15:03:27 :) 15:03:28 wow 15:03:40 okay, I've publicly admitted that I haven't read it in a whole. oops. 15:03:51 lol 15:03:55 * dtantsur will correct that 15:04:02 reviews ++ 15:04:18 wrt priorities. it was discussed at ptg, about 'lazy consensus' and some time period. Did we decide? If not, shall we decide now? 1 week after PTG or something like that? 15:04:37 in API SIG we did it this way: 15:04:48 rloo: I was thinking by the next meeting at worst, tbh 15:04:57 when the core team (2-4 ppl) is pleased with a document, we announce it and the deadline 15:05:06 after the deadline the change is merged if no objections 15:05:15 Regarding our next topic, we have no action items from the prior meeting 15:05:21 for us it could be something like once 3x +2 are there, merge in 1 week 15:05:23 dtantsur: That seems reasonable 15:05:35 dtantsur: is taht the entire core team or majority or ? 15:05:58 rloo: API SIG has had 4 cores top during its whole history 15:06:09 only 2 nowdays 15:06:22 so it is more of an external feedback mechanism in then 15:06:26 which also makes sense 15:06:28 dtantsur: so 3x +2 is 3 core members regardless of total number of cores. 15:06:41 for us - yes 15:06:44 our existing policy is agreement from all cores. 15:06:52 right, this is what lazy consensus is about 15:07:04 once 3x+2 are there, we invoke it (with an announce here/on the ML) 15:07:11 if no objections are recorded, say, in a week, merge it 15:07:11 Perhaps we should discuss it during the discussion topic 15:07:15 yeah, sorry 15:07:15 Mark Goddard proposed openstack/bifrost master: Fix bifrost_inventory.py with node nodes enrolled https://review.opendev.org/734115 15:07:17 if we will need to wait for all cores we will have problems =) 15:07:18 no worries! 15:07:30 iurygregory: yeah... That is what I'm seeking to avoid 15:07:37 Mark Goddard proposed openstack/bifrost master: Fix bifrost_inventory.py with no nodes enrolled https://review.opendev.org/734115 15:07:42 agree 15:08:09 So crazy thought, do we even need to visit subteam status reports this week? 15:08:15 since the priority document is up in review? 15:08:25 I've put some stuff about deploy steps 15:08:34 no need to spend 10 mins on it though :) 15:08:57 Then lets jump directly to priorities for the coming week. The review list is huge 15:09:26 to put it mildly 15:09:31 yeah.... 15:09:37 We also haven't held a meeting in a little while 15:10:00 #topic Deciding on priorities for the coming week 15:10:10 #link https://etherpad.opendev.org/p/IronicWhiteBoard 15:10:34 Starting at line 130, we have a number of items that merged... and we have a huge number of items that could use reviews and or feedback. 15:10:41 So any reviews this week would be super appreciated by contributors 15:10:49 * TheJulia removes the merged items from the list 15:10:54 remove deployment API for now, it's not finished 15:10:59 * let's 15:11:01 ack 15:11:58 wow, that was a lot 15:12:16 Now, new items for this week starting at line 183. If there any objections, now would be good to raise them 15:13:05 we need to move "for the next meeting" to the bottom, right? 15:13:21 basically, but I'm cleaning it up now since there may be duplicates 15:15:43 A lot of the items on that list are fairly small to medium patches 15:15:53 so some time reviewing would be good this week. 15:16:36 Also a number of those items are documentation improvements 15:16:55 Questions/objections? 15:16:58 Things to add? 15:17:20 I hope we can manage it :) 15:17:25 I'll be out Thu-Fri 15:17:40 ack 15:17:43 (the whole RH is out on Fri IIUC) 15:17:45 it looks ok, probably at the limit 15:17:55 looks like we got some duplicates in there too 15:18:00 * TheJulia tires to clean some of that up 15:18:20 Mark Goddard proposed openstack/bifrost master: Fix bifrost_inventory.py with no nodes enrolled https://review.opendev.org/734115 15:20:22 I concur, it makes up for last week :) 15:20:31 Well, I'll take this as lazy consensus and we can move on! 15:20:35 ++ 15:20:44 #topic Discussion 15:20:59 I added one topic, "What did everyone think of the virtual PTG?" 15:21:11 Comments/suggestions/ideas? 15:21:12 + Productive, easy to attend (for me) 15:21:17 - Missing evenings together 15:21:17 I really like we had a brainstorming session 15:21:34 + Inclusive (towards people who cannot travel) 15:21:51 it was awesome, but i have the feeling it requires more energy =) 15:21:58 Indeed. Some technical problems though 15:22:06 it was the first time 15:22:19 Our two hour time boxing might have been a little too little at least for the first day or two 15:22:22 it felt a bit "compressed" 15:22:29 That is a good way to put it 15:22:35 sad that I missed it. Clashed with kolla :( 15:22:41 + Having normal work time in-between allowed follow-uping 15:22:46 There was no topic evolution, no "oh, we should talk about xyz too! Schedule it!" 15:23:01 Yeah, normal work was also a source of interrupts 15:23:12 I felt super drained every day, but a lot is going on right now 15:23:33 Well, if there is nothing else! Time to move on to the BareMetal SIG 15:23:42 - hallway chatter with ironic and non-ironic folks so felt more like 'only ironic' 15:23:54 yeah, there was pretty much no hallway chatter 15:23:59 yeah, less interaction with other groups 15:24:05 cross-project communication was basically non-existant 15:24:12 i guess 'missing community interaction' ? 15:24:13 except in the pre-defined verticals 15:24:36 but maybe pros outweigh cons. 15:24:58 I concur, and I think everything said so far echos what my feedback to the foundation contained 15:25:01 well, we don't know when large-scale travel is enabled again... 15:25:18 - no maple syrup :/ 15:25:38 - and no mountain views from the venue 15:25:46 * rloo gets some from fridge and passes to rpittau 15:25:54 and sea planes, and the Tap & Barrel 15:26:00 rloo: thanks for the thought :) 15:26:10 #LongingForTheVancouverConventionCenter 15:26:24 I'd really, really love the Summit in Berlin to happen, but.. 15:26:30 ditto 15:26:31 * iurygregory never went to Vancouver =( 15:26:39 Anyway, onward to the SIG 15:26:43 #topic BareMetal SIG 15:26:59 Looks like arne_wiebalck is pushing things forward (Thanks Arne!) 15:27:15 A doodle has been posted for a final review of the whitepaper 15:27:23 #link https://doodle.com/poll/afwgy9zs8fi55wqe 15:27:25 * arne_wiebalck wakes up 15:27:27 arne_wiebalck: is there anything else? 15:27:46 good morning arne_wiebalck :) 15:27:50 I don't think so, I hope we can get things into a final state on Wed. 15:27:54 * iurygregory saw an email from arne_wiebalck to the ml 15:28:06 (assuming we meet on Wed) 15:28:36 those of us from RH have a conflicting meeting on Tue IIRC 15:28:54 ok, Wed seems fine so far ... TheJulia ? 15:29:27 opening other calendar 15:29:59 I believe wed is the only option for us 15:30:07 Wednesday works nicely 15:30:18 ok, I will close the doodle and sent out an invite 15:30:24 *send 15:30:54 ok 15:31:12 Then onward to RFE review? 15:31:15 ++ 15:31:30 #topic RFE Review 15:31:44 dtantsur has two topics for us to discuss in terms of RFEs 15:31:53 * TheJulia gives dtantsur the stage 15:31:56 thx :) 15:32:08 #link https://storyboard.openstack.org/#!/story/2007771 [dtantsur] Limited deployment capabilities without BMC credentials 15:32:29 so, this stems from the never-ending (at least on our side) requests to have at least $something working without the power control 15:32:49 I've noticed that we pretty much have all the bits in place and with a few fixes we can define a flow that will work 15:32:59 I read this first one and it seemed logical, the very last part seemed a little vague, but it seems reasonable to me 15:33:01 this RFE is about trying it, fixing lose ends and documenting 15:33:50 I had a quick look and it seems ok to me 15:33:53 seems really reasonable, since yeah, the bits are all there really 15:35:39 * dtantsur hears nice June crickets 15:36:00 heh 15:36:04 next rfe? 15:36:16 after initial deployment, someone would need to change the boot order and boot into disk? 15:36:16 #link https://storyboard.openstack.org/#!/story/2007610 [dtantsur] Networking boot fallback for ironic 15:36:18 i'm still reading/groking the rfe 15:36:27 kaifeng: the 2nd link is actually about it 15:36:27 but can give feedback later. 15:36:36 seems like a bit of a hack to me, but still thinking about it 15:36:36 rloo: the 2 RFEs are related, go ahead 15:36:54 it is a bit of a hack, yes, but a hack that a lot of people will likely appreciate 15:37:05 dtantsur: oh, i will / should read both first then. sorry, i can't give good feedback on it in this meeting. (cuz i'm slow) 15:37:16 turns out a lot of people want slightly different workflows 15:37:18 no worries, the goal is to highlight the RFEs 15:37:27 not to make all possible decisions here and now 15:37:41 the 2nd RFE is about the boot order indeed 15:38:02 in addition to RFE #1, there are two more cases where we cannot assert the boot order the way we want: SNMP and Redfish 15:38:17 the latter is affected by the lack of (a supported) way to configure persistent boot on some hardware 15:38:29 the idea is simple (and I've learned today foreman does it too): 15:38:52 generate PXE environment for active nodes with local boot, so that if they boot from network, we tell them to boot from disk 15:39:04 I feel a deploy interface cfeature compatability matrix coming soon to a review queue near you 15:39:09 the 2nd one seems reasonable to me for I just hit a machine failed to perform a uefi local boot today :) 15:39:09 heh 15:39:13 s/cfeature/feature/ 15:39:39 kaifeng: as in the efibootmgr failed to set things up? 15:39:45 " if they boot from network, we tell them to boot from disk" O.o 15:40:09 essentially, an iPXE file with 'sanboot' 15:40:26 we do it for whole disk images with netboot already 15:40:34 TheJulia: I think it's a bmc firmware issue, it just can't even manually change settings in the bios 15:40:49 kaifeng: fun! :( 15:41:23 Well, if there is nothing to discuss further on the RFEs, time for open discussion 15:42:17 yep 15:42:20 #topic Open Discussion 15:42:41 I am hitting this issue https://bugzilla.redhat.com/show_bug.cgi?id=1033739 hard :( 15:42:41 bugzilla.redhat.com bug 1033739 in libvirt "Dnsmasq need prefix in dhcp-range for dhcpv6" [Unspecified,Closed: upstream] - Assigned to libvirt-maint 15:43:22 oh no wrong link, sorry 15:43:30 ok :) 15:43:46 https://bugzilla.redhat.com/show_bug.cgi?id=1575026 15:43:47 bugzilla.redhat.com bug 1575026 in dnsmasq "Can't PXE/iPXE boot with dnsmasq and DHCPv6" [Medium,Closed: wontfix] - Assigned to pemensik 15:44:04 I am trying to make ironic inspector supports ipv6 discovery 15:44:12 you're using RHEL 7? 15:44:45 I am using centos7, this is the closest thing I googled 15:44:55 is CentOS 8 an option for you? 15:45:00 It has/will be fixed there 15:45:26 hjensas and bfournie have had success with IPv6 and ironic-inspector, but on RHEL 8 15:46:43 the weird thing is we never hit this issue in non-ironic related area, i can't tell the difference 15:47:15 IIRC the problem is with several times we do DHCP during iPXE boot 15:47:33 kaifeng: There are a couple different issues, but some in part is the way dnsmasq works and how the ipxe client re-does dhcp with a different client identifier 15:47:39 also, part of this is how dhcpv6 was designed 15:48:12 * dtantsur is of opinion that IPv6 was designed to show the masses that computers are cursed 15:48:28 v4 I guess was the "just work" version 15:48:36 and v6 was the "lets over-engineer!" 15:48:40 we are not using ipxe, just pxe 15:48:54 kaifeng: instances failing to dhcp? 15:49:42 hi rpittau do u remember the pxe could not boot issue i post days ago? 15:50:07 It seems a bit similiar to the kaifeng post issue. 15:50:09 I guess we're done with today's meeting. Thanks everyone. I'm sure we'll continue to discuss IPv6 for a little while 15:50:10 no, i am still trying the introspection poart 15:50:21 part 15:50:30 kaifeng: ramdisk failing to get an address? 15:51:02 the interesting thing is first time works, and if we wait for a while, it also works, but i can't estimate how long 15:51:32 hjensas: ^^^ 15:51:33 this does sound like an issue with leases 15:51:39 if we do two consecutive introspection, dnsmasq would complain no addresses on the second one 15:51:44 * dtantsur suggests we wrap it up indeed and continue off-meeting 15:51:48 yeah, sounds slightly different :\ 15:51:52 ++ 15:51:55 Thanks everyone! 15:52:01 Have a wonderful week! 15:52:02 introspection using neutron, or using dnsmasq standalone with a dynamic pool? 15:52:03 #endmeeting