15:00:00 #startmeeting ironic 15:00:01 Meeting started Mon Jan 27 15:00:00 2020 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:02 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:04 The meeting name has been set to 'ironic' 15:00:04 o/ everyone! 15:00:18 \o 15:00:20 \o 15:00:24 o/ 15:00:27 o/ 15:00:27 o/ 15:00:28 Our agenda for this morning can be found on the wiki 15:00:31 #link https://wiki.openstack.org/wiki/Meetings/Ironic 15:00:36 o/ 15:00:40 o/ 15:00:59 #topic Announcements/Reminders 15:01:16 #info OpenStack Ironic Mid-cycle Meetup, 25-26 February 2020, CERN, Geneva, Switzerland 15:01:17 \o 15:01:18 o/ 15:01:18 #link https://indico.cern.ch/event/863986/ 15:01:41 o/ 15:01:50 #link https://etherpad.openstack.org/p/ironic-ussuri-midcycle 15:01:53 \o 15:02:27 #info Opendev + PTG Vancouver, Dates: Mon, Jun 8, 2020 - Thu, Jun 11, 2020 15:02:52 o/ 15:03:20 Early Bird tickets are for sale now! 15:03:23 #link https://www.eventbrite.com/e/opendev-ptg-vancouver-2020-tickets-88923270897 15:04:00 #info Python 3.8 non-voting jobs are being added to Ussuri 15:04:02 #link https://review.opendev.org/693401 15:04:03 patch 693401 - openstack-zuul-jobs - Add non-voting tox py38 jobs to Ussuri templates (MERGED) - 3 patch sets 15:04:13 just 6 months in advance \o/ 15:04:18 (this is re tickets, not jobs) 15:04:26 Does anyone have anything else to bring up for announcements/reminders? 15:04:41 This is actually a good thing :) 15:06:30 * TheJulia guesses we're good to proceed 15:07:15 We appear to have had no action items last week, so I think we can skip that section and move into subteam status updates 15:08:28 #topic Review subteam status reports 15:08:52 #link https://etherpad.openstack.org/p/IronicWhiteBoard 15:08:55 Line 278 15:10:30 Regarding the Bare Metal Program/SIG whitepaper, I did go back and forth with the foundation in email last week, and I have a few more emails to read. Seems like there was a disconnect that occurred :( 15:11:13 ok 15:11:51 maybe we should just try to polish the rest as we do not move fwd since a while now 15:12:27 * arne_wiebalck certainly used this as a excuse to not do anything :) 15:12:42 arne_wiebalck: that is an option. I'm still trying to get the RH case study approved for release :( I'll have to follow-up this week on that again 15:12:53 ok, thanks! 15:13:06 Looks like tzumainn's mutlitenancy related work could use some reviews 15:14:05 etingof: looks like your needing the spec reviewed? 15:14:23 very much so! 15:15:20 arne_wiebalck: re software raid: are we just needing the ipa patches rebased 15:15:47 TheJulia: there is some more to do, I think 15:15:55 * dtantsur wants to look into device selection for software RAID 15:16:09 part of the functionality has been moved to another patch 15:16:17 I think 15:16:27 also: I did not yet test this on a real machine 15:16:37 I'll try to do this this week 15:16:49 dtantsur: ++ 15:16:58 If we can get the section updated with a current status, I think that would help so we can focus reviews 15:17:41 On the subject of v6 CI, still not had time to revisit :( 15:18:11 iurygregory: your UEFI patch has taken some functionality from the UEIF RAID patch, nno? 15:18:30 arne_wiebalck, only the part to look for partition 15:18:44 in utils.py 15:18:55 right, but this then needs to be taken out of xXraphXx 's pacth set 15:19:20 I have some other UEFI things to do and can fold this in 15:19:26 I hope 15:19:30 :) 15:20:06 Merged openstack/bifrost master: Use upper-constraints everywhere in ironic installation https://review.opendev.org/703998 15:20:49 the retirement spec is mostly implemented, if someone is interested to review (it's quite straight forward I think) 15:21:04 Thanks rpittau for the first comments! 15:21:15 arne_wiebalck: awesome 15:21:19 arne_wiebalck: my pleasure :) 15:21:53 kudos to dtantsur who delivered the blue print in fact 15:22:04 Are we good to move to Priorities for the week? 15:22:28 arne_wiebalck, yeah I pushed in a separate patch after talking with him =) 15:22:41 did it? :) 15:22:46 did I? :) 15:23:24 I studied your node protection patch :) 15:23:29 heh 15:23:47 ah, right, you mentioned it the other day :) 15:23:51 * arne_wiebalck thinks "studied" sounds so much nicer than "copied" 15:24:22 * TheJulia concurs 15:24:31 So lets move on to priorities 15:24:41 #topic Priorities for the coming week 15:24:57 #link https://etherpad.openstack.org/p/IronicWhiteBoard 15:24:59 Line 169 15:25:37 * TheJulia cleans up the bifrost entry 15:25:51 Merged openstack/sushy master: Add Task Monitor support for async operations https://review.opendev.org/670360 15:26:34 Looks like there is only one outstanding mergable patch in that series, but we stil need to get the other jobs working 15:26:44 I know there are a few others out there in review though 15:27:17 Centos 8 is very close, at least with tinyipa 15:27:24 yeah, this ^^^ 15:27:43 rpittau: well, DIB images are known to work for CentOS 8, using it as a base OS is a problem :) 15:28:19 awesome 15:28:22 it's a new major release for a reason :) 15:28:36 Looks like some of the revisions marked approved have merged 15:28:38 Marking them as such 15:29:53 Is anyone aware of any additional patches that should be added to the list? 15:30:40 I added the scoped rules for introspection, hope that's ok. 15:31:53 I noticed, and tagged dmitry on it. It looks okay to me, but i would really like a practical use example in the documentation 15:32:11 TheJulia: agreed 15:32:32 it's on my list, just never got to it (imagine the size of the list, given that it barely shrinks..) 15:32:58 no worries, not urgent :) 15:32:59 Merged openstack/bifrost master: Set Ansible default version to 2.8 https://review.opendev.org/703633 15:33:11 dtantsur: yeah, the lists never get smaller :( 15:33:50 I guess I'm good with the priorities, we just need to update the notes/remove the items as stuff merges 15:35:09 Did we ever get to have ops midcycle feedback? I know johnthetubaguy is not here 15:36:28 So moving on, I guess we can skip baremetal sig and move directly to Open Discussion? 15:36:31 arne_wiebalck: ^^? 15:36:37 yes 15:36:45 ++ 15:37:38 #topic Open discussion 15:38:00 So I'm thinking there should be coffee handed out at the beginning of ironic meetings? Or maybe a pre-meeting coffee reminder? :) 15:38:07 Just to wake everyone up :) 15:38:21 so, a thought (caused by downstream discussion). with iPXE we can have a hybrid legacy/UEFI boot environment. shall we implement it? 15:38:25 also coffee++ 15:38:50 dtantsur: could you elaborate what this means? 15:38:50 dtantsur: do you mean secure boot? 15:38:55 mmm, no 15:39:05 currently an operator has to choose a boot mode for a node 15:39:07 Then clarification required 15:39:13 oh! 15:39:19 if they fail to (and the driver cannot enforce a boot mode), the boot will fail 15:39:32 Iury Gregory Melo Ferreira proposed openstack/ironic master: WIP - Native zuulv3 grenade job for ironic https://review.opendev.org/703098 15:39:33 actually, for ironic-inspector we tend to build environments that happily support both legacy and UEFI 15:39:34 dtantsur: right, I had this this week :) 15:39:45 we could do it for cleaning/deployment with iPXE 15:39:48 so offer pointers to both options based on arch? 15:39:52 yeah 15:40:07 That seems logical, we've been doing similar stuff in standalone for a while 15:40:11 I'm kind of surprised we've not done it 15:40:22 line in https://opendev.org/openstack/puppet-ironic/src/branch/master/templates/inspector_dnsmasq_http.erb 15:40:37 I only asked about secure boot because someone downstream also asked me about secure boot and that is one of those subjects I kind of want to avoid like the plague since nobody will like the answer 15:40:42 :D 15:40:57 I suspect one day we'll have to do it... 15:41:00 "You want secure boot.... " "uhh... have you gotten MSFT to sign your ipxe.efi binary?" 15:41:16 "What?!? why can't I use shim?!?" 15:41:18 well, we can go easier on them: "have you inserted your keys in your UEFI ROM"? 15:41:23 :D 15:41:37 ++ and sign the ipxe loader! 15:41:51 I mean... you can do it, right? and then we can support it. 15:41:56 let's do virtual media boot instead 15:42:01 etingof: ++ 15:42:14 * dtantsur hopes we don't have to sign the virtual media ISO 15:42:24 but yes, it is a pretty valid answer 15:42:30 dtantsur: you can actually.... but only on certian hardware 15:42:32 maybe we should try it and see if it works out of box 15:42:58 I don't think the iso needs to be signed, just the kernel/shim/grub on the iso 15:42:58 TheJulia: well, let us call it commercial advantage of "certain hardware" :) 15:43:13 dtantsur: ++ 15:43:37 I do think we should try it with virtual media on one of our labs and if it works - document it 15:43:53 ++ 15:43:59 * etingof never heard of the necessity to sign ISOs 15:44:16 yolanda: ^ might be a good path to follow 15:44:47 * TheJulia wonders if we have anything else to discuss or plot this morning 15:44:53 hi! just wanted to mention the node lessee spec again - https://review.opendev.org/701061 - mostly curious if, at a high-level, people think I'm completely off base in the approach (in which case I can start coming up with a new one) 15:44:53 patch 701061 - ironic-specs - Create node lessee field - 2 patch sets 15:45:17 tzumainn: I'm nearly slacking off on it to wait for the midcycle :) 15:45:37 dtantsur, that's fair too! 15:45:38 I'd like to move a bit more careful than with owner to avoid making the same mistakes 15:46:09 btw we have johnthetubaguy here to discuss the ops meetup 15:46:18 o/ johnthetubaguy 15:46:29 hey 15:46:33 dtantsur, yep, definitely - I'd like to tie it in with allocation from the start and wire it into... other stuff I might not know about 15:46:38 tzumainn: dtantsur: yeah, leasee just becomes another layer of complexity :\ 15:46:50 johnthetubaguy: How was the ops midcycle? 15:46:53 #link https://etherpad.openstack.org/p/LON-2020-OPS-AGENDA 15:47:04 so I think it was generally really good 15:47:41 about 50% UK, 40% other bits of Europe, 10% others 15:47:41 Any takeaways for the ironic community? Feedback? needs? etc? 15:47:47 nice! 15:47:53 so we had a session 15:48:02 #link https://etherpad.openstack.org/p/LON-2020-OPS-BARE-METAL 15:48:23 a big bit of feedback was where to get started and what to read 15:48:58 looking at etherpad, "Sometimes hard to know where to report issues". Curious to know where they looked, or where we need to advertise... 15:49:03 I mentioned the "universe from nothing" using kayboe inside a VM 15:49:12 or maybe it was cuz they didn't know if it was ironic or nova or neutron 15:49:15 rloo: that was more a debugging issue 15:49:16 yeah 15:49:18 that 15:49:40 understandable johnthetubaguy. thx. 15:49:41 not loads of actual users, but lots of interest 15:50:17 mentioning how networking-generic switch / the ansible one works made people a bit less confused about how it could possibly work 15:50:39 so in summary, I think its the getting started bits 15:50:50 that plus the steep learning curve ocmment makes me think we need less developer orietned documentation 15:50:59 oriented getting started documentaiton 15:51:06 now there were some compliments for the docs details, but yeah, what you just said 15:51:16 sooo many ways to use ironic. where to start, which way to use? :) 15:51:43 yeah, we tried to go down the path of detailing options for that and got stuck in basically the back and forth of nitpicking archtectures 15:52:04 but even that, is not getting started at the level likely needed 15:52:09 I ended up pointing them at this: http://www.stackhpc.com/universe-from-nothing.html 15:52:21 * TheJulia adds note to read that 15:52:22 because it needs one 32 GB machine and you get to poke it with a stick 15:52:42 maybe we should just start by itemizing the different ways. then elaborate later, link to stuff like kayobe 15:52:43 johnthetubaguy: can we turn that in a playbook that people can clone and execute? 15:52:56 and the same with bifrost, some sort of "very quick start" 15:53:17 dtantsur: well, it kind of has that already playbook wise 15:53:25 dtantsur: its more transferable skills to help do it "properly" with real hardware 15:53:36 The learning curve is the thing, providing more tooling is not going to solve that 15:53:43 it is only going to make it harder to understand what/how/why 15:54:12 TheJulia: I mean, even simpler. https://docs.openstack.org/bifrost/latest/install/index.html but as one script. 15:54:15 rloo: I've been thinking that is kind of the thing we're missing 15:54:23 Merged openstack/bifrost stable/train: Use upper-constraints everywhere in ironic installation https://review.opendev.org/704011 15:54:24 Merged openstack/bifrost stable/stein: Use upper-constraints everywhere in ironic installation https://review.opendev.org/704000 15:54:27 maybe even a script you can `curl | sudo sh` (SIGH) 15:54:42 I think its the docs more than the tools, I could be wrong 15:54:47 +1 rloo's point 15:54:49 rloo: in terms of expressing "Here is the what" in terms of ways, and how goes from there 15:54:53 https://docs.openstack.org/bifrost/latest/ could use an improvement 15:54:58 in terms of landing experience 15:55:20 so here's our doc landing page: https://docs.openstack.org/ironic/latest/ 15:55:20 more of a tasting menu, rather than a phone book explanation? 15:55:47 maybe under 'User Guide' ? not sure. 15:55:55 so... that is actually kind of where the doc folks wanted projects to start heding towards... but... that got hard resistance :( 15:56:17 well, we have a certain degree of freedom now 15:56:19 * TheJulia wonders if we just need a tasting-menu.rst file 15:56:24 ++ 15:56:39 "Here are the fine wines^Wmodes of install 15:56:41 " 15:56:43 tasting-menu.rst for dinner 15:56:57 Okay, I suspect we're reaching the end of our meeting 15:57:20 To install ironic, first obtain a whiskey^Wblank machine 15:57:21 I should say, there was an apatite for hearing about how people succeeded 15:57:26 * iurygregory : we will teach how to prepare good drinks also right? 15:57:31 that item about baremetal SIG and case studies might be really useful 15:57:32 johnthetubaguy: that is... a really interesting point 15:57:58 johnthetubaguy: by chance did you pick up on those operators perferred news/information channels? 15:57:59 so more real experiences / use cases ? 15:58:26 Yeah 15:58:34 we could document them somewhere 15:58:34 TheJulia: only consensus was screw the mailing list as no one reads it (only partially a joke) 15:58:40 like a collection of articles 15:58:52 Part of what happened in the early decline in interest is conferences got burned out on anything with openstack in the submission 15:58:59 so people gave up submitting talks 15:59:17 maybe on the docs site, you include the use cases, or links to where they live 15:59:30 just s/openstack/virtualization/g 15:59:31 people do seem to find the docs, at least the ones that actually care do 15:59:35 The last CFP review I participated, similar feelings on the word "container" was occuring 15:59:39 mayb in the wiki ? 15:59:51 * TheJulia needs to nuke the wiki from high orbit.... 15:59:56 :D 16:00:05 so maybe not... 16:00:10 Russians work on that already 16:00:46 twitter apparently worked for the ops summit, FWIW 16:00:48 for now, put it under 'User Guide' section, add list of diff ways to use ironic, then put down which companies use it with links (or not) to docs elsewhere. 16:01:17 Yeah, They have mentioned twitter before 16:01:30 rloo: that could work, then tweet that we did it 16:01:32 rloo: we could link to the whitepaper too 16:01:33 rloo: that could be a good place, yeah 16:01:41 I feel like we're reaching consensus 16:01:48 Anyway, we're past time, thanks everyone! 16:01:51 it is a start. the info could go elsewhere too, later. but let's just start :) 16:02:09 o/ 16:02:19 Thank you! 16:02:22 thanks! 16:02:33 #endmeeting