15:00:34 #startmeeting ironic 15:00:35 Meeting started Mon Apr 6 15:00:34 2020 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:35 o/ 15:00:35 o/ 15:00:36 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:38 The meeting name has been set to 'ironic' 15:00:40 o/ 15:00:43 o/ 15:00:45 \o 15:00:50 o/ 15:00:50 \o 15:00:51 o/ 15:01:02 arne_wiebalck: I think the ipa patch merged, the alternative is waiting an entire cycle because it is a breaking change for older IPA images 15:01:10 Good morning everyone! 15:01:11 o/ 15:01:25 Our agenda can be found on the wiki! 15:01:27 #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting 15:01:30 o/ 15:01:39 We have a number of items, so we should go ahead and get started! 15:01:53 #topic Announcements / Reminders 15:01:53 o/ 15:02:15 TheJulia: ok, let's take this after the meeting if you have a moment 15:02:36 o/ 15:02:59 April 9th is the Ussuri-3 milestone which means Cycle highlights are due this week. If there is anything people wish for me to specifically mention, please let me know ASAP. 15:03:25 Openstack's RC1 deadline is April 23rd, which means we need to cut our release for the ussuri cycle before then. 15:03:35 Any questions on that? 15:03:47 Does anyone heave anything else to announce or remind us of? 15:04:02 TheJulia: next monday is bank holiday I think 15:04:10 ironic-tempest-ipa-wholedisk-direct-tinyipa-multinode is back voting =) 15:04:11 and Friday too for some of us 15:04:27 ++ for dtantsur said 15:04:39 * TheJulia senses she needs to take a day or two off 15:04:39 yep, but for next monday I guess we'll have to cancel the meeting 15:04:59 That seems like it is viable 15:05:01 Any objections? 15:05:13 o/ 15:05:29 (sorry - showing up late - that was not an objection) 15:05:34 * TheJulia hears no objections :) 15:05:36 o/ 15:05:42 cdearborn: I figured that :) 15:05:49 * iurygregory was about to ask if it was XD 15:06:22 #agreed Ironic will not be holding a meeting on the 13th because of the Holiday. 15:06:34 Can I get a volunteer to send a note to the mailing list? 15:06:53 I can do it 15:06:55 \o 15:07:05 iurygregory: go ahead if you want to :) 15:07:11 Excellent! Thanks! 15:07:18 dtantsur, if you are ok with it =) 15:07:32 I'm always ok with less work ;) 15:07:55 action item on me =) 15:08:17 #action iurygregory to send a note to the mailing list indicating we will not be meeting next week. 15:08:24 Onward! 15:08:27 #topic Review action items from previous meeting 15:08:59 We had one action item last week, which was for me to send out another doodle for whitepaper meeting, but sadly I did not upon starting to perceive the stress everyone was under. 15:09:10 arne_wiebalck: Is this what your topic item is on, trying to figure out a time to revisit? 15:09:34 TheJulia: yes, I think we should keep the momentum 15:09:55 TheJulia: there were quite some additions and reviews 15:09:59 arne_wiebalck: would you be willing to send out the doodle? 15:10:02 ++ 15:10:06 TheJulia: sure 15:10:11 arne_wiebalck: excellent! 15:10:20 Okay then, ward if there are no objections! 15:11:03 #topic Review subteam status reports 15:11:13 #link https://etherpad.openstack.org/p/IronicWhiteBoard 15:11:17 Line 287 15:12:01 mgoddard: dtantsur: looks like the deploy steps stuffs could use an update on the whiteboard. 15:12:16 it's probably pretty much correct, lemme see 15:13:17 arne_wiebalck: Looks like software raid only has just the doc patch and the ironic patch we were just talking about left. \o/ 15:13:50 done 15:13:53 That seems like it is it to me. Is everyone read to proceed onward? 15:13:55 dtantsur: thanks! 15:13:59 TheJulia: \o/ 15:14:13 ++ 15:14:41 Looks like ?rpittau? is adding some stuff to CI improvements 15:14:54 just one thing, done 15:15:01 thanks 15:15:05 Onward! 15:15:12 #topic Deciding on priorities for the coming week! 15:15:20 #link https://etherpad.openstack.org/p/IronicWhiteBoard 15:15:43 Starting at line 130 15:16:03 On a positive note, a ton of patches merged last week! Lets keep up the momentum because I need to write the cycle highlights :) 15:16:03 I've added a few dependency clean-ups there, although most of them have been approved (thanks!) 15:16:13 * TheJulia cleans up the merged items now 15:19:17 How does that look for everyone? 15:19:38 ok 15:19:49 ok 15:20:14 LGTM 15:20:15 Seems like a lot, but 5 or 6 of them are already in-flight or will be approved quickly because they are really small patches. 15:20:49 Anyway, if we're good with that, I guess it is time to proceed to discussion! 15:20:56 Any objections? 15:21:07 ++ 15:21:19 * TheJulia takes that as no :) 15:21:25 #topic Discussion 15:21:39 First topic is rpittau's regarding the bioic OVMF package. 15:21:45 #link https://bugs.launchpad.net/ubuntu/+source/edk2/+bug/1821729 15:21:46 Launchpad bug 1821729 in edk2 (Ubuntu Cosmic) "UEFI in ovmf package causes kernel panic" [Undecided,Invalid] 15:21:51 rpittau: the microphone is yours 15:21:52 thanks 15:22:09 I'm testing the ovmf native package on bionic using dib 15:22:30 Verification of a change to openstack/python-ironic-inspector-client failed: Make oslo.i18n an optional dependency https://review.opendev.org/717730 15:22:31 Verification of a change to openstack/python-ironic-inspector-client failed: Get rid of the oslo.utils requirement https://review.opendev.org/717731 15:22:32 Verification of a change to openstack/python-ironic-inspector-client failed: Move osc-lib to test-requirements https://review.opendev.org/717733 15:22:45 at the moment whtat I'm seeing is that the node gets the ip but it hangs before getting the NBP 15:22:51 * TheJulia glares at openstackgerrit bot... :) 15:22:59 this is what I'm seeing http://paste.openstack.org/show/791675/ 15:23:29 thee console log stops right after teh Station IP address 15:23:45 and the only anomaly is the tftpd error in journal 15:23:51 connection tracking/firewalling blocking the return connection? 15:24:21 I checked the firewall but I can't see anything wrong, apparently, also nothing changed on that side :/ 15:24:45 Well bionic did make default connection tracker helper chagnes 15:24:57 But I thought we accounted for that in our CI jobs 15:25:08 yes, and we have a workaround for that in place 15:25:32 otherwise many other things would not work 15:25:38 yeah 15:26:16 it could be that ovmf now requires another rule in iptables, but I can't really say what 15:26:16 connection refused makes me think that the tftp daemon is not able to send packets to OVMF code running in the VM which makes me think that maybe they broke their tftp code?!? 15:26:36 O.o 15:26:36 I feel like the socket needs to be tcpdumped 15:26:42 woa 15:26:49 just tcpdump the entire thing and see what we've got going back and forth 15:27:17 if we don't see ovmf emitting a icmp packet to the TFTP server address, then we know it is hypervisor firewalling 15:27:39 ok, I'll see to that 15:27:51 rpittau: the ipv6 ci job has an example of using tcpdump in it if memory serves 15:28:05 Hopefully that helps 15:28:10 the thing that's making me crazy is thtat updating the package and restarting the server creation works 15:28:22 ugh :( 15:28:49 Well, lets at least get a tcpdump.. because if ovmf is telling the server "send me the file on this port" and then responds with ICMP, it is clearly an OVMF bug 15:28:55 ok 15:29:18 Are we good to move on? 15:29:25 both updating package and restarting server create togheter. Or both those are workaround? 15:29:26 I am :) 15:29:58 hjensas: starting with xenial pkg, creating node, then desotryong node, upgrading package, creating node again works 15:30:20 creating node directly with bionic package does not work 15:30:30 Oo 15:30:31 s/node/server 15:30:40 rpittau: wow. Thanks, crystal clear. 15:30:44 :( 15:31:06 Next topic is dtantsur's on his proposal to make ironic a top-level opendev project. 15:31:13 #link http://lists.openstack.org/pipermail/openstack-discuss/2020-April/013757.html 15:31:15 Merged openstack/ironic master: Change multinode job to voting https://review.opendev.org/717348 15:31:16 Merged openstack/python-ironicclient master: Make oslo.i18n an optional dependency https://review.opendev.org/717729 15:31:17 dtantsur: the floor is yours 15:31:18 Merged openstack/python-ironicclient master: A standalone CLI for ironic https://review.opendev.org/717273 15:31:30 well, I mostly wanted to make sure that all parties here have read and commented 15:31:46 I know some of us have discussed it previously, but the discussion stalled 15:32:02 a lot of interesting stuff in the thread, highly recommend 15:32:21 That thread is also a little hard to read because there is a strong undertone of hurt and fear 15:32:31 yeah 15:32:39 not without that indeed. but also quite a few useful thoughts. 15:32:47 Indeed 15:33:04 I think it's important that people who care about ironic (many of them are present) read, think and comment 15:33:18 otherwise the thread will stall and end up with nothing 15:33:19 ++ 15:33:36 even if we don't end up breaking up from openstack, I'm sure there can be a lot of useful follow-ups 15:33:57 * dtantsur needs to provide a new round of responses to people, send spoons! 15:34:21 It is a discussion that needs to happen, and I love that it returning us to the long debated what is openstack thread 15:35:20 I seem to be great at opening cans of worms :D 15:35:25 dtantsur: I'm running low, but maybe http://www.cutestpaw.com/articles/50-cute-cats-make-your-life-happier/ might help 15:35:32 * hjensas likes that the thread started stirring a bit about how to "get to market" that opentack is a collection of services. That can be used standalone or just 1 or 2. 15:35:46 TheJulia: that may work, thanks! 15:36:01 hjensas: that's a possible useful outcome 15:36:15 hjensas: and it is highlighting a lot of major disconnects that need mending... somehow 15:37:21 indeed 15:37:47 Anyway, dtantsur thank you for raising the topic, and if you need more cut pet links, I'll happily dig for them because I need spoons too. 15:37:56 more cut pet links \o/ 15:37:59 np :) 15:38:17 cut pet links should be shared =) 15:38:27 #agreed All cute pet photos should be shared. 15:38:35 Shall we proceed onward? 15:38:37 The MAAS web site link etingof shared provided insight into what we're competing with: https://maas.io/ 15:39:21 that's me following the thread in question: http://www.cutestpaw.com/wp-content/uploads/2011/11/Kitten.jpg 15:39:34 * kaifeng_ actually investigated and compared with ironic before 15:39:35 * etingof thinks MaaS is a tough competitor, at least from marketing and PR standpoints 15:39:49 rpioso: exactly. somebody (mnaser?) asked how the side should look - this is how probably 15:39:56 * site 15:40:02 And we have no marketing or PR. Or to put it more precisely, we are the marketing and PR team as well. 15:40:11 kaifeng_: I'd be curious to hear about your experience! 15:40:40 TheJulia: yeah, we cannot get to the same level.. but we may get a nice landing page! 15:40:51 * arne_wiebalck was also quite intrigued by the https://rebar.digital/ link etingof posted earlier 15:40:53 dtantsur: interesting you note that, there was an effort that was pushed many moons ago to try and simplify project landing pages and not have them full of links of descriptie text 15:41:20 My feeling on maas is it is a project to quickly bring a data center into a public cloud 15:41:21 for that we need to *have* project landing pages 15:41:22 it flopped of course because some still wanted the index style landing page 15:41:43 I don't believe docs.o.o/$project is 15:42:04 no, it isen't, and the intent of the docs team was to try and craft them into that 15:42:10 Sadly, things change too 15:42:21 Anyway, lots of good discussion, and we should move onto the SIG! 15:42:29 ++ 15:42:34 #topic Baremetal SIG 15:42:51 arne_wiebalck: Is there anything else to discuss that we didn't cover earlier? 15:43:15 TheJulia: Not relly. Thanks everyone for contributing, looks much better now! 15:43:18 really 15:43:34 * etingof will push his bits in a bit 15:43:46 etingof: tanks! 15:43:55 * arne_wiebalck cannot type 15:44:01 dtantsur: fyi, this is me right now http://www.cutestpaw.com/wp-content/uploads/2011/11/To-infinity-and-beyond.jpeg needing more coffee 15:44:01 I'm a bit behind on my part, sorry 15:44:13 Same, I'm getting pulled in many directions 15:44:18 looks like you indeed :) 15:44:31 same :/ 15:44:41 I should have time this week, we've not been running the 3d rpinters around the clock since Saturday morning. 15:45:20 printers 15:45:31 Anyway, onward to Open Discussion 15:45:34 #topic Open Discussion 15:45:47 I have a follow-up question to https://review.opendev.org/#/c/717273/ 15:45:48 patch 717273 - python-ironicclient - A standalone CLI for ironic (MERGED) - 9 patch sets 15:45:57 first, yes, we now have a 'baremetal' command without OSC 15:46:04 \o/ 15:46:08 NobodyCam: ^^^^^ 15:46:17 second, what would people think if I enable a way for the same command to add ironic-inspector-client commands? 15:46:38 i.e. if python-ironic-inspector-client is installed, then 'baremetal introspection blah-blah' is a thing 15:46:42 dtantsur: Might be interesting, do you think you could get it up in the next 24 hours? 15:46:56 TheJulia: tomorrow morning quite likely 15:47:03 I say do it! 15:47:19 Good morning folks and WooHoo 15:47:24 I got close to it, but then stumbled upon an issue in ironic-inspector-client 15:47:28 I think I can work around it 15:47:47 Okay, awesome 15:47:49 Thanks dtantsur ! 15:50:12 Do we have anything else to discuss? 15:50:46 justa heads up for the cores to look at the changes in the ironic-tempest-plugin https://review.opendev.org/712041 15:50:46 patch 712041 - ironic-tempest-plugin - Split Tempest tests between ipmi and redfish - 8 patch sets 15:50:47 =) 15:51:03 * etingof pulls the mic 15:51:17 go for it etingof o/ 15:51:17 iurygregory: I may have already added those.. 15:51:19 etingof: ? 15:51:29 it turned out that ironic can't vmedia boot Supermicros because they do vmedia it in their own OEM way. what we could do is to introduce a new sushy extension package - sushy-oem-supermicro. so I am wondering how useful that would be...? 15:51:30 TheJulia, oh thanks =) 15:51:48 etingof: not just package, also a new driver, right? 15:51:56 living in ironic-staging-drivers since it's not vendor-backed? 15:52:13 well, ideally - just the mediation layer well-hidden inside sushy extension... 15:52:17 s/driver/hardware type/ (heh, of all people, I'm not supposed to confuse these) 15:52:35 etingof: the way sushy works now, OEM stuff won't be magically picked up 15:53:16 yeah, ideally that should be magical whenever possible 15:53:36 but, yes, explicit vendor driver would do as well 15:54:11 I'm trying to understand the scope of your proposal 15:54:19 it seems the magic is not impossible 15:54:31 do you propose to also modify sushy and/or the redfish hw type to magically pick up OEM changes? 15:54:37 similarly to what we discussed in GVA? 15:55:19 yeah, that seems ideal to me... look, if sushy/can learn from BMC what kind of vendor is that, sushy could probably pick the right extension to mediate... 15:55:35 yeah, or it can be a driver_info field 15:55:42 I'd read a detailed RFE on this 15:55:58 right 15:56:08 ++ 15:56:31 perhaps the case of supermicro could serve as a precedent for sushy OEM automation... 15:56:35 yeah, I'm not exactly groking what and how at the moment so I think something written might be helpful 15:57:07 etingof: Would the iDRAC vmedia diffs benefit from the mediation layer? 15:57:47 I guess it depends on what is going to be proposed? 15:57:50 rpioso, ideally, if we could hide all idrac-redfish differences in sushy extension... 15:58:58 etingof: We would still have a need to be able to offer our vendor value add. Not sure how that would be handled. 15:59:25 rpioso, like what? 15:59:29 via selling hardware? ;) 15:59:35 sorry, couldn't resist 15:59:48 dtantsur: heh 16:00:01 I don't think "you need a non-standard driver for our hardware" is a selling point actually 16:00:02 for the ironic operator it is important to deploy the machine 16:00:11 dtantsur++ 16:00:13 etingof: Server Configuration Profile (SCP). 16:00:20 * rpioso looks for the story 16:00:31 ah, you mean, you have more features coming? 16:00:44 this is fine. I think the current discussion is to make all base features of 'redfish' hw type work 16:00:47 * TheJulia wonders if we are done with the meeting 16:00:51 we probably are 16:01:02 etingof: https://storyboard.openstack.org/#!/story/2003594 16:01:27 TheJulia should now dismiss us 16:01:37 Excellent! 16:01:51 Have a wonderful week everyone! Stay safe! And remember to share cute pet photos! 16:02:01 Also photos of ferrets. My wife <3s them. 16:02:05 #endmeeting