15:00:16 #startmeeting ironic 15:00:16 Meeting started Mon Sep 28 15:00:16 2020 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:17 o/ 15:00:18 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:20 The meeting name has been set to 'ironic' 15:00:27 o/ 15:00:29 o/ 15:00:29 \o 15:00:29 o/ 15:00:32 o/ 15:00:34 o/ 15:00:34 o/ 15:00:40 Good morning everyone! 15:00:42 \o 15:00:48 Our agenda today can be found on the wiki, as always. 15:00:50 #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting 15:00:57 o/ 15:01:07 #topic Announcements / Reminders 15:01:35 #info An etherpad for PTG items has been created. Please add your crazy and not so crazy ideas to it. 15:01:43 #link https://etherpad.opendev.org/p/ironic-wallaby-ptg 15:01:53 #info This week is Victoria R-2 15:02:28 In other words, we're near the end of the cycle. Priority as such remains CI and getting patches reviewed/merged. 15:03:38 #info All ironic-core members now have rights on stable branches in ironic projects. As a reminder, stable branch policy is to be adhered to. The caveat to that is as an operator focused/hardware coupled project, we do have some leeway. If in any doubt, please ask questions! 15:03:46 #link https://docs.openstack.org/project-team-guide/stable-branches.html#review-guidelines 15:04:17 #info TheJulia has proposed to return JayF to the ironic-core group. 15:04:26 Does anyone have anything else to announce or remind us of this week? 15:05:37 * TheJulia plays the sound of crickets 15:05:43 I guess not? 15:06:01 mm, releases? 15:06:19 Yes, we do need to go ahead and get releases out the door very_soon() 15:06:20 I mean, we should finish Victoria ASAP :) 15:06:26 exactly 15:06:55 I guess we can move on. Do we feel we should worry about subteam status reports this week at all since the important item is likely review priorites? 15:07:30 not a lot of new information, as it seems 15:07:35 Did someone replace the coffee with Decaf? 15:07:46 yes, me 15:07:55 at this point, the only thing if it hasn't been done, is maybe indicate which things will not be done in victoria? 15:08:14 rloo: I think the process is under way on the whiteboard 15:08:21 ++ 15:08:50 then onwards to priorities? 15:09:46 #topic review subteam status reports 15:10:41 arne_wiebalck: looks like your recent testing revealed bugs around raid/uefi/partitions. Seems like they are all in flight? 15:10:52 yes 15:11:04 I'm trying to fix these asap. 15:11:20 One is a pretty big one :-S 15:11:23 arne_wiebalck: lets get the patches on the review priorites 15:11:24 ugh fun 15:11:43 TheJulia: ++, thanks 15:12:14 Speaking of which, looks like the whiteboard is basically otherwise up to date feature/code wise. 15:12:17 So onward! 15:12:20 #topic Deciding on priorities for the coming week 15:12:25 I'm very close to fixing the race in hash ring. I still have no clear understanding of how it happens, but I have a patch. 15:12:32 #link https://etherpad.opendev.org/p/IronicWhiteBoard 15:12:35 dtantsur: oh my :( 15:12:40 dtantsur: and thanks! 15:13:01 Starting at line 128 15:13:07 * TheJulia removes merged items 15:13:07 TheJulia may u pay sometime help me fix up the ramdisk deploy interface issue this week 15:13:29 or anyone else has experience with ramdisk deploy interface will be welcome. 15:14:10 I am on a very closed deadline, the it stucks on last step of ramdisk deploy now. 15:15:16 QianbiaoNG: Lets discuss during open discussion 15:15:26 Looks like the items to add are at line 185 15:15:29 ok sure. 15:15:44 They look all reasonable to add to me. Thoughts? 15:16:07 Dmitry Tantsur proposed openstack/ironic master: Fix a race condition in the hash ring code https://review.opendev.org/754362 15:16:08 TheJulia: ^^^ 15:16:24 this also fixes the multinode job (although the workaround will likely merge quicker) 15:18:07 Awesome 15:18:28 folks, please don't recheck anything at least either of the two fixes merges 15:18:33 (see the whiteboard for the status) 15:18:43 Okay, I've updated the list of patches starting at line 125. I'm going to remove specs at this time, we can revisit them in the new cycle 15:18:54 ++ let's concentrate and what should+can merge for V 15:18:58 s/and/on/ 15:19:22 ++ 15:19:50 Are we good to proceed? 15:20:14 +1 15:20:19 ++ 15:20:51 Since we have no explicit discussion topics, next is Baremetal SIG 15:20:55 #topic Baremetal SIG 15:21:03 First meeting planned for next Tue, Oct 6 at 2pm UTC 15:21:10 Will send a mail 15:21:28 Awesome 15:21:37 Anything else? 15:21:42 I also added the SIG's etherpad to the wiki as a reference and for the agenda 15:21:57 I think that's it. 15:23:19 Okay! 15:23:42 Well, looking at the RFE's it looks like Iury didn't have a chance to get back to them and I believe today is a holiday so we can move directly to Open Discussion 15:23:56 #topic Open Discussion 15:24:24 newbee question, what is SIG. i did not google anything revelant 15:24:33 Special Interest Group 15:24:39 aha, ok 15:25:01 QianbiaoNG: https://wiki.openstack.org/wiki/OpenStack_SIGs 15:25:31 dtantsur thanks. 15:25:48 Hello, folks, anyone has experience with ramdisk deploy interface ever. 15:26:20 i need some help with building ramdisk image. story is here: https://storyboard.openstack.org/#!/story/2008126 15:26:26 QianbiaoNG: what was the link you pasted earlier to what your encountering? Last I remember you were encountering issues with your ramdisk understanding what to do once it started booting 15:26:52 yes, the OS fails to boot correctly. 15:27:09 ends with "Warning: /dev/ram0 does not exist" 15:27:55 TheJulia i record all details there https://storyboard.openstack.org/#!/story/2008126 15:28:26 yeah, and that has to do with the contents attached. Reading you got the inspection ramdisk to boot via the ramdisk deploy interface? 15:28:40 Yes 15:28:48 inspect ramdisk works good. 15:28:59 and dib basically hasn't moved afaik on the bug with the ramdisk-image-create issue 15:29:26 is ramdisk-image-create still supported or used? 15:29:30 And i try to add "ramdisk" dib element, but it can build image succefully. 15:29:52 wait, adding the ramdisk element now works? 15:29:57 or does it still fail? 15:30:05 TheJulia dtantsur i remember one of you two told me that ramdisk-image-create is deprecated. 15:30:19 TheJulia ramdisk element has bugs i think 15:30:27 At least we no longer use it, I cannot say for the DIB team 15:30:32 Checking here: https://bugs.launchpad.net/diskimage-builder/+bug/1815954 15:30:33 Launchpad bug 1815954 in diskimage-builder "_BASE_ELEMENT_DIR: unbound variable" [Undecided,New] 15:30:52 I comment at last 15:31:10 seems like the right thing to do is to fix dib 15:31:27 It seems like that either way, when QianbiaoNG figures it out, we should probably document a working process for getting a ramdisk built that can run under ramdisk deploy driver as well. Maybe even add it as a feature to ipa-builder? 15:31:34 and then we need to document how to build suitable ramdisks 15:31:39 right :) 15:31:46 well, IPA-builder builds IPA. this is not IPA. 15:31:59 maybe we need a CI job on DIB that checks our procedure? 15:31:59 JayF yes, that is exactly what i want. 15:32:27 we may fix ramdisk issue on dib, and document it. 15:32:27 oh, I guess it's obvious, it's even in the name, that it's probably not a suitable use case for that repo :) 15:34:11 I am +2 for fix dib and hopefully have a job that double checks somewhere that it is working as expected. I think the base condundrum is that it was broken and nothing caught it. Then documentation on ways to build ramdisks 15:34:34 Maybe add ramdisk driver to our CI, including a DIB build for a ramdisk? 15:35:03 TheJulia I will be on this channel all the time this week 15:35:03 IDK if I love the idea of *our* ci breaking if DIB breaks ramdisks 15:35:11 maybe they'd take CI to validate ramdisk building... 15:35:13 JayF: it presently runs an iso image 15:35:28 I believe the ramdisk support in DIB has only ever been used by ironic 15:35:28 if u got any idea about where to start fix dib, I can start the fix job. 15:35:30 I'd hope they would but only time will tell 15:36:03 QianbiaoNG: I think we would all be starting at the same place, but I'd likely start backtracking what happened to BASE_ELEMENT_DIR 15:36:44 I guess the major issue I have with trying to run dib actively in our CI is caches being out of date and our CI being dead as a result 15:37:01 can be non-voting 15:37:04 yeah 15:37:24 Another, minor topic for open discussion -- https://review.opendev.org/#/c/746774/ removal of select from IPA heartbeating appears to pass CI in it's current form, after all the CI fixes hit. Would appreciate reviews on it now. 15:37:24 patch 746774 - ironic-python-agent - Simplify heartbeating by removing use of select() - 28 patch sets 15:37:25 I fix the BASE_ELEMENT_DIR by adding it manually, and it got more issue after that :) 15:38:04 Hey, on a plus side, tempest should work with an non-iso ramdisk :) 15:38:10 I think.... 15:38:14 JayF: awesome! 15:38:22 QianbiaoNG: have you posted that as a patch? 15:38:32 * JayF is still suspicious it was a lucky pass, but will recheck if no other feedback 15:38:43 QianbiaoNG Not yet, i set the env manually. not by update code. 15:38:49 TheJulia ^^ 15:38:52 troubleshooting what you suspect to be a random failure while other random failures exist is the hurt zone :) 15:39:26 :) 15:40:56 It seems ramdisk deploy interface is rarely used. 15:40:58 QianbiaoNG: ahh, maybe go back a cycle or two in the code base and see if the paramter was previously being set? 15:41:24 TheJulia sure i will start with that 15:41:44 QianbiaoNG: It has a very specific user scenario that most people don't need nor want 15:42:05 Hmm. That's not a bad idea. Install progressively older diskimage-builder versions, then see the last one it worked at 15:42:13 and you can potentially suss out the problem from there? 15:42:24 maybe even do a `git bisect` from there? 15:43:02 Yeah, it obviously existed in the past, just need to figure out where it broke 15:44:27 great idea 15:45:04 binary search works in this situatino. :) 15:45:15 Will choose a pretty old version to start. 15:45:31 I'd go back ?2? cycles 15:45:36 I bet that will work, tbh 15:45:48 back to before we really started pushing ipa-b 15:47:06 TheJulia the default version of dib in ubuntu18 does not work too, it's 2.28 IIRC 15:47:28 it's 3.3.1 currently. 15:47:31 dtantsur: In regards to your comment on the security spec, its my understanding that we want a generic security-interface before implementing any Keylime specifics. Is this not correct? There is a Keylime spec in the works at the moment too. 15:48:16 QianbiaoNG: hmm, 3.x was likely for python3, so maybe back even further :( 15:49:06 TheJulia sure, i will test and keep you update 15:49:14 Seems like we're done with the meeting today... 15:49:19 Thanks everyone! 15:49:38 Thanks everyone, have a good day. 15:49:57 Thanks TheJulia o/ 15:50:36 #endmeeting