| opendevreview | Merged openstack/ironic master: Fix DB/Lock session handling issues https://review.opendev.org/c/openstack/ironic/+/881765 | 01:34 |
|---|---|---|
| opendevreview | Merged openstack/ironic-python-agent master: Disable MD5 image checksums https://review.opendev.org/c/openstack/ironic-python-agent/+/865190 | 06:41 |
| opendevreview | Verification of a change to openstack/ironic-python-agent stable/zed failed: Fix UTF-16 result handling for efibootmgr https://review.opendev.org/c/openstack/ironic-python-agent/+/881658 | 06:42 |
| opendevreview | Merged openstack/ironic-python-agent stable/yoga: Fix UTF-16 result handling for efibootmgr https://review.opendev.org/c/openstack/ironic-python-agent/+/881659 | 07:07 |
| opendevreview | Merged openstack/ironic-python-agent stable/xena: Fix UTF-16 result handling for efibootmgr https://review.opendev.org/c/openstack/ironic-python-agent/+/881762 | 07:07 |
| opendevreview | Merged openstack/ironic-python-agent stable/wallaby: Fix UTF-16 result handling for efibootmgr https://review.opendev.org/c/openstack/ironic-python-agent/+/881763 | 07:07 |
| rpittau | good morning ironic! o/ | 07:15 |
| opendevreview | OpenStack Release Bot proposed openstack/ironic bugfix/18.0: Update .gitreview for bugfix/18.0 https://review.opendev.org/c/openstack/ironic/+/881942 | 07:21 |
| opendevreview | Steven Relf proposed openstack/networking-generic-switch master: fixes #2010720. This ensures that if the driver has an empty plug/unplug to bond interface function, the code will continue down the old path prior to https://github.com/openstack/networking-generic-switch/commit/aea67080442272e64238fda840bcdeeb5db9ff94 https://review.opendev.org/c/openstack/networking-generic-switch/+/881945 | 08:36 |
| opendevreview | Steven Relf proposed openstack/networking-generic-switch master: fixes #2010720. This ensures that if the driver has an empty plug/unplug to bond interface function, the code will continue down the old path prior to https://github.com/openstack/networking-generic-switch/commit/aea67080442272e64238fda840bcdeeb5db9ff94 https://review.opendev.org/c/openstack/networking-generic-switch/+/881946 | 08:43 |
| dtantsur | JayF, TheJulia, we use WAL with SQLite in Metal3 already, so unless you want to break it, we're good :) | 09:14 |
| dtantsur | https://github.com/metal3-io/ironic-image/blob/main/Dockerfile#L60-L61 | 09:15 |
| opendevreview | Merged openstack/ironic-python-agent stable/2023.1: Fix UTF-16 result handling for efibootmgr https://review.opendev.org/c/openstack/ironic-python-agent/+/881657 | 09:32 |
| iurygregory | good morning Ironic | 11:09 |
| opendevreview | Steven Relf proposed openstack/networking-generic-switch master: fixes #2010720. This ensures that if the driver has an empty plug/unplug to bond interface function, the code will continue down the old path prior to https://github.com/openstack/networking-generic-switch/commit/aea67080442272e64238fda840bcdeeb5db9ff94 https://review.opendev.org/c/openstack/networking-generic-switch/+/881958 | 11:09 |
| opendevreview | Steven Relf proposed openstack/networking-generic-switch master: Ensures that if a switch plugin has an empty plug/unplug bond to network function that the old code path of plug/unplug port to network is followed, otherwise this can result in a no-op happening https://review.opendev.org/c/openstack/networking-generic-switch/+/881959 | 11:17 |
| kaloyank | morning ironic /o | 11:20 |
| opendevreview | Jacob Anders proposed openstack/ironic master: Handle MissingAttributeError when using OOB inspections to fetch MACs https://review.opendev.org/c/openstack/ironic/+/880575 | 12:09 |
| *** sfinucan is now known as stephenfin | 13:04 | |
| TheJulia | dtantsur: ahh, well, the engine actually has to know to use it in WAL mode \o/ | 13:49 |
| dtantsur | ohrly? | 13:49 |
| dtantsur | so we haven't ACTUALLY used WAL all this time? | 13:49 |
| dtantsur | also, good morning TheJulia | 13:49 |
| TheJulia | dtantsur: I believe that is the case | 13:49 |
| dtantsur | okay, that could explain certain results we were getting back in the days.. | 13:50 |
| TheJulia | we went from hundreds of errors to no errors at all with debug logging running | 13:50 |
| TheJulia | (which *really* slowed things down) | 13:50 |
| TheJulia | Good morning | 13:50 |
| * TheJulia tries to wake up | 13:50 | |
| JayF | dtantsur: I can confirm that does ~nothing. PRAGMA in sqlite is a client setting per connection | 14:21 |
| dtantsur | JayF: I think WAL is an exception of this rule and is set on database creation time. | 14:21 |
| dtantsur | maybe it also needs to be in the session, dunno | 14:21 |
| JayF | It 100% must be done in the session | 14:21 |
| dtantsur | okay, so their documentation is lacking then.. sweet. I wonder why they cannot just read the format the database was created in. | 14:22 |
| JayF | What's the full story on https://review.opendev.org/c/openstack/ironic-lib/+/879211 | 14:47 |
| JayF | Like, the code seems fine, I just think it's weird to approve a library change w/o any idea where/how it's going to be used | 14:47 |
| opendevreview | Julia Kreger proposed openstack/ironic-python-agent master: Fix Bandit errors https://review.opendev.org/c/openstack/ironic-python-agent/+/879912 | 14:48 |
| opendevreview | Jay Faulkner proposed openstack/ironic master: Make metal3 job voting https://review.opendev.org/c/openstack/ironic/+/882073 | 15:15 |
| rpittau | oh wow | 15:31 |
| TheJulia | JayF: ++ | 15:31 |
| rpittau | thanks, that means a lot :) | 15:32 |
| TheJulia | JayF: presently you cannot run ironic conductors on anything aside from the default port | 15:32 |
| JayF | rpittau: making the job voting? | 15:32 |
| rpittau | yeah | 15:32 |
| JayF | rpittau: I view it more as, thanks for making a job to test sqlite support :D | 15:32 |
| rpittau | :D | 15:33 |
| TheJulia | or as configured from the remote side, so basically you can't run multiple conductors on the same machine today | 15:33 |
| JayF | rpittau: also, we'd co-gate with any similar project in openstack; the fact it's CNCF is immaterial to me | 15:33 |
| TheJulia | at least, with jsonrpc | 15:33 |
| JayF | TheJulia: ack; that was my assumption but wanted to validate | 15:33 |
| rpittau | well I see this as a huge step in the collaboration/integration with metal3, I'm glad we're here :) | 15:34 |
| JayF | all I did was two lines in a yaml file | 15:35 |
| JayF | TheJulia fixed sqlite and you wrote the CI job | 15:35 |
| TheJulia | sqlite+sqlalchemy == the bane of TheJulia | 15:37 |
| * TheJulia prefers lolcats | 15:37 | |
| zigo | Heya! FYI, I had my first baremetal node discovered by IPA and reported to ironic-inspector today! :) | 15:42 |
| dtantsur | \o/ | 15:42 |
| zigo | I used a custom live-image built with live-build (Debian based). | 15:43 |
| zigo | Though one question: is Ironic rewriting the grub config dynamically? | 15:43 |
| zigo | (it felt like it does...) | 15:43 |
| TheJulia | zigo: congrats | 15:43 |
| TheJulia | zigo: the overall code path *really* depends on the contents of the image | 15:43 |
| zigo | One thing we want to do, is to setup special networking using FRR bgp-to-the-host when nodes are getting installed. Is there ways to hook some kind of scripts during the install process or what? | 15:45 |
| JayF | zigo: are you talking about post-install inside-image configuration? | 15:46 |
| JayF | zigo: or something else? | 15:46 |
| zigo | JayF: Yeah, something like that. | 15:46 |
| zigo | Like, once the image is burnt on the physical disk, then we need some kind of custom postinst of our own... | 15:46 |
| JayF | zigo: yeah, we recommend use of cloud-init or similar technology for that. We generally do not do post-install configuration beyond what's absolutely neccessary (e.g. bootloader configs for partition images) | 15:46 |
| JayF | zigo: it's possible you could use Ironic deploy steps to implement that; but I'd consider it a mild abuse of the feature. That being said; you don't have to tell me if you abuse the feature ;) | 15:47 |
| zigo | JayF: The problem with cloud-init and friends, is that this means an L2 connectivity with DHCP. We want to avoid it... | 15:47 |
| zigo | Oh ok. | 15:47 |
| JayF | zigo: you can use cloud init to, for instance, run a script to configure whatever networking you want | 15:47 |
| JayF | zigo: I'm suggesting, in general, using an on-first-boot mechanism embedded in the image to do that setup on first boot | 15:48 |
| zigo | JayF: Yeah, but cloud-init wont be able to fetch the metadata, and find the IP address to configure, because we don't want to do any L2 connectivity at all, and that's what I'm worrying about. | 15:48 |
| JayF | ah, I didn't realize you were operating without config drive | 15:48 |
| zigo | Oh, config drive ... | 15:48 |
| JayF | yep :D | 15:49 |
| zigo | I always wonder, how does it work with Ironic then? | 15:49 |
| JayF | we make a partition and plunk it in there | 15:49 |
| zigo | Does it use the virtual-media of the BMC ? | 15:49 |
| zigo | Oh ok. | 15:49 |
| zigo | Great! So I have a solution then! :) | 15:49 |
| JayF | I wrote the patch a decade-ish ago to cloud-init to be willing to read config data off a physical disk parition :D | 15:49 |
| * JayF not sure if he ended up being the one to upstream that | 15:49 | |
| JayF | but it was literally just filtering out non-virtual disks. silly cloud-init :P | 15:49 |
| TheJulia | zigo: if you have a UEFI ready image with a UEFI partition and all, we hunt for the bootloader record csv file and load that into NVRAM | 15:51 |
| TheJulia | Partition images are where we end up having to do a lot to make the bootloader happy/work | 15:51 |
| zigo | My next step is probably advanced networking now... :P | 16:04 |
| dtantsur | I've tried using virtual media for a configdrive. But it conflicts with using virtual media for IPA itself. | 16:12 |
| rpittau | good night! o/ | 16:22 |
| JayF | o/ | 16:27 |
| JayF | FYI: I'm leaving after the 1800 UTC technical committee meeting. If you need something from me today, ask soon :) | 16:46 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic-python-agent master: Add network interface speed to the inventory https://review.opendev.org/c/openstack/ironic-python-agent/+/879156 | 16:47 |
| dtantsur | TheJulia: I'm putting the MD5 deprecation on the tomorrow's metal3 meeting agenda. If you can attend - awesome, otherwise I can cover it. | 16:48 |
| TheJulia | dtantsur: what time is it? | 16:51 |
| JayF | dtantsur: https://review.opendev.org/c/openstack/ironic/+/882073 wanted to make sure you had visibility on this | 16:52 |
| JayF | dtantsur: to be appropriately excited or horrified depending on your opinion ;) | 16:52 |
| dtantsur | TheJulia: 2pm UTC | 16:52 |
| dtantsur | JayF: I'm happy to see it. Just the usual caveats of co-gating: sometimes we'll have to make it temporary non-voting. | 16:53 |
| JayF | and that'll be fine. I think we're all in favor of practicality :D | 16:53 |
| JayF | I just don't want someone to be able to ignore the job and land something; we have a large number of NV jobs; but I think the sqlite/autocommit interactions are too easy to break to be worth leaving it NV | 16:54 |
| dtantsur | JayF: btw, have you seen https://groups.google.com/g/metal3-dev/c/5lDaVhwajYQ ? | 16:55 |
| JayF | I do not follow metal3-dev | 16:55 |
| JayF | perhaps I should? | 16:55 |
| dtantsur | JayF: it's very low traffic, you probably should | 16:55 |
| JayF | ugh | 16:56 |
| JayF | why are metal3 docs locked! | 16:56 |
| dtantsur | JayF: you need to subscribe to the ML with your google account | 16:56 |
| JayF | everytime I see a google docs request access screen after clicking a link on a public OSS list, I want to chuck my laptop across the room :( | 16:56 |
| dtantsur | yeah, dunno. anti-spam measure, I assume. | 16:56 |
| JayF | eh, a gate is a gate. we need to fling things open and let people in. at least to view/comment if not edit | 16:56 |
| dtantsur | anyway, the keep part is the PTG doodle: https://doodle.com/meeting/participate/id/eV7WnQoa | 16:59 |
| JayF | dtantsur: looking at the PTG doc; I'm not sure I would provide value if I attended. Do you disagree? I don't mind attending, and love the warm-fuzzies of cross-attending-PTGs, but don't wanna just be a spectator | 16:59 |
| dtantsur | JayF: it's quite possible. I'm just making sure you don't miss the opportunity. | 16:59 |
| JayF | I am 100000% glad you pointed it out to me. | 17:00 |
| JayF | The temptation to attend even just to watch is high, but I will likely leave the doodle alone and attend opportunistically if it fits. | 17:00 |
| JayF | If you all find a topic you need my brain for (maybe something IPA?) feel free to specifically ask | 17:00 |
| JayF | even if it's like, 2am PST I can make it work :D | 17:00 |
| dtantsur | Time-wise, it's going to be ~2pm UTC. The day is the question. | 17:00 |
| dtantsur | Also, noted. If I don't forget, I'll announce the date and time on our ML. | 17:01 |
| JayF | heh. I don't have meetings at 5am local time so it doesn't matter so much ;) | 17:01 |
| dtantsur | 5am, ugh | 17:01 |
| JayF | dtantsur: I'd very, very much appreciate that from the same warm-fuzzy standpoint lol | 17:01 |
| JayF | dtantsur: you know, for me that's less awful than for some. Most mornings I'm awake by 6, 6:15am no alarm | 17:01 |
| dtantsur | wow, a morning person! | 17:02 |
| dtantsur | Everything before ~ 7:30am I'm considering a resurrection, not awakening. | 17:02 |
| dtantsur | Time for a walk, talk to you tomorrow o/ | 17:05 |
| kaloyank | quick question: when a node times out a soft reboot, is there a hard reboot like there is for powering off? | 17:07 |
| dtantsur | kaloyank: when installing - yes. When you just use power API, I think no. We had to implement that in Metal3, so maybe your question is a sign for us to add this feature to Ironic.. | 17:10 |
| * dtantsur now leaves for real, will check the scrollback tomorrow | 17:10 | |
| JayF | dtantsur: my wife is a teacher; so I work 7a-4p local time (that works out to 1400-2300 UTC during DST; 1500-0000 UTC otherwise) | 17:27 |
| JayF | and my workday ends more or less when she's opening the garage door to come home :D | 17:27 |
| TheJulia | ... wait... resurection?! | 17:39 |
| TheJulia | :) | 17:39 |
| TheJulia | kaloyank: .... if *memory* serves, we just send the message, if the OS never responds we don't have a way to really know | 17:39 |
| TheJulia | kaloyank: sort of like with IPMI, we send the message, we have no idea if it was acted upon | 17:42 |
| TheJulia | you *can* send a hard reboot though | 17:42 |
| TheJulia | hard I think is we just turn the power off | 17:42 |
| kaloyank | TheJulia: Yeah, I found out the hard reboot but I'd rather not fall back to it. This spurs out two new questions: | 17:46 |
| kaloyank | 1. When I tried to issue a soft reboot on a node deployed by Nova, I got an error saying I can't reboot an active node. Is this expected? | 17:48 |
| kaloyank | I just forgot the second one, long day | 17:50 |
| TheJulia | kaloyank: no, that is not expected behavior | 18:00 |
| TheJulia | I'm curious where that would be generated from | 18:00 |
| kaloyank | hmm, give me a sec to dig some logs | 18:01 |
| NobodyCam | good morning ironic folks | 18:03 |
| TheJulia | good morning | 18:08 |
| kaloyank | TheJulia: here it is https://pastebin.com/qe3LV3nh | 18:08 |
| kaloyank | jftr, I'm running yoga | 18:13 |
| TheJulia | I'm not sure set provision state is right | 18:15 |
| TheJulia | I'll need to look | 18:15 |
| TheJulia | Hey, anyone want to workflow https://review.opendev.org/c/openstack/ironic/+/862832 ? | 18:21 |
| JayF | lookin | 18:22 |
| JayF | good thing too, I put in my daily upstream update we had landed that (I thought we had!) | 18:22 |
| TheJulia | heh | 18:22 |
| kaloyank | TheJulia: well, shame on me I guess | 18:49 |
| JayF | kaloyank: I think you want https://github.com/openstack/openstacksdk/blob/master/openstack/baremetal/v1/node.py#L636 | 18:51 |
| JayF | (that looked like openstacksdk?) | 18:51 |
| kaloyank | yes, exactly | 18:52 |
| opendevreview | Verification of a change to openstack/ironic master failed: Remove autocommit, again. https://review.opendev.org/c/openstack/ironic/+/862832 | 19:35 |
| opendevreview | Julia Kreger proposed openstack/ironic-tempest-plugin master: Clarify checksum for wholedisk images. https://review.opendev.org/c/openstack/ironic-tempest-plugin/+/882096 | 19:58 |
| opendevreview | Julia Kreger proposed openstack/ironic master: Change wholedisk image checksum to sha256 https://review.opendev.org/c/openstack/ironic/+/882097 | 20:03 |
| TheJulia | Do I get an award for breaking tempest? | 20:19 |
| stevebaker[m] | <<< C:\ >>> | 20:20 |
| stevebaker[m] | here you go | 20:21 |
| iurygregory | \( ﾟヮﾟ)/ | 20:33 |
| iurygregory | for you TheJulia =) | 20:33 |
| * TheJulia glares at stevebaker[m] ;) | 20:43 | |
| * stevebaker[m] whistles | 20:44 | |
| TheJulia | so looks like by default we lean towards sha256 with the ramdisk | 20:55 |
| TheJulia | on a forced raw image specifically | 20:55 |
| iurygregory | forum session accepted but I'm not traveling .-. yay? .-. | 22:13 |
| TheJulia | ... could someone else facilitate? | 22:17 |
| iurygregory | I'm going to check some of the prices to see | 22:17 |
| iurygregory | I'm going to check if I can use the voucher from the travel to the PTG that got canceled | 22:19 |
| * TheJulia wonders how my title is back to "Cloud Automation and Distribution Engineer" | 23:10 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!