queensly[m] | Good morning Ironic | 08:03 |
---|---|---|
queensly[m] | Have a great week ahead! | 08:03 |
rpittau | good morning ironic! o/ | 08:09 |
abongale | good morning ironic! o/ | 08:20 |
opendevreview | Alex Welsh proposed openstack/bifrost master: Fix deployment image checksum validation https://review.opendev.org/c/openstack/bifrost/+/953080 | 09:27 |
dtantsur | morning folks, I'm back and without a lot of IRC history | 11:48 |
TheJulia | dtantsur: Last week was relatively quiet | 12:59 |
dtantsur | that's a rare luck! | 13:32 |
TheJulia | dtantsur: Well, Thursday was a holiday in the US, and I know some folks took it off | 13:41 |
dtantsur | right, same in many parts of Europe | 13:41 |
TheJulia | And then why work friday when you have a thursday off ;) | 13:41 |
dtantsur | ++ | 13:41 |
opendevreview | Takashi Kajinami proposed openstack/sushy-tools master: Drop test logic for Python < 3.8 https://review.opendev.org/c/openstack/sushy-tools/+/953103 | 13:42 |
opendevreview | Merged openstack/sushy-tools master: Drop test logic for Python < 3.8 https://review.opendev.org/c/openstack/sushy-tools/+/953103 | 14:22 |
cardoe | So dtantsur I'm thinking of adding a top level sushy.quirks. It'll be a class with members that are probably mostly booleans but they can be named something long and crazy and unique. | 14:26 |
cardoe | VirtualMediaRequiresAuth: bool = Field(default=False) | 14:27 |
cardoe | rough pseudo code | 14:27 |
cardoe | The quirks object can be passed to a session / connection so that operations can reference it to make choices. | 14:27 |
dtantsur | mmm, yeah, this sounds pretty good. even with auto-detection, we can make quirks somewhat more structured this way. | 14:28 |
cardoe | We can eventually add stuff like sushy.quirks.Dell as an instance of it with some defaults for say all Dell devices. | 14:28 |
cardoe | But anyone can instantiate one with whatever quirks they want it to do. | 14:29 |
dtantsur | +++ | 14:30 |
cardoe | Obviously that quirks object can add members as needed but we'll require they always have a default value. | 14:30 |
cardoe | I think there was a bug report about making a spec for this? | 14:30 |
guilhermesp | good morning ironic team! wondering if you can give me a quick hand here .. ive been fighting to get a software raid1 for 2x960 disks and RAID0 for 2x3TB disks... the layout is pretty weird after following the documentation https://docs.openstack.org/ironic/latest/admin/raid.html ... that is the final result https://paste.openstack.org/raw/b6zA4cFE5pysKOwUSR9Z/ which i would expect nvme0 and nvme1 for OS and nvme2 | 14:31 |
guilhermesp | and nvme3 for data disk | 14:31 |
dtantsur | cardoe: it does not sound spec-worthy to me, unless you're also planning Ironic API additions | 14:31 |
cardoe | I'm struggling to find it but I can write up the spec first or I can craft a rough POC to give an idea and then write the spec and then finalize the patch. | 14:31 |
dtantsur | the sushy part can be done without IMO | 14:31 |
cardoe | Definitely don't wanna make Ironic API changes for it for at all possible to avoid. | 14:31 |
cardoe | okay I'll do the POC | 14:31 |
dtantsur | yep, please do | 14:33 |
dtantsur | guilhermesp: I need to remind myself how software RAID works.. | 14:33 |
dtantsur | https://docs.openstack.org/ironic/latest/admin/raid.html mentions " software RAID which always spans all disks" which makes me believe what you're looking for is not supported | 14:34 |
dtantsur | ... and then we show an example with physical_disks :D | 14:35 |
guilhermesp | no worries! appreciate any clues... i tried defining physical disks as well. it works well for RAID1 but when it comes to RAID0, it tries partitioning the disks and ends up breaking like https://gist.githubusercontent.com/guilhermesteinmuller/242dc443de068f918ed267e65e5b6002/raw/76e073f51a43530e1e077e50297c1502f9985bac/gistfile1.txt | 14:35 |
dtantsur | guilhermesp: have you tried adding physical_disks? our documentation is inconclusive on whether it will work | 14:35 |
dtantsur | got it | 14:35 |
dtantsur | I guess you're hitting a bug, let me try to understand why it happens | 14:36 |
guilhermesp | appreciate that dtantsur ! | 14:37 |
guilhermesp | i also faced another bug which is already fixed which makes me to define volume_name where docs says is not supported for software raid; https://bugs.launchpad.net/ironic-python-agent/+bug/2073406 | 14:37 |
dtantsur | I see. Are you using an older release? | 14:38 |
guilhermesp | im using 2024.1 but we havent backported that yet in atmosphere but we are working on getting that in | 14:38 |
opendevreview | cid proposed openstack/ironic master: Migrate conductor threads to a thread pool https://review.opendev.org/c/openstack/ironic/+/952939 | 14:41 |
dtantsur | guilhermesp: for the future: I highly recommend running IPA in debug mode | 14:44 |
guilhermesp | thaats noted. If it helps you i can try enabling that and getting some more info | 14:45 |
dtantsur | guilhermesp: I think I get a rough idea of where things went wrong. If I prepare a draft patch, will you be able to test it? | 14:46 |
guilhermesp | yep i think we can easily backport that and test | 14:46 |
guilhermesp | it will probably need a new image on our side for amotpshere which would need us to patch a new release but i will sync that with mnaser :) | 14:48 |
dtantsur | guilhermesp: meanwhile, could you file a bug about the physical_disks behavior? | 14:48 |
guilhermesp | sure | 14:49 |
guilhermesp | that would be on https://bugs.launchpad.net/ironic-python-agent i believe? | 14:50 |
cid | ++ | 14:50 |
dtantsur | yes | 14:51 |
TheJulia | Software raid has a lot of very specific quarks | 14:54 |
TheJulia | Which I think are noted in IPA docs not ironic admin raid docs. | 14:55 |
opendevreview | Dmitry Tantsur proposed openstack/ironic-python-agent master: WIP software RAID on different devices https://review.opendev.org/c/openstack/ironic-python-agent/+/953122 | 14:55 |
dtantsur | guilhermesp: try this ^^^. if it does not work, please collect logs with DEBUG on. | 14:55 |
TheJulia | would be interesting to see the results of that :) | 14:56 |
TheJulia | #startmeeting ironic | 15:00 |
opendevmeet | Meeting started Mon Jun 23 15:00:20 2025 UTC and is due to finish in 60 minutes. The chair is TheJulia. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
opendevmeet | The meeting name has been set to 'ironic' | 15:00 |
guilhermesp | thank you very much!!! btw https://bugs.launchpad.net/ironic-python-agent/+bug/2115211 which i believe this patch will be associated | 15:00 |
guilhermesp | have a good meeting ! | 15:00 |
TheJulia | Greetings folks! Who is chairing our meeting today? | 15:00 |
* TheJulia suspects she is, but also wonders if we even have quorum today | 15:01 | |
alegacy | o/ | 15:01 |
TheJulia | o/ alegacy | 15:01 |
* TheJulia begins to suspect meeting.quorum() is returning false this week | 15:02 | |
cid | I think, no quorum too :D | 15:04 |
rpittau | o/ | 15:05 |
TheJulia | Folks, given we don't have a quorum this week, please be mindful of reviews. Specifically those tagged ironic-week-prio. We have a couple in ironic-specs which could use reviews. We'll meet again next week. | 15:05 |
queensly[m] | o/ | 15:05 |
TheJulia | As a heads up, the week of July 7th, I'll be working somewhat odd hours, so I likely won't be able to be around during the weekly meeting. | 15:05 |
TheJulia | Thanks everyone, have a great week! | 15:06 |
TheJulia | #endmeeting | 15:06 |
opendevmeet | Meeting ended Mon Jun 23 15:06:08 2025 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:06 |
opendevmeet | Minutes: https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-06-23-15.00.html | 15:06 |
opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-06-23-15.00.txt | 15:06 |
opendevmeet | Log: https://meetings.opendev.org/meetings/ironic/2025/ironic.2025-06-23-15.00.log.html | 15:06 |
* TheJulia brews more coffee | 15:06 | |
cid | Talking about reviews... something tells me this change may or may not pass CI this time around. Is that all the thread work required? Or is there something I'm missing? | 15:06 |
cid | This >> https://review.opendev.org/c/openstack/ironic/+/952939 | 15:06 |
dtantsur | wow, I got distracted for a few minutes, and the meeting is over :D | 15:06 |
TheJulia | cid: oh yeah, I'd suspect that is going to take a bit more work before we can do that specifically | 15:07 |
dtantsur | cid: I don't know how the CI will feel about launching 300 threads or whatever we try to do | 15:07 |
dtantsur | see my calculations on the etherpad | 15:07 |
clif | I too missed the meeting, but I'm around o/ | 15:07 |
TheJulia | cid: downstream, last week, I created items to hopefully get us to try and benchmark *before* and *after* that change specifically to try and isolate that and provide guidance around the actual thread settings | 15:08 |
TheJulia | yeah, 300 threads is a bit much | 15:08 |
TheJulia | we might OOM stuff in fun ways ;) | 15:08 |
dtantsur | I expect Linux to be smart about stack size nowadays, but yeah | 15:09 |
TheJulia | Yeah, its a huge behavior change, we need to do some measurements and move very carefully. I *also* suspect because we're changing __init__.py, other things will break without additional work | 15:10 |
TheJulia | (but we do need to start figuring it out) | 15:10 |
cid | We do. 300 does sounds like a lot. | 15:10 |
cid | I just queue the change in the priority list and expect those feedback | 15:10 |
opendevreview | Takashi Kajinami proposed openstack/sushy master: Restore flake8-import-order https://review.opendev.org/c/openstack/sushy/+/953124 | 15:11 |
* cid *I'll just queue the change in the priority list and expect those feedback | 15:11 | |
TheJulia | I've not looked at the etherpad in like two weeks | 15:11 |
opendevreview | Takashi Kajinami proposed openstack/sushy-tools master: Switch to pre-commit https://review.opendev.org/c/openstack/sushy-tools/+/953126 | 15:14 |
TheJulia | That being said, I'd expect a month or so of us toying with it and thinking "is this the right path, are these the right settings" to move forward iwth | 15:17 |
cid | Oh, alright. Will that be happening in the CI or by downstreams? | 15:19 |
TheJulia | I think it will need to likely be a blend, the biggest question which comes to mind is how does the profile change and we can't really do that entirely in CI. | 15:25 |
opendevreview | Dmitry Tantsur proposed openstack/ironic-python-agent master: WIP software RAID on different devices https://review.opendev.org/c/openstack/ironic-python-agent/+/953122 | 15:26 |
TheJulia | last call for https://review.opendev.org/c/openstack/ironic-specs/+/945642/7..9 | 18:36 |
TheJulia | err | 18:36 |
TheJulia | https://review.opendev.org/c/openstack/ironic-specs/+/945642/ | 18:36 |
TheJulia | alegacy: o/ you around? | 18:55 |
TheJulia | doh | 18:58 |
TheJulia | alegacy_: o/ | 18:58 |
TheJulia | alegacy_: I left a +2 on https://review.opendev.org/c/openstack/ironic-specs/+/952126, I suspect you might want to discuss one or two of the comments, its all good though. | 19:08 |
TheJulia | Of note, I +2'ed kaifeng's networking spec *as well*. I'm ont a huge fan of it, but it is rather simple of an approach and would align with neutron usage patterns | 19:26 |
alegacy_ | Thanks TheJulia... will take a look at your comments and update the spec accordingly! | 19:30 |
TheJulia | alegacy: if you want to discuss the PXE stuff, just let me know, I'm about to start sprint planning with my team so... it might be a little while but also have time tomorrow if that helps | 20:14 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!