| opendevreview | Michal Nasiadka proposed opendev/zuul-providers master: Add image builds and provider config for Debian Bookworm https://review.opendev.org/c/opendev/zuul-providers/+/949127 | 09:28 |
|---|---|---|
| opendevreview | Michal Nasiadka proposed opendev/zuul-providers master: Add bookworm/arm64 image builds https://review.opendev.org/c/opendev/zuul-providers/+/949482 | 10:18 |
| opendevreview | Michal Nasiadka proposed openstack/diskimage-builder master: bootloader: Fix searching for grub-mkconfig https://review.opendev.org/c/openstack/diskimage-builder/+/949493 | 11:51 |
| opendevreview | Michal Nasiadka proposed opendev/zuul-providers master: Add bookworm/arm64 image builds https://review.opendev.org/c/opendev/zuul-providers/+/949482 | 11:52 |
| mnasiadka | Hello, I don't think the zuul-providers jobs support a depends-on on DIB code - I'd be happy for somebody to have a look at https://review.opendev.org/c/opendev/zuul-providers/+/949482 - maybe I missed something essential and the DIB ,,bug'' is not the thing on which the debian arm64 builds are failing | 12:25 |
| *** iurygregory_ is now known as iurygregory | 13:08 | |
| corvus | no they don't.... but they use a role in zuul-jobs that installs dib, and we could update that to detect a depends-on to dib and install it from source instead | 13:24 |
| corvus | er, from git i mean | 13:24 |
| mnasiadka | right | 13:46 |
| mnasiadka | corvus: in the meantime Bookworm x86 is ready - https://review.opendev.org/c/opendev/zuul-providers/+/949127 | 13:47 |
| clarkb | mnasiadka: corvus: I think I found a git organization issue in that change ^ | 14:48 |
| clarkb | but otherwise it looks good to me | 14:48 |
| opendevreview | Michal Nasiadka proposed opendev/zuul-providers master: Add image builds and provider config for Debian Bookworm https://review.opendev.org/c/opendev/zuul-providers/+/949127 | 14:48 |
| opendevreview | Michal Nasiadka proposed zuul/zuul-jobs master: Add support for using depends-on in ensure-dib role https://review.opendev.org/c/zuul/zuul-jobs/+/949514 | 14:51 |
| opendevreview | Michal Nasiadka proposed opendev/zuul-providers master: Add bookworm/arm64 image builds https://review.opendev.org/c/opendev/zuul-providers/+/949482 | 14:52 |
| corvus | oh, well, if you're updating that change, can you drop the "normal" label? :) | 15:01 |
| opendevreview | Michal Nasiadka proposed zuul/zuul-jobs master: Add support for using depends-on in ensure-dib role https://review.opendev.org/c/zuul/zuul-jobs/+/949514 | 15:13 |
| mnasiadka | corvus: sorry, missed that, but I can drop all normal labels in a followup | 15:34 |
| mnasiadka | corvus: Did I get the path wrong in 949514? See https://zuul.opendev.org/t/opendev/build/65499ea1ec564b9cb9bd6f2ff467d6d9 (in the past there was some ,,debug'' information on cloned repo paths, now I don't really know where to double check) | 15:35 |
| clarkb | EMS sent email warning us that "All clients must support native OIDC via MSC3861" due to a migration of how authentication works. They are migrating us on May 19 | 15:42 |
| corvus | mnasiadka: that looks like the executor path being used on the worker node | 15:46 |
| fungi | i suspect the weechat matrix plugin i'm working is going to break in a week | 15:47 |
| fungi | s/working/using/ | 15:47 |
| corvus | mnasiadka: left a comment on https://review.opendev.org/c/zuul/zuul-jobs/+/949514 | 15:47 |
| clarkb | fungi: ya the other things would be gerritbot and the eavesdrop bot. I think the eavesdrop one is probably fine as it uses a well supported python lib iirc | 15:48 |
| clarkb | not sure about gerritbots haskell. We may need an image rebuild? no tristanC in here... | 15:48 |
| fungi | development seems to have picked back up in https://github.com/poljar/weechat-matrix-rs so maybe it's time to finally give it a try | 15:50 |
| opendevreview | Michal Nasiadka proposed zuul/zuul-jobs master: Add support for using depends-on in ensure-dib role https://review.opendev.org/c/zuul/zuul-jobs/+/949514 | 16:33 |
| mnasiadka | corvus: relative path fails as well - do I need to copy the repository dir to the worker? | 17:00 |
| corvus | no that should be automatic... the output section of https://zuul.opendev.org/t/opendev/build/e7a61e247b4e46f5a4e3a53518b5b0a3/console#0/3/11/ubuntu-noble-arm64 looks like it's there | 17:19 |
| corvus | mnasiadka: i guess try with {{ansible_user_dir}} -- though i don't know why relative wouldn't work there. perhaps something with the ansible pip module | 17:21 |
| JayF | Something seems to be very broken on the openstack.org profile editor? https://usercontent.irccloud-cdn.com/file/yIrj0c7J/image.png | 17:45 |
| clarkb | JayF: I'll share your screenshot with those managing that service | 17:46 |
| JayF | thank you | 17:46 |
| clarkb | JayF: they are able to reproduce and are opening a ticket. I'll let you know if there is anythign else to share. But my guess is they can figure it out from here since they can reproduce | 17:55 |
| JayF | Honestly, it's just a coincidence I bumped up against it | 17:55 |
| JayF | was looking for that text around what "affiliated" means | 17:55 |
| JayF | so it's OK if you don't wanna close the loop on this one, I'm not blocked or anything | 17:56 |
| clarkb | ack | 17:56 |
| opendevreview | Michal Nasiadka proposed zuul/zuul-jobs master: Add support for using depends-on in ensure-dib role https://review.opendev.org/c/zuul/zuul-jobs/+/949514 | 18:11 |
| clarkb | I'm looking at meeting agenda updatse. I think we can drop the ovh log upload errors if no one has noticed more of them. I'll update the mailman log rotation item with a link to fungi's change. Gerrit server swap can be cleaned up and replaced with Gerrit 3.11 upgrade planning. Anything happen last week that I should tack onto the agenda? or anything else? | 18:23 |
| fungi | nothing on my end. just trying to dig myself out of a travel hole for now | 18:30 |
| clarkb | yup I'm in the same boat | 18:31 |
| clarkb | but preparing for tomorrow's meeting is semi time sensitive so want ot make sure I don't drop the ball on that | 18:31 |
| fungi | admirable! | 18:31 |
| clarkb | I did manage to pick up my bike from the shop saturday before the rain arrived. And its been raining since otherwise I would mow because the spring weather means the grass grows at a rapid rate | 18:32 |
| mnasiadka | corvus: thanks, that worked | 18:41 |
| corvus | these should get us niz grafana dashboards: https://review.opendev.org/q/hashtag:opendev-niz+status:open | 18:43 |
| mnasiadka | Ok, anyway that bug in diskimage-builder is rather cosmetic - I'll need to have a look why on debian arm64 we're missing /boot/efi/EFI/debian | 18:50 |
| mnasiadka | If anybody has an idea - comments welcome in https://review.opendev.org/c/opendev/zuul-providers/+/949482 | 18:50 |
| mnasiadka | I'll have a look in the existing nodepool configs for images tomorrow | 18:50 |
| clarkb | I approved the grafyaml update and will recheck the new dashboards change once that lands | 18:52 |
| clarkb | but I don't expect problems and can approve the dashboard update once I can see the dashboards | 18:52 |
| opendevreview | Merged opendev/grafyaml master: Add regex support to template query https://review.opendev.org/c/opendev/grafyaml/+/949248 | 18:52 |
| clarkb | I guess I needt owait for the image to promote too but that won't take long | 18:53 |
| fungi | mnasiadka: typically, /boot/efi is mounted as a separate device so it can be fat32/vfat-formatted. maybe we're not including the additional device? | 18:56 |
| clarkb | iirc the arm images have a whole disk layout specification because they have to be efi/gpt and ya that | 18:57 |
| mnasiadka | actually we only pass the disk specification for Debian | 18:57 |
| mnasiadka | it works for x86, but doesn't for arm64 | 18:58 |
| mnasiadka | let me try removing the disk layout spec | 18:58 |
| clarkb | I think if you don't have a disk layout spec you get mbr which doesn't boot | 18:58 |
| clarkb | but its worth a shot | 18:58 |
| opendevreview | Michal Nasiadka proposed opendev/zuul-providers master: Add bookworm/arm64 image builds https://review.opendev.org/c/opendev/zuul-providers/+/949482 | 18:58 |
| mnasiadka | it works for Ubuntu without disk layout spec, block-device-efi should give us a ,,proper'' default layout, so let's try | 18:59 |
| clarkb | corvus: there is a bug in the grafana dashboards. The new template uses `title: 'Nodepool: %PROVIDER%'` and the nodepool templates due the same thing and you can't have those duplicatse. I can fix it if you like changing Nodepool to Zuul Launcher ? | 19:00 |
| clarkb | or I'm happy to review an update if you want to push one | 19:00 |
| corvus | ill fix | 19:02 |
| opendevreview | James E. Blair proposed openstack/project-config master: Add zuul-launcher dashboards https://review.opendev.org/c/openstack/project-config/+/949241 | 19:03 |
| opendevreview | Merged openstack/project-config master: Add zuul-launcher dashboards https://review.opendev.org/c/openstack/project-config/+/949241 | 19:57 |
| mnasiadka | Ok then, removing the storage layout helped | 20:01 |
| mnasiadka | I’ll shape up the Debian arm64 patch tomorrow | 20:01 |
| mnasiadka | (10pm here) | 20:01 |
| fungi | no need to feel compelled to be up so late for this! | 20:02 |
| clarkb | the zuul launcher graphs are now in grafana | 20:07 |
| corvus | a number of nodes are still in ready state due to being associated with requests which were later dequeued. a change merged last week to fix that, but left the requests orphaned. i will manually delete them from zk now. | 20:26 |
| fungi | thanks! | 20:27 |
| clarkb | ok after a first pass on the meeting agenda things trimmed way down. Let me know if I should add/edit/remove anything else | 20:29 |
| corvus | okay, those zombie nodes should be gone now | 20:36 |
| clarkb | the rackspace flex graph seems to reflect the change | 20:37 |
| clarkb | corvus: does that have to be done via the api? | 20:37 |
| clarkb | (just wondering what the mechanics are for doing that cleanup) | 20:37 |
| corvus | manual zk; api for that not implemented yet | 20:37 |
| clarkb | got it. Do you update the state to used so they get deleted automatically? | 20:38 |
| corvus | just deleted the requests and the launcher took care of the rest | 20:39 |
| clarkb | aha | 20:39 |
| corvus | i added a max-ready-age for a bunch of the labels last week | 20:39 |
| corvus | so they became unallocated ready nodes past their max-age and were auto-deleted | 20:40 |
| corvus | we can revert out the max-ready-age soon i think. i just added it to clean up mistakes like this | 20:40 |
| tristanC[m] | clarkb: I'm not sure to understand how to implement the matrix next-gen auth MSC3861 for the gerritbot. If I understand correctly, the access token must be refreshed (and the refresh token also rotate each time this happen), and the m.login.password is also not an option. | 20:41 |
| tristanC[m] | So it looks like that you would initialize both tokens first, and then the client must store the refreshed token locally? And if the state is lost, then you would have to recreate the tokens. It's a bit confusing, I'll have to re-read the spec. | 20:43 |
| clarkb | tristanC[m]: I'm not sure I was just following up on an email they sent which indicated clients may need to be updated. | 20:43 |
| clarkb | tristanC[m]: https://matrix.org/blog/2024/10/29/matrix-2.0-is-here/#2-next-generation-auth and https://element-hq.github.io/matrix-authentication-service/ are the two documents they linked to | 20:44 |
| clarkb | not sure if those are helpful in better undertsanding the mechanics | 20:44 |
| tristanC[m] | The service-account/bot usage doesn't seem well defined. Also the spec doesn't seem to be implemented outside of synapse, so it's difficult to test with the other servers like conduit. | 21:01 |
| corvus | i think i read something about matrix.org scheduling its cutover for april 7; if they did that, perhaps the matrix.org server could be used for testing | 21:02 |
| tristanC[m] | It would be good to know if you can still create long lived token without refresh for such account, or may I use the m.login.password legacy auth to ensure a valid refresh token? https://element-hq.github.io/matrix-authentication-service/topics/authorization.html#compatibility-sessions seems to say so, but I can't tell if it's compatible with "All clients must support native OIDC via MSC3861" | 21:12 |
| tristanC[m] | In anycase, it doesn't sound complicated to support for the gerritbot, it's just that I don't know what needs to be done:) On matrix.org, if you don't set refresh_token: true when creating a token, then you get the regular token, which I think is compatible with the next gen auth. | 21:17 |
| fungi | yeah, i think i had to do that for my nonstandard client | 21:18 |
| fungi | so maybe it won't break after all | 21:19 |
| clarkb | last call on meeting agenda updates. I'll get that sent out in the next 15-30 minutes | 22:13 |
| clarkb | going to be a few extra minutes. Fastmail logged me out due to inactivity despite me literally using the web client an hour ago | 22:58 |
| clarkb | looks like they also logged ou my mobile app | 22:58 |
| fungi | some service probably got restarted and cleared stored session ids | 23:00 |
| clarkb | it should be in your inboxes nowish | 23:04 |
| fungi | 'tis indeed | 23:07 |
| clarkb | of course as soon as I send that I see there is a new gitea release... I'll get a change p for that | 23:12 |
| opendevreview | Clark Boylan proposed opendev/system-config master: Update to gitea 1.23.8 https://review.opendev.org/c/opendev/system-config/+/949544 | 23:16 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
