| opendevreview | melanie witt proposed openstack/devstack master: Configure oslo.limit endpoint discovery for Nova https://review.opendev.org/c/openstack/devstack/+/961326 | 00:29 |
|---|---|---|
| *** mhen_ is now known as mhen | 01:51 | |
| opendevreview | Takashi Kajinami proposed openstack/devstack master: wip: Query endpoint id https://review.opendev.org/c/openstack/devstack/+/961328 | 02:27 |
| ykarel | frickler, hi when you get chance please check https://review.opendev.org/c/openstack/devstack/+/961222 | 03:22 |
| *** dmellad is now known as dmellado | 07:14 | |
| ykarel | @sean-k-mooney, can you please check/merge https://review.opendev.org/c/openstack/devstack/+/961222 | 11:19 |
| sean-k-mooney | i was just replying | 11:20 |
| sean-k-mooney | yes but givens fedora direction im not sure the centos 10 is going to be a good long term testign target upstream at least not for fips | 11:20 |
| sean-k-mooney | there was some dicussion in the tc channel about not wanting to provdier special fips images in ci | 11:21 |
| sean-k-mooney | so if centos stream looses the ablilty to enable fips post install then we may need to explore other distos to test fips as a result | 11:22 |
| sean-k-mooney | or test it only via third party ci | 11:22 |
| ykarel | sean-k-mooney, ok thx | 11:23 |
| ykarel | will atleast check this week enabling fips post boot with 10-stream | 11:24 |
| sean-k-mooney | i think if we can do it even if it requried rebooting the vm to do it it shoudl be fine | 11:24 |
| sean-k-mooney | i.e. if we have to unistall non fips package then reboot into it it woudl be ok | 11:25 |
| ykarel | fips role reboots even now | 11:25 |
| sean-k-mooney | ya so that pattern is already established | 11:25 |
| sean-k-mooney | what woudl be problematic is if it was not possibel to enabel at all after the inital install | 11:25 |
| ykarel | i started with https://review.opendev.org/q/topic:%22centos-10-stream%22 | 11:26 |
| ykarel | so will continue on that | 11:26 |
| ykarel | but later this week | 11:26 |
| sean-k-mooney | the concern is there are other fips equvelents in other juristiction and we dont nessisarly want to give FIPS special treatment but also dont want the foundation to have to maintain special imagfe for each | 11:26 |
| sean-k-mooney | coll ill try an take a look when you think they are getting ready | 11:28 |
| sean-k-mooney | are there other fips josb beyond neutron | 11:29 |
| sean-k-mooney | well centos jobs | 11:29 |
| sean-k-mooney | or is that whree most of the centos testing is done | 11:29 |
| ykarel | some projects seems to have it apart from neutron https://codesearch.opendev.org/?q=enable_fips%3A&i=nope&literal=nope&files=&excludeFiles=&repos= | 11:32 |
| ykarel | and all those are on 9-stream | 11:33 |
| sean-k-mooney | it looks like it really only neutron swift and tempest | 11:42 |
| sean-k-mooney | the reset are really just in supprot of those | 11:42 |
| sean-k-mooney | frickler: clarkb fungi is this statement still true? https://opendev.org/opendev/zuul-providers/src/branch/master/zuul.d/image-build-jobs.yaml#L387-L392 | 11:52 |
| sean-k-mooney | how would i check if the mirrors for debian trixie are now aviable? | 11:53 |
| sean-k-mooney | also i feel like this config moved? it uses to be in the project-config repo right? but moved as part fo nodepool in zuul? | 11:54 |
| sean-k-mooney | ya it did https://github.com/openstack/project-config/commit/b22bdcf480bf91562709da61252f14ba3c2f6f46 | 11:55 |
| sean-k-mooney | i guess at least ovh has not updated https://mirror.bhs1.ovh.opendev.org/debian/dists/ | 11:59 |
| sean-k-mooney | based on https://docs.opendev.org/opendev/infra-specs/latest/specs/unified_mirrors.html there shoudl eb a mirror per region/cloud in that format | 12:00 |
| sean-k-mooney | and i guess https://opendev.org/opendev/system-config/src/branch/master/hiera/common.yaml#L39-L48 are the corrent mierros | 12:01 |
| sean-k-mooney | the reason im askign is to knwo beyond a rebase what are ther crital path item that need to be doen to move https://review.opendev.org/c/openstack/devstack/+/954653 forward so that we can supprot debian 13 jobs as early in 2026.1 as possible | 12:05 |
| sean-k-mooney | im guessing it involves updating https://opendev.org/opendev/system-config/src/branch/master/playbooks/roles/reprepro/files/debian/config/distributions | 12:07 |
| sean-k-mooney | to add trixie | 12:07 |
| sean-k-mooney | so i will need to create a patch similar to https://opendev.org/opendev/system-config/commit/17a005a69b9f15353e9b22b179015928136e862e | 12:07 |
| frickler | sean-k-mooney: it also currently depends on a) fungi finishing the afs server upgrades and b) infra admins creating enough free space for the new mirror, not sure right away whether we have enough old stuff we can clean up or whether we'd also need to add new volumes to the afs servers | 12:33 |
| sean-k-mooney | ack. frickler is having the mirror a requirement to add the devstack supprot? | 12:34 |
| frickler | sean-k-mooney: another option I've been pondering which might also be helpful in the future is to teach our dib builds and also devstack and supporting roles to run without local mirrors | 12:34 |
| sean-k-mooney | i.e. would we be ok with movign forward with your orginal patch without using hosted mirros and just using the upstream ones? | 12:34 |
| frickler | currently this is not possible, but working on it would be worthwhile IMO | 12:34 |
| sean-k-mooney | well you kind of made it work by passign the upstream mirros in teh job | 12:35 |
| sean-k-mooney | https://review.opendev.org/c/openstack/devstack/+/954653/9/.zuul.yaml#763 | 12:35 |
| frickler | sean-k-mooney: yes, but also with the additional hack of removing the broken pip.conf that results from that | 12:35 |
| sean-k-mooney | ah right i see | 12:35 |
| frickler | https://review.opendev.org/c/openstack/devstack/+/954653/9/stack.sh#2 | 12:36 |
| sean-k-mooney | we could overrid pypi_mirror | 12:36 |
| frickler | if you make that conditional for trixie, we could merge that and amend later, yes | 12:36 |
| sean-k-mooney | well what i was thinking is defineing https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/defaults/main.yaml#L11 | 12:37 |
| sean-k-mooney | so that it also points to the upstream or pypi or otherwise seeign if it woudl be posible to make that work. | 12:38 |
| sean-k-mooney | but sure i coudl put it behind a conditional for now | 12:38 |
| frickler | whatever works | 12:39 |
| sean-k-mooney | ok if you dont mind, ill rebase your previous change and comment out hte removal of the pip.conf to see what the orginal erro was since the logs have roated | 12:40 |
| frickler | sure | 12:40 |
| sean-k-mooney | and then based on that ill look at add a "REMOVE_PIP_CONF" varibale and set that to true in the debian job | 12:41 |
| sean-k-mooney | once there is space for the mirros we can just drop the two job overrieds at a futrue date. | 12:42 |
| opendevreview | sean mooney proposed openstack/devstack master: Add support for trixie and platform job https://review.opendev.org/c/openstack/devstack/+/954653 | 12:51 |
| sean-k-mooney | looking at https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/defaults/main.yaml#L3-L11 im pretty sure we can make the pypi mirror work by setting pypi_fqdn and wheel_mirror to use the provider mirror directly in the job or via a pre playbook but if this is not going to be needed long term we may or may not want to do that | 12:54 |
| sean-k-mooney | im expectin gthis to fail becasue it wil try to use deb.debian.org/pypi/simple | 12:55 |
| sean-k-mooney | which is not a thing | 12:55 |
| sean-k-mooney | same for teh wheel cache | 12:55 |
| sean-k-mooney | but i think setting pypi_fqdn: "{{ zuul_site_mirror_fqdn | default(omit) }}" woudl work | 12:56 |
| sean-k-mooney | i might push a patch to the zuul jobs repo to alter the defaulting slightly as a test. with tha tsaid the devstack job will fail pretty quickly so ill wait for that to complete first | 12:58 |
| opendevreview | Merged openstack/devstack master: Restore os_CODENAME for old rhel distros https://review.opendev.org/c/openstack/devstack/+/961222 | 13:16 |
| opendevreview | Fernando Ferraz proposed openstack/devstack-plugin-nfs master: Exclude more tempest tests from NFS job https://review.opendev.org/c/openstack/devstack-plugin-nfs/+/952476 | 13:20 |
| opendevreview | Fernando Ferraz proposed openstack/devstack-plugin-nfs master: Exclude more tempest tests from NFS job https://review.opendev.org/c/openstack/devstack-plugin-nfs/+/952476 | 13:23 |
| sean-k-mooney | frickler: i think https://review.opendev.org/c/zuul/zuul-jobs/+/961369 could be a reasonable approch to making mirros optional in the jobs | 13:30 |
| sean-k-mooney | ill test that with a depend on and try usign that in the debain job and see how that goes for now | 13:31 |
| sean-k-mooney | im wonderign if a third condtional for the wheel chaching woudl also make sense but that could alwasy be a follow up | 13:32 |
| opendevreview | Fernando Ferraz proposed openstack/devstack-plugin-nfs master: Exclude more tempest tests from NFS job https://review.opendev.org/c/openstack/devstack-plugin-nfs/+/952476 | 13:35 |
| opendevreview | sean mooney proposed openstack/devstack master: Add support for trixie and platform job https://review.opendev.org/c/openstack/devstack/+/954653 | 13:46 |
| *** clarkb is now known as Guest26673 | 13:57 | |
| *** Guest26673 is now known as clarkb | 14:46 | |
| clarkb | sean-k-mooney: looking at the dists list in our mirrors is the easy way to know which debian releases are available: https://mirror.dfw.rax.opendev.org/debian/dists/ | 14:47 |
| sean-k-mooney | yep | 14:47 |
| sean-k-mooney | that what i ended up doing | 14:47 |
| clarkb | sean-k-mooney: frickler: I don't think the pip conf needs any changing. We never added a wheel mirror for noble and never changed its pip conf as far as I know | 14:47 |
| sean-k-mooney | clarkb: it does becasue of how frickler overrode the mirrors | 14:48 |
| sean-k-mooney | https://opendev.org/zuul/zuul-jobs/src/branch/master/roles/configure-mirrors/defaults/main.yaml#L3 | 14:48 |
| clarkb | oh we're using the mirror_fqdn value and configuring upstream like a mirror I see | 14:48 |
| sean-k-mooney | ya | 14:48 |
| sean-k-mooney | so i pushed a patch to zuul-jobs to make things condtional | 14:49 |
| sean-k-mooney | but it looks like that is not in requried projects | 14:49 |
| sean-k-mooney | so it dint get pulled into my current run i think | 14:49 |
| sean-k-mooney | im just reviewing the last test run to confirm | 14:49 |
| clarkb | if you use depends on it shouldn't matter. The problem is more like that configure-mirrors runs in a trusted context so isn't being applied speculatively | 14:49 |
| sean-k-mooney | ah | 14:50 |
| sean-k-mooney | ok right that makes sense | 14:50 |
| sean-k-mooney | so for now i can revert to pointign at the upstream debian mirros without the depends on | 14:50 |
| sean-k-mooney | and override teh pypi and mirror urls to wheel_mirror: "https://pypi.org/simple/" | 14:50 |
| sean-k-mooney | but if https://review.opendev.org/c/zuul/zuul-jobs/+/961369 is somethign folks are ok with it woudl be nicer to do it that way long term | 14:51 |
| opendevreview | sean mooney proposed openstack/devstack master: Add support for trixie and platform job https://review.opendev.org/c/openstack/devstack/+/954653 | 14:56 |
| opendevreview | sean mooney proposed openstack/devstack master: Add support for trixie and platform job https://review.opendev.org/c/openstack/devstack/+/954653 | 16:23 |
| opendevreview | sean mooney proposed openstack/devstack master: Add support for trixie and platform job https://review.opendev.org/c/openstack/devstack/+/954653 | 16:34 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!