| *** blarnath is now known as d34dh0r53 | 15:01 | |
| opendevreview | daniel.pawlik proposed openstack/ci-log-processing master: Add information about CloudFormation status on AWS https://review.opendev.org/c/openstack/ci-log-processing/+/905244 | 15:46 |
|---|---|---|
| opendevreview | Ian Wienand proposed openstack/openstack-zuul-jobs master: openafs-rpm-package-build: update to openafs 1.8.10 https://review.opendev.org/c/openstack/openstack-zuul-jobs/+/905270 | 21:26 |
| clarkb | ianw: ^ is that what is required to make centos stream work? | 21:26 |
| clarkb | maybe its an idea anyway | 21:26 |
| ianw | likely -- it looks like i started on it with https://review.opendev.org/c/openstack/openstack-zuul-jobs/+/885908?usp=search to use 1.8.10-pre1 but got distracted i guess ... | 21:27 |
| ianw | the best solution would be for us to copy the wheels back through the executor so each build node didn't need to install openafs | 21:28 |
| ianw | but as you say; more and more critical things are moving to anylinux and providing wheels anyway ... so perhaps just saying we don't support it is fine too | 21:29 |
| clarkb | ya I think the main thing that is a problem now is libvirt-python | 21:30 |
| clarkb | even lxml is providing wheels these days | 21:30 |
| ianw | openafs-rpm-package-build-centos-8-stream-arm64 - error: Failed build dependencies: | 21:33 |
| ianw | kernel-devel-aarch64 = 4.18.0-522.el8 is needed by openafs-1.8.10-1.el8.aarch64 | 21:33 |
| ianw | that suggests to me that the arm64 node trying to do that build is not running the same kernel as in the mirrors | 21:33 |
| ianw | either the nodes aren't building and updating, or something up with the mirror :/ | 21:34 |
| ianw | same thing with centos-9 - https://zuul.opendev.org/t/openstack/build/b22b1b7c308f4c64bfc4fac25ffaab09 | 21:35 |
| fungi | ouch, yeah i guess https://pypi.org/project/libvirt-python/#files is still just sdists | 21:36 |
| clarkb | looks like arm64 images are failing to build | 21:36 |
| clarkb | if we fix that in theory we get new images then maybe we get working openafs package builds | 21:36 |
| clarkb | fungi: ya iirc the libvirt devs / red hat lawyers have some sort of concern about publishing binary packages to pypi due to licensing terms | 21:37 |
| clarkb | /opt is full on nb04 | 21:38 |
| clarkb | I'll go ahead and stop the builder and see what I can do about cleaning it up | 21:38 |
| fungi | thanks! | 21:38 |
| ianw | ++ thanks! x86 looking better and building https://zuul.opendev.org/t/openstack/stream/63aed614abdf4b6bac52ab1f75f08a6e?logfile=console.log | 21:40 |
| clarkb | was going to run du against dib_tmp first to figure out how much space I can save but this is slow. Might be best to just start deleting things and see where we end up | 21:41 |
| fungi | yeah, df, rm, df | 21:41 |
| fungi | those dib caches and tempdirs are full of tiny files | 21:41 |
| tonyb | clarkb: in no way am I suggesting you don't send that email. but I'll work on fixing the wheel builds. | 21:42 |
| fungi | measuring them all takes an age | 21:42 |
| ianw | i think we'll get ephemeral build environments working before we figure out how to get dib to stop leaking :) | 21:42 |
| tonyb | possibly by using kAFS etc etc | 21:42 |
| clarkb | tonyb: ok in that case I'm happy to pull on this thread (and others) first before we need to send an email. We can always update the email for a different deadline | 21:42 |
| ianw | i think if we can get 1.8.10 built it will solve the immediate issue; but yes working on any of the long-wanted updates like copying via the executor/kafs etc. all good | 21:44 |
| tonyb | I feel like setting a deadline is good | 21:48 |
| clarkb | tonyb: ya but then people accuse me of being anti $thing when I say I want to turn it off | 21:57 |
| clarkb | now 33GB are free. This is going to take a while | 22:22 |
| tonyb | clarkb: Okay. How about I send the message that you've drafted. Will that mean the grumbles end up in my direction? | 22:23 |
| fungi | if you enjoy having grumbles in your direction, we can probably find better ways to achieve that | 22:24 |
| clarkb | tonyb: heh I think that might help a bit given the specific ocntent of this message (centos things) | 22:24 |
| fungi | you're tired of being known as "clark the centos hater?" | 22:25 |
| clarkb | ya exactly :) I'd rather be "clark the person indifferent to centos as long as clark doesn't have to fix all the centos things" :) | 22:27 |
| fungi | that's a lot more of a mouthful | 22:29 |
| JayF | clark who doesn't give two centos's is much more pithy | 22:30 |
| tonyb | JayF: ++ | 22:36 |
| tonyb | fungi: I'm happy to be grumbled at but only for specific reasons | 22:37 |
| tonyb | clarkb: So once we're happy with the draft email I'll send it. | 22:38 |
| fungi | what's funny is that we have two centoses (8 and 9, or x64 and aarch64 depending on how you look at it) | 22:40 |
| clarkb | tonyb: I think I'm ahppy with it as is. But was hoping that frickler could weigh in as frickler was the initial reporter of thei ssues | 22:41 |
| tonyb | clarkb: Okay. I'll wait for that then. | 22:42 |
| *** jph4 is now known as jph | 23:48 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!