| ianw | JayF: AIUI, and https://setuptools.pypa.io/en/latest/deprecated/zip_safe.html goes into a bit -- egg's can be .zip files but it's not a wheel thing. i don't think we've ever had zip_safe specified, but there's no way dib would have worked as it's all relative scripts calling each other, and nobody complained about it | 00:36 |
|---|---|---|
| ianw | i think we're fine to just offset off file() | 00:37 |
| JayF | I believe you, but will verify ... it's crazy to me they'd create such a wacky interface for an incredible edge case | 03:00 |
| JayF | clarkb when you get a chance, if you could revisit https://review.opendev.org/c/openstack/diskimage-builder/+/907757 I'd appreciate it. I expect Sam to move forward (on his own; we got his ubuntu account fixed) with the next follow-ups for this once it lands. | 15:26 |
| clarkb | JayF: can you rebase 907904 so that it is on the latest version of the fixup change making it mergeable? I'd like to approve the two changes together to avoid regressions | 16:50 |
| opendevreview | Jay Faulkner proposed openstack/diskimage-builder master: Re-enable voting on Gentoo DIB CI job https://review.opendev.org/c/openstack/diskimage-builder/+/907904 | 16:51 |
| JayF | sweet, I did it without pushing the first one too | 16:51 |
| clarkb | oh the file mode for disk-image-create is still set to executable | 16:53 |
| clarkb | sorry I missed that on pulling it up this mornng beacuse I looked at hte patchset delta | 16:53 |
| JayF | yes, as commented it was not originally committed intentionally, but upon review it is the correct fix | 16:54 |
| clarkb | and now I see the comment | 16:54 |
| JayF | because otherwise you can't run it in a venv without doing that manually | 16:54 |
| clarkb | well I'm not convinced it is correct | 16:54 |
| clarkb | but I don't think it will hurt either | 16:54 |
| JayF | at least we see to talk about it | 16:54 |
| JayF | gerrit++ | 16:54 |
| clarkb | the python package install will install a disk-image-create python script | 16:56 |
| clarkb | I almost wonder if the shell disk-imgae-create is dead code | 16:56 |
| clarkb | oh no the python execs the shell | 16:57 |
| clarkb | ok ya this is fine | 16:57 |
| opendevreview | Merged openstack/diskimage-builder master: Fix various minor issues with Gentoo; make CI pass https://review.opendev.org/c/openstack/diskimage-builder/+/907757 | 21:55 |
| JayF | There's something slightly flakey about the dib-functests fedora image | 22:40 |
| JayF | it keeps failing, and looking at the logs, it's like it's failing to find the right image to dl | 22:41 |
| JayF | well, "keeps failing" == 2x fails and like, 3x success since I've been looking | 22:41 |
| JayF | enough to raise an eyebrow | 22:41 |
| ianw | > oh no the python execs the shell | 22:59 |
| ianw | yeah, there was a thought that over time the python would become more and more sane (argparse, etc.) and the shell script smaller and smaller | 23:00 |
| clarkb | JayF: can you link to the failures? | 23:09 |
| * clarkb is secretly on a mission to get everyone talking about job failures to always link the location of the failurse as well | 23:09 | |
| JayF | https://zuul.opendev.org/t/openstack/build/82acbb872a6948549a20632fa110caf5 specifically... | 23:09 |
| JayF | https://zuul.opendev.org/t/openstack/build/82acbb872a6948549a20632fa110caf5/log/logs/fedora_build-succeeds.FAIL.log#279 | 23:10 |
| JayF | A cursory look makes me wonder/suspect that Fedora might have a bad machine in rotation | 23:10 |
| clarkb | /fedora/linux/releases/37/Cloud/x86_64/images/Fedora-Cloud-Base-37-1.7.x86_64.qcow2 is the actual path but we're hitting /fedora/linux/releases/37/Cloud/x86_64/images/Fedora-Cloud-Base-37-.x86_64.qcow2 | 23:11 |
| JayF | https://zuul.opendev.org/t/openstack/build/82acbb872a6948549a20632fa110caf5/log/logs/fedora_build-succeeds.FAIL.log#236 | 23:12 |
| JayF | yeah, I think here, when it's getting the image list and grepping outta it | 23:12 |
| JayF | it's getting a bad response there "N" amount of times, is my hypothesis | 23:12 |
| JayF | I've never locally seen a result from that URL that wouldn't have given a better BASE_IMAGE_* result | 23:12 |
| clarkb | huh so ya a bad host providing a list that isn't valid generally (or at all) then a request for the resulting file fails against other backends? | 23:15 |
| clarkb | https://download.fedoraproject.org/pub/fedora/linux/releases/37/Cloud/x86_64/images this is where the list comes from | 23:16 |
| clarkb | I get back two SUBRELEASES of 1.7 when I run this locally | 23:19 |
| clarkb | I guess bash may flatten that for us | 23:19 |
| opendevreview | Clark Boylan proposed openstack/diskimage-builder master: Write fedora download redirect info to stderr https://review.opendev.org/c/openstack/diskimage-builder/+/908207 | 23:28 |
| clarkb | JayF: something like that may give us pointers to the bad backend | 23:28 |
| JayF | +1 | 23:28 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!