| opendevreview | Steve Baker proposed openstack/diskimage-builder master: Depend on dib-python to set DIB_PYTHON_VERSION https://review.opendev.org/c/openstack/diskimage-builder/+/802408 | 00:03 |
|---|---|---|
| *** rpittau|afk is now known as rpittau | 07:29 | |
| *** rpittau is now known as rpittau|afk | 16:40 | |
| JayF | Hey, curious if you all think you'd take a patch here https://github.com/openstack/diskimage-builder/blob/master/diskimage_builder/elements/yum-minimal/root.d/08-yum-chroot#L74 to allow me to specify more bootstrap packages for the chroot | 17:37 |
| JayF | trying to get this working with a downstream reskin distro, which has more `*-release` files I have to install | 17:38 |
| JayF | I JFDI'd a patch. Will look for feedback on it :) | 17:47 |
| opendevreview | Jay Faulkner proposed openstack/diskimage-builder master: Permit specification of extra bootstrap packages https://review.opendev.org/c/openstack/diskimage-builder/+/802592 | 17:47 |
| JayF | https://zuul.opendev.org/t/openstack/build/476a5f5f04ab42ffaf929f0aeab40fd0 I've dug through the failure here, and I can't find the actual failure output anywhere; what am I missing? | 20:09 |
| ianw | JayF: in that case, the nodepool is failing to build a centos-7 image | 20:20 |
| ianw | https://zuul.opendev.org/t/openstack/build/476a5f5f04ab42ffaf929f0aeab40fd0/log/nodepool/builds/ are the attempts to build | 20:20 |
| ianw | https://zuul.opendev.org/t/openstack/build/476a5f5f04ab42ffaf929f0aeab40fd0/log/nodepool/builds/test-image-0000000004.log#313 appears to be the cause | 20:20 |
| ianw | 2021-07-27 18:45:24.333 | /opt/dib/tmp/dib_build.M8wXyJz9/hooks/root.d/08-yum-chroot: line 74: DIB_YUM_MINIMAL_BOOTSTRAP_PACKAGES: unbound variable | 20:20 |
| JayF | that's exactly the file I'm looking for; thank you | 20:21 |
| JayF | How did I miss that, I thought I opened every file :| | 20:21 |
| JayF | thanks | 20:21 |
| ianw | yeah, unfortunately it doesn't give you a big red arrow pointing to the failure. all you see is our polling of "is the image built and booted" timing out in the main logs | 20:22 |
| JayF | yeah; I figured out nodepool was doing the work but just must have overlooked those files somehow | 20:23 |
| JayF | I'll get that variable happy so DIB is happy with it unset, thanks for the pointer | 20:24 |
| ianw | yeah, just a DIB_YUM_MINIMAL_BOOTSTRAP_PACKAGES:- usually does it | 20:27 |
| JayF | you OK with the change in concept? | 20:28 |
| JayF | Trying to not have to fork yum-minimal just to support a downstream rhel/centos skin : | 20:28 |
| JayF | :| | 20:28 |
| ianw | i can see that installing things at this stage might be required for that | 20:29 |
| ianw | only thing i'd say is that we've already moved to containerfile for fedora, and it would be my strong preference to continue in that direction | 20:30 |
| ianw | i don't see yum-minimal having too much more life, since it relies heavily on alien rpm which always feels like it's on thin ic3e | 20:31 |
| ianw | ice | 20:31 |
| ianw | (the ice actually breaking when fedora 34 introduced a new compression format for rpms) | 20:32 |
| JayF | I don't know what `containerfile` is | 20:41 |
| JayF | but I'm extremely skeptical there's going to be a move that will work as well for IPA ramdisks :) | 20:41 |
| JayF | the -minimal part matters greatly, and since I'm consuming the IPA elements and they expect a -minimal base, I'm going to keep down my path for now | 20:42 |
| opendevreview | Jay Faulkner proposed openstack/diskimage-builder master: Permit specification of extra bootstrap packages https://review.opendev.org/c/openstack/diskimage-builder/+/802592 | 20:47 |
| ianw | ah; yeah basically we're using upstream docker images as the initial chroot, rather than building our own | 20:48 |
| JayF | That sounds suspiciously similar to how we used to do IPA builds -- we'd make a chroot using docker tooling (with an export to tgz at the end) and then chroot into that from a CoreOS ramdisk | 20:49 |
| JayF | I approve, that's GREAT tooling to use for that. | 20:49 |
| ianw | yeah, i mean it's sort of where dib started; by being a framework to modify upstream .iso. but that was a constant headache as distros would constantly add or remove things because they weren't really meant to be consumed for modification | 20:52 |
| JayF | https://github.com/openstack/ironic-python-agent/blob/stable/pike/Dockerfile if you go back far enough we still have a dockerfile :D | 20:52 |
| ianw | but with containers it's a different mindset; these are things meant to be built upon | 20:52 |
| JayF | that proxy.sh stuff might be something DIB would be interested in, basically adding support for socks proxies to docker building | 20:52 |
| JayF | we had a lot of trouble with that in IPA upstream back then, because intel contributors couldn't build 'em until we got the proxy working | 20:53 |
| ianw | yeah, i mean you can do a lot of what dib elements do in a Dockerfile | 20:53 |
| ianw | except instead of having a script that runs and can be edited, you're munging a multi-line string in the Dockerfile | 20:54 |
| JayF | yeah; I don't mind docker as a base for stuff like this at all, it'll just be painful when we go to migrate IPA I'm sure | 20:55 |
| JayF | in the meantime, I'll keep getting my $employer-rhel working :D | 20:55 |
| ianw | in terms of what dib outputs, it looks the same | 20:56 |
| JayF | https://review.opendev.org/c/openstack/diskimage-builder/+/802592 is passing CI now, ty for the help | 22:23 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!