| *** Guest7381 is now known as bbezak | 14:56 | |
| opendevreview | Rafal Lewandowski proposed openstack/diskimage-builder master: Change grub variables for style and timeout https://review.opendev.org/c/openstack/diskimage-builder/+/937684 | 15:52 |
|---|---|---|
| opendevreview | Rafal Lewandowski proposed openstack/diskimage-builder master: Change grub variables for style and timeout https://review.opendev.org/c/openstack/diskimage-builder/+/937684 | 15:53 |
| opendevreview | Rafal Lewandowski proposed openstack/diskimage-builder master: Change grub variables for style and timeout https://review.opendev.org/c/openstack/diskimage-builder/+/937684 | 18:34 |
| JayF | Do we care about documenting bugs through launchpad? Or is it just about release notes | 20:22 |
| JayF | my downstream reported a bug (really an edge case one of our features misses), and I'd like to JFDI the fix | 20:22 |
| opendevreview | Jay Faulkner proposed openstack/diskimage-builder master: wip: ubuntu-minimal: respect DIB_APT_SOURCES https://review.opendev.org/c/openstack/diskimage-builder/+/940407 | 20:41 |
| clarkb | I'm not too troubled by JFDI and edit release notes | 20:55 |
| clarkb | out of curiousity does the apt-sources element not work with -minimal because it needs to boostrap debootstrap before it applies the source file? | 20:57 |
| clarkb | or maybe we're overriding it | 20:57 |
| clarkb | ya we're overriding instead. JayF I think you can just skip writing anything if apt-sources is used? | 20:57 |
| clarkb | rather tahn writing things again | 20:58 |
| JayF | it operates at the wrong phase | 21:00 |
| JayF | it works in extra-data.d, we need it in early root | 21:00 |
| JayF | these -minimal builds sorta have the apt sources logic mixed in | 21:00 |
| JayF | I was more going to lean the route of deprecating the apt-sources module | 21:00 |
| clarkb | why not just make apt-sources run in root then? | 21:01 |
| JayF | because if we significant change the execution order, it's already an API break, and it'll be pretty difficult to rework all the apt sources logic to factor it outta all those -minimal elements | 21:01 |
| clarkb | I don't think having it run in root is an api break | 21:01 |
| JayF | clarkb: debootstrap has root.d/00-blah where it does the pull. we'd have to move that to 01, make it aware of the arg as well | 21:01 |
| clarkb | I guess it could stop overriding things and get overridden? | 21:01 |
| JayF | that element, btw, is 8 years old no update | 21:02 |
| JayF | and I suspect has been functionally nonworking in most cases for a while, after reading some of the ubuntu elements | 21:02 |
| clarkb | I'm not sure that is meanginful fwiw | 21:02 |
| clarkb | most elements in dib are old and don't get updates | 21:02 |
| clarkb | looking at the file you chagned its 75- not 00- | 21:02 |
| clarkb | so we'd just need to run before that point | 21:02 |
| JayF | oh. OH. | 21:02 |
| JayF | aw hell | 21:02 |
| JayF | okay, back to the drawing board | 21:03 |
| JayF | thanks for talking me through it | 21:03 |
| clarkb | and ya you'd need to be aware of not writing a file later if the var is set but if we can centralize all the error checking for validity across the various deb based systems that would be nice | 21:03 |
| clarkb | otherwise we need to have that in a bunch of different places | 21:03 |
| clarkb | (I think that approach would work if we are forced into it, I just want to make sure it really is the lesser evil) | 21:05 |
| JayF | well let me put it this way, you already reset this from a "5 line minimal change" to something beefier, so I'll do the better/beefier change | 21:19 |
| JayF | so there's a chicken/egg thing here | 21:42 |
| JayF | there's no root to install a sources.list in until after debootstrap runs | 21:42 |
| JayF | afaict at least | 21:42 |
| clarkb | oh hrm in 08-debootstrap it does point debootstrap at the targetroot | 21:44 |
| clarkb | but debootstrap itself takes the mirror host as an argument | 21:44 |
| JayF | yeah | 21:44 |
| JayF | I'm thinking this might be a "debootstrap is not the right shape to solve this problem" issue as much as anything | 21:44 |
| clarkb | so I think if you run between 08-debootstrap and 75-ubuntu-minimal-baseinstall it would work | 21:44 |
| JayF | which would be /really/ unfortunate given my set of requirements | 21:44 |
| clarkb | but you'd still not solve this for debootstrap itself | 21:45 |
| JayF | 08-debootstrap hits the network | 21:45 |
| clarkb | yes. why can't you use the mirror argument? | 21:45 |
| JayF | I basically have an isolated network build machine, and the mirror we have isn't "shaped" like debian expects | 21:45 |
| * JayF redacting you an example | 21:45 | |
| clarkb | I would expect that to be a problem for deboostrap regardless of our sources config? | 21:46 |
| clarkb | but I'm not super familiar with debootstrap to know if you can configure it with something like a richer sources config | 21:46 |
| JayF | https://www.irccloud.com/pastebin/6epyg5c0/ | 21:46 |
| JayF | basically you can't just swap out the hostname, you need to set the path, and the path is unique per ... component? | 21:47 |
| JayF | and all the debootstrap stuff AFAICT is built 100% assuming your mirror is debian-shaped, which just isn't a good assumption in most corporate environments which would have a single dedicated artifact store | 21:48 |
| clarkb | ya so you've got a separate top level dir for each dists/ entry rather than letting all the dists/ entries coexist like: https://mirror.dfw.rax.opendev.org/ubuntu/dists/ | 21:48 |
| clarkb | which works if you have a proper sources.list config but I'm not sure about getting debootstrap to undersatnd that. It may be possible | 21:49 |
| JayF | yes, exactly | 21:49 |
| JayF | this is why I wanna sub in the full sources.list | 21:49 |
| JayF | and was going to hook it up to -minimal because ... that's just a better overall image tbh | 21:49 |
| clarkb | https://wiki.debian.org/Multistrap | 21:52 |
| JayF | at least debootstrap is in shell, I can dig that deep | 21:52 |
| clarkb | though reading that I'm wondering if debootstrap will pull packages from anything but the base distro release repo | 21:52 |
| JayF | that's EXACTLY what I'm looking for now | 21:52 |
| clarkb | so you may still get away with it pointed at https://some-hostname/deb/ubuntu-22.04-x86_64/some-string-jammy ? | 21:52 |
| JayF | exactly, I'm checking that now | 21:53 |
| JayF | multistrap isn't packaged widely, which makes it mostly a nogo | 21:53 |
| clarkb | ya it only runs on debian apparently | 21:53 |
| JayF | debootstrap looks like it's relying on implementation details of the mirror in setup_apt_sources afaict | 21:55 |
| JayF | https://salsa.debian.org/installer-team/debootstrap/-/blob/master/functions?ref_type=heads#L1198 | 21:55 |
| JayF | I'll give that to downstream to attempt though, it'll be a useful error | 21:56 |
| clarkb | that may only affect EXTRA_SUITES? | 21:57 |
| clarkb | so if you don't add -security etc it may work? | 21:57 |
| clarkb | you might also be able to get away with symlinks in your mirror or something along those lines | 21:58 |
| clarkb | but also its not really a mirror if it doesn't reflect upstream :/ | 21:58 |
| clarkb | (we, opendev, do hacks like that in other places so I get why it happens but you do have to be careful it continues to work) | 21:58 |
| JayF | thanks for the help, I'm going to talk to my downstream and see if this is a situation where dib is even the right tool | 22:07 |
| JayF | because I think we're going to run into other limitations | 22:07 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!