[00:00] "This document is entirely a description of something that does not yet exist" [00:00] I like that :) [00:01] sarnold: turns out writing documentation for your new feature is quite a good way to shake out stupid ideas1 [00:01] ! [00:02] mwhudson: yes! I tried that once and realized that as much as I wanted to write one specific tool, I had no idea what I wanted that tool to do :D [00:02] mwhudson: is it the YAML section you're trying to format? [00:03] mwhudson: if so, might be useful - now YAML is used an many places - to ask the sysadmin team to add https://moinmo.in/ParserMarket/YAML to MoinMoin [00:03] mwhudson: *thanks* for the "creating an autoinstall file" section :) that's got ot be the number one complaint with preseed files, no one knows how to make one or what to put in them. [00:04] sarnold: heh yes although that part is extra double not existing [00:05] mwhudson: but you're *thinking* about it, which is promising [00:05] any other comments welcome of course, it's probably almost time to send it to ubuntu-devel@ [00:07] mwhudson: can you include the netplan yaml directly inline into the network: field? [00:07] sarnold: yes [00:08] mwhudson: alright, how about cloud-init configs? [00:08] mwhudson: ssh keys from launchpad / github etc? [00:09] sarnold: cloud-init no, pending actual sensible use case [00:09] mwhudson: heh, fair, I wondered even as I typed it :) [00:09] mwhudson: ssh hostkey? [00:09] sarnold: not sure i see value importing keys vs just having the key there, i guess it's easier to type [00:09] mwhudson: a random blob to shove into /dev/random? [00:10] mwhudson: how about a way to say packages that *shouldn't* be included? [00:13] sarnold: not sure i understand [00:13] mwhudson: which part? [00:13] sarnold: late_commands supplies you with an indefinite amount of rope to tie around your own neck... [00:13] mwhudson: heh [00:13] sarnold: either of the last two? [00:13] sarnold: you mean packages that are usually part of ubuntu server to not install? that doesn't seem like a great idea [00:13] but installing a package to just then apt-get purge it is a bit funny vs just not installing it itn the first place. not a huge deal I guess.. [00:14] sarnold: install is blatting a squashfs onto disk [00:14] AH [00:14] you don't get to only blat part of it :) [00:14] man this thing's gonna be *fast* [00:15] mwhudson: the random blob is to try to give a machine enough entropy to do things like generate ssh host key without taking ten minutes or asking someone to type gibberish on the keyboard [00:20] sarnold: how does that work? do you drop a file on disk somewhere for the kernel to read on first boot? [00:22] anyway none of the ways of getting the autoinstall config to the installer seem especially secure so i'm not sure how wise putting secrets in there would be [00:24] mwhudson: hm. good point. :/ [00:25] i guess we can support fetching it over https [00:25] So we can have zstd squashfses, when initramfses? :3 [00:25] mwhudson: you can cat random data into /dev/random and the kernel will use it, but not account for it; you've got to issue an ioctl to get the kernel to account for it, whiuch usually means using rng-tools [00:34] mwhudson: okay I think one last one.. raid configs? [00:34] sarnold: are supported by curtin [00:35] mwhudson: nice. I remember folks saying that the newish installer didn't support their favourite raid setup as well as debian-installer [00:38] i think i've fixed most of that now [00:38] yay :) [00:38] thanks mwhudson [02:34] https://bugs.launchpad.net/ubuntu/+source/python-acme/+bug/1836823 sounds like fun! [02:34] Launchpad bug 1836823 in python-acme (Ubuntu) "python-acme will break on November 1st" [Undecided,New] === cpaelzer__ is now known as cpaelzer [08:58] mwhudson: it's still catching up. I ran casper manually for you. [09:42] why is getgroups() not working on launchpad builders? [09:42] https://github.com/gridcf/gct/issues/95 [09:50] rbasak: thanks [09:55] LocutusOfBorg: a bug report on launchpad-buildd would be a much better way to ask that question [09:56] cjwatson, if you say that it might be a bug I would happy to report it :) [09:57] I'm trying to reproduce locally before opening a bug [09:57] I mean if you have a local launchpad-buildd setup already then sure :) [09:57] cjwatson, I was trying with pbuilder [09:57] and other sbuild isntances [09:57] they may or may not be close enough [09:57] but it works... [09:58] I know of no particular reason why we'd want the group setup to be invalid for this, so it seems likely to be a detail nobody else has run into before for whatever reason [09:59] thanks I'm not usually inclined to just open bugs because meh, its mostly always my fault :) [09:59] I'm trying debomatic and then I'll open it [10:01] Might be something like a mismatch between the buildd gid in the base VM and the chroot which nothing fixes up [10:05] LP: 1836870 [10:05] Launchpad bug 1836870 in launchpad-buildd "getgroups() fails on launchpad builders." [Undecided,New] https://launchpad.net/bugs/1836870 [11:14] Logan, hello, can you please do something for groovy? sync/merge? === ricab is now known as ricab|lunch [12:07] //win 8 === JanC is now known as Guest31310 === ricab|lunch is now known as ricab === JanC_ is now known as JanC [20:18] Laney: uploaded a new gcc-9. I doubt that things will be magically fixed, but would appreciate a glib2.0 rebuild and mutter check [23:29] Which channel would the ubuntu graphics driver PPA hangout on?