[07:37] <lordievader> Good morning
[10:35] <flaf> Hi @all. Here is someone who knows “subiquity” (new installer for Ubuntu server) and can tell me if it's possible to make a full automatic and _*minimal*_ install of Ubuntu server, for instance without snapd (in clear just with openssh-server installed)? Thx for your help.
[10:37] <flaf> (in my case it's Ubuntu Focal Fossa 20.04)
[10:47] <flaf> (and doc of subiquity is here https://ubuntu.com/server/docs/install/autoinstall-quickstart)
[11:56] <Obscenity> a thought comes to mind when joing here, anyone noticed the newer installers are labeled "live" and not "installer"? its confused me a few times
[13:45] <dbungert> flaf: have you played with the late-commands portion of autoinstall?  I expect it should do what you want but haven't tested that case yet.
[13:47] <flaf> dbungert: yes, I see it. So, in this case, I should have to run something like “apt-get purge <package I don't wan't>”. But I'm wondering if it can be possible to limit installation to a minimum of packages from the start.
[13:48] <flaf> dbungert: but I think it's impossible, I'm afraid.
[13:48] <dbungert> flaf: for your exact case, no.
[13:49] <flaf> dbungert: ok, thx for your confirmation.
[13:49] <dbungert> we have some work in progress on a install source that is 500MBish smaller and "not intended for humans", but it doesn't address your snapless scenario
[13:50] <flaf> dbungert: "not intended for humanss"?
[13:51] <dbungert> flaf: ie to help save space we remove stuff that people using the device interactively might want - but things like apt are still there so you can human-ify it if needed :)
[13:53] <flaf> dbungert: interesting. How can I follow this project and know when it will be released?
[13:55] <dbungert> new subiquity will support it for the upcoming Impish release (and subsequent releases), not sure if it will apply to Focal
[14:01] <ogra> isnt subiquity itself moving to become a snap (for upgraeability during install) ?
[14:01] <dbungert> ogra: is already
[14:01] <dbungert> and it can do the upgrade during install, if the user chooses to do so when prompted
[14:02] <flaf> ogra: subiquity is a snap during the installer "session".
[14:04] <flaf> but the target installation can be installed without snap (in theory).
[14:33] <ogra> dbungert, right, that was what i remembered ... i was just wondering if that might have any impact on the target system ... but seems like not
[14:40] <dbungert> ogra: I think I'm missing something, would you elaborate?
[14:41] <ogra> dbungert, i was wondering if the snap or parts of it would be needed/required in the installed target for some reason ... but since thats not the case snapd is indeed removabe from the target ... all fine
[14:41] <dbungert> ah yes, subiquity has no business on the installed target system
[14:41] <ogra> right
[15:37] <flaf> dbungert: do you know why, during the installation via subiquity, the OS installer try to connect https://api.snapcraft.io?
[15:38] <dbungert> it enumerates a list of suggested snaps for the snaps page
[15:39] <dbungert> also might be involved in subiquity's check for a new version of subiquity
[15:40] <flaf> dbungert: in my case, the VM has no Internet access, only a proxy for APT. But it seems to not be a problem during my auto-installation.
[18:24] <flaf> dbungert: sorry for the ping, I have tried this in late-commands => “curtin in-target --target=/target -- apt-get -y -q purge snapd” during my automatic install. But I can see the error message “finish: subiquity/Late/run: ... purge snapd']' returned non-zero exit status 100.” Is there a log file where I can see the output of the command to see the error (maybe)?
[18:25] <dbungert> flaf: check journalctl, it outputs additional information.  I'd love to have all that in the subiquity log also but for now it's in the journal.
[18:26] <flaf> Ok, thx dbungert.
[18:33] <dbungert> flaf: just be wary of LP: #1939176, which I do expect we'll have fixed in the next release
[18:33] <dbungert> (late-commands related hang)
[18:39] <flaf> dbungert: ok, thx for the information. In my case, it's a problem during the purge where a postscript try a ummount in /run/... which fails.
[18:42] <dbungert> flaf: that one sounds like we should file a bug, if you don't mind.  https://bugs.launchpad.net/subiquity/+filebug
[18:46] <flaf> dbungert: I can file a bug but I'm not sure it's a bug in this case. During the "curtin in-target --target=/target -- apt-get purge snapd", a postscript wants to unmount a mountpoint in /run/snap/... (sorry I don't remember the path exactly). And it's logic that the directory in not yet mounted because the target OS is not started yet. So for me it's not a bug, no?
[18:48] <dbungert> flaf: maybe, maybe not.  In my opinion late-commands should be a fairly rich mechanism for customizing the install - an escape hatch for things we don't have direct support for.  If it's as easy as mounting something it needs, that sounds worthwile. 
[21:44] <mwhudson> dbungert, paride: ah seems the test answer files all need sources: sections adding