[08:15] <psivaa> cjwatson: curious if you'd be able to take a look at: https://bugs.launchpad.net/ubuntu/+source/base-installer/+bug/1359712 please?
[08:15] <ubot2> Launchpad bug 1359712 in base-installer (Ubuntu) ""base-installer: error: /cdrom/install/filesystem.squashfs has failed to be mounted as squashfs." is seen on trusty server instalaltions" [Undecided,New]
[08:15] <psivaa> cjwatson: plars mentioned someone was asking for trusty server install failures
[08:15] <psivaa> s/for/about
[10:37] <cjwatson> psivaa: ack
[10:53] <fish_> re
[10:54] <fish_> cjwatson: not directly installer related but just saw you were involved in 'biosdevname on poweredge'. so I just installed trusty (sucessfuly this time ;)) but now my 2nd interface sometimes is called eth1, sometimes em2
[10:55] <fish_> so well, there is this problem now and I'm wondering whether I should care to fix it or jsut disable biosdevname
[11:01] <cjwatson> I would just disable it if it is causing trouble
[11:01] <fish_> franky, given that dell 'invented' this doesn't make me more confident in using it - even on dell boxes.
[11:01] <cjwatson> it ... wasn't really my idea to enable it by default
[11:02] <fish_> what should it solve anyway? it's not very clear to me.. the device ordering should be the same across all my systems if the hw is the same. and if I change the hw, I reinstall the systems anyways
[11:02] <fish_> cjwatson: ok, all the discussion sounded like that :) thanks for confirming, I'll just disable it
[11:02] <cjwatson> network device enumeration isn't necessarily stable across reboots even with the same hardware
[11:03] <cjwatson> can depend on such things as which order the PCI bus happens to respond in
[11:03] <fish_> cjwatson: I've heard about that but I never saw that happen across reboots
[11:04] <fish_> IIRC the order gets encoded as udev rules during install anyway, right? but I never saw it during reinstalls either
[11:05] <fish_> well, actually a few years ago I had such issue once when bootstrapping debian but it was somehow caused by odd kernel parameters (aka me)
[11:09] <cjwatson> my understanding is that the udev rule approach is not in fact reliable and can e.g. result in interfaces getting stuck with renamed_foo names
[11:11] <fish_> hrm well, I see how it makes sense to have something like biosdevname.. well, but since I was thinking 'I need to reinstall those systems, let's just install trusty now' and this already took me longer than I hoped I'll stick with the good ol' eth naming for now
[11:47] <fish_> but things are really odd with trusty. why on earth does it run dhcp even though /etc/network/interfaces doesn't set that up?
[11:56] <fish_> ah never mind, that's my fault
[13:27] <cjwatson> psivaa: ah, just a d-i kernel abi bump needed, I'll do that
[13:27] <psivaa> cjwatson: ack, thanks.
[15:29] <fish__> now I hoped I've fixed the lvm volume creation in my late script but now the ubuntu-vg vg can't be found
[15:39] <fish__> is it possible to set the default lvm volume group name? it defaults to the hostname which I don't want (and don't think is a good idea in general)
[16:19] <fish__> I can't even find the sources for whatever is defining the volume group name
[16:20] <fish__> could someone point me to that?
[16:28] <fish__> finally I found it: partman-auto-lvm/new_vg_name - would be great of those things are documented somewhere. it's not in here for example: https://help.ubuntu.com/lts/installation-guide/example-preseed.txt
[16:40] <fish__> so now my last problem with this install: somehow it didn't install ifenslave even though I specified it in my preseed: d-i pkgsel/include string openssh-server lxc-docker btrfs-tools apparmor ifenslave
[16:40] <fish__> the log shows that it's installing apparmor but not ifenslave
[16:59] <fish__> I'll reinstall this machine again to make sure. it sounds so unlikely that even I can't believe it ;)