Fails to reboot after installing Ubuntu GNOME 14.10 release candidate in KVM

[Marius Gedminas]

I was testing Ubuntu GNOME utopic-desktop-amd64.iso (final release candidate, downloaded today) with testdrive in KVM .

When the installation was complete and I was prompted to reboot, I accepted. Then I ended up staring at this text screen for 10 minutes (I timed it): https://twitter.com/mgedmin/status/524825755566751744

During this time I could switch between virtual consoles with Alt-F1, Alt-F2, Alt-F3, Alt-F7.

Pressing Enter did nothing in any of the consoles.

I did not think to try Ctrl+Alt+Del, and I couldn't use Alt+SysRq in KVM.

ProblemType: Bug
DistroRelease: Ubuntu 14.10
Package: ubiquity (not installed)
ProcVersionSignature: Ubuntu 3.16.0-23.31-generic 3.16.4
Uname: Linux 3.16.0-23-generic x86_64
ApportVersion: 2.14.7-0ubuntu8
Architecture: amd64
CurrentDesktop: GNOME
Date: Wed Oct 22 10:47:18 2014
InstallCmdLine: file=/cdrom/preseed/ubuntu-gnome.seed boot=casper initrd=/casper/initrd.lz quiet splash -- maybe-ubiquity
InstallationDate: Installed on 2014-10-22 (0 days ago)
InstallationMedia: Ubuntu-GNOME 14.10 "Utopic Unicorn" - Release amd64 (20141022)
ProcEnviron:
 TERM=xterm
 PATH=(custom, no user)
 XDG_RUNTIME_DIR=<set>
 LANG=lt_LT.UTF-8
 SHELL=/bin/bash
SourcePackage: ubiquity
UpgradeStatus: No upgrade log present (probably fresh install)

[Marius Gedminas (mgedmin)]

Tried this again (but with LVM and encryption enabled), got stuck at the reboot stage again.

Ctrl+Alt+Del in the VM emits a broadcast message on /dev/tty1 about rebooting, but doesn't reboot.

I figured out how to send magic SysRq commands (Ctrl-Alt-2, sendkey alt-sysrq-s, then Ctrl-Alt-1 and Alt-Left/Right to make it refresh the screen, otherwise the alt-sysrq command output does not appear). Alt+SysRq+l/t/w are disabled by the kernel and my qemu froze before I got to Alt+SysRq+e/i/k.

[Ubuntu QA Website (ubuntuqa)]

This bug has been reported on the Ubuntu ISO testing tracker.

A list of all reports related to this bug can be found here:
http://iso.qa.ubuntu.com/qatracker/reports/bugs/1384095

[Nicholas Skaggs (nskaggs)]

Marius, I suspect this is an issue with the virtualization software and not ubiquity itself. I assume issuing a reboot via the reset button on the pc (or in this case, sending a poweroff or reboot signal) should cause the installation to reboot just fine and load into your new installation.

[Phillip Sz (phillip-sz)]

@4 No, I also could not shutdown after using the live-session. Probably the same problem. (On real hardware: http://phillw.net/hardware/7785845)

[Phillip Susi (psusi)]

This doesn't seem to have anything to do with ubiquity. To figure out where the problem is, we need you to attach your /var/log/syslog file. You should be able to boot the live cd and mount the hard disk to access the file and attach it to this report.

[Marius Gedminas (mgedmin)]

I don't think this is related to KVM: I can reboot fine in the installed system, it's just the livecd/installer session that hangs.
I also remember having this problem on real hardware back when I last installed Ubuntu. I regret not reporting it back then.

This is not a critical bug: hitting the reset button (or Alt+SysRq+S,U,B) reboots into the freshly-installed system fine.

Yeah, I'm sure it's not ubiquity's fault. It could be upstart maybe? Or the particular upstart config files present on the LiveCD? I filed this bug against upstart since the ISO testing tracker told me that's where to file bugs for problems that happen "during the installation itself".

I can reproduce this if I boot the .ISO in KVM, then Ctrl+Alt+F1 (i.e. Ctrl+Alt+2 to the QEMU monitor and 'sendkeys ctrl-alt-f1', then Ctrl+Alt+1 to go back to the VM) and 'sudo reboot'.

I had 'tail -f /var/log/syslog' on /dev/tty1 when I ran 'sudo reboot' on tty2. The reboot produced exactly zero additional messages before tail got killed.

I've no idea how to extract /var/log/syslog from the livecd session. AFAIU it lives in RAM and disappears as soon as I reboot. Maybe if I could dump the entire memory of the VM to a file and do a grep on it? Or if I knew how to set up remote syslog (assuming the network doesn't go down too soon)? Ideas welcome.

[Marius Gedminas (mgedmin)]

You dump the first gigabyte of the physical memory of a guest VM in KVM by Ctrl+Alt+2 and 'pmemdump 0 1073741824 memdump'

The only log message appearing there after I 'sudo reboot' is

    Oct 23 06:02:18 ubuntu-gnome rsyslogd: [origin software="rsyslogd" swVersion="7.4.4" x-pid="1022" x-info="http://www.rsyslog.com"] exiting on signal 15.

I enabled full SysRq with 'echo 1 > /proc/sys/kernel/sysrq' before attempting sudo reboot. Unfortunately alt-sysrq-t/w showed absolutely nothing.

Alt-SysRq+E produced the following messages on /dev/tty1:

    [ 150.506887] SysRq : Terminate All Tasks
    [ 150.507817] init: rc main process (3232) killed by TERM signal
    [ 150.508340] init: plymouth-shutdown main process (3297) killed by TERM signal

[Marius Gedminas (mgedmin)]

The reboot hangs in /etc/rc6.d/K11casper, in 'plymouth watch-keystroke': http://i.imgur.com/LGiRyOu.png (I added 'set -x' to the top of that file, and also to the top of /etc/rc6.d/K12reboot).

So Plymouth is running, but for some reason it's not displaying any graphics (nor text) and it's not reacting to keyboard input.

[Marius Gedminas (mgedmin)]

This is pretty much a duplicate of 966480.