boot freeze on non cloud

[Guy Taylor]

Installing cloud-init on a non cloud system causes the boot sequence to freeze with no error or warning.

[Pas (pasthelod)]

Hi, I suspect this is the same bug which causes (caused?) almost the same symptoms the only difference is that the last line is "init: ureadahead-other main process (***) terminated with status 4" ( http://ubuntuforums.org/showthread.php?t=1514521 -- this was probably bad blood between the ATi driver and plymouth-splash , https://bugs.launchpad.net/ubuntu/+bug/517770 https://bugs.launchpad.net/ubuntu/+bug/545536 ).

It's probably a deadlock/gridlock between plymouth processes and upstart. I've read about bugs in the plymouth-splash daemon, alas there's no easy way to just switch to an other mountall daemon that'd help eliminate possibilities for bugs like this.

[Guy Taylor (thebiggerguy)]

Managed to get a shell using http://upstart.ubuntu.com/wiki/OMGBroken
I think this is a mountall bug. Running 'stop mountall' then 'start mountall' made all the services start again including x11 (which made the screen go unreadable, I guess as openvt was running).
On restart the same freeze happened again, but 3 or 4 restarts later (after a guess of 30 restarts) fsck ran through plymouth successfully. So I think that defiantly rules out the fsck/plymouth bugs.

[Guy Taylor (thebiggerguy)]

!! Epic fail on my part !!

I have been developing on Amazon's ec2 recently and it seems I accidentally installed 'clout-init' on my system and not the one in the cloud. Anyway that is what broke the boot sequence.

I am redirecting this bug to cloud-init as I feel that any package that breaks the boot sequence without warning or error message is a problem.

[Guy Taylor (thebiggerguy)]

Old Bug Description for reference

During boot the system gets 'stuck' on the Plymouth loading screen. This is not due to fsck.

Dropping into text mode the last line is that the file system is clean 'Root: clean , */* files */* blocks'. The system reacts to ctrl-alt-del and starts safely running the shutdown script 'looking of unattended updates, de-configuring network, .... '.

Rebooting with '--verbose' kernel flag shows that the root fs on a MD raid is successfully assembled, mounted and checked for the need for fsck. The last thing the kernel does is run init. After this the same situation happens as above, all the correct init scripts are run with no errors. The system still reacts to events such as 'init: Handling usb-device-removed event' and even the screen enters and resumes power-saving. BUT the system never drops into any form of a shell on runlevel 1 or into GDM/X11.

Running the kernel with 'break=init' and entering the busybox shell. I can successfully list the root fs and run init, but the same situation as above happens.

Initially I thought it was due to nVidia's driver so i reinstalled without installing it and the same happens. The system does reboot successfully multiple times before I installed all the lucid updates (I started from 10.04-1).

In '--verbose' Init states that plymouth-splash 'exited normally', 'goal changed from start to stop', ....., 'post-stop to waiting' and 'init: Handling stopped event'. That is the last thing that is printed to the screen during boot intill a event is run.