[lxc-users] Container starts on incorrect runlevel

Davide Baldini baldiniebaldini at gmail.com
Thu Oct 15 13:22:37 UTC 2015


I actually don't care about micro-management of individual containers, such as 
fstab or fsck, as long as they work and do their duty. That's one reason why I 
moved away from qemu-KVM.

> Wouldn't an fsck that runs when an fs is mounted read-write potentially cause an error to the fs?

But that is scaring. Issuing
     mount -o remount,ro -t ext4 /dev/md0 /
from inside the guest after it has fully booted, results in:
     mount: / is busy
so it doesn't touch the fs. However, starting the container in foreground, with
     lxc-start -F -P /media/raid1/ -n ve-200
executes fsck during boot, as part of its rcS.d scripts, and prints on console:
     Checking file systems...fsck from util-linux 2.20.1
     done.
     Mounting local filesystems...done.

The filesystem is an ext4 on block device /dev/md0, mounted on /media/raid1, and 
indeed is already mounted rw when the guest boots. The fs is also shared with 
other containers, and mounted/used by the host, too.

Do you know if the guest actually has the power to alter this filesystem, in 
case fsck wants to? That would seriously be a problem.

-- Davide


More information about the lxc-users mailing list