3.16 kernel in jwheezy-backports sometimes fails to boot on systems with LVM - needs newer udev
We (as in the company I work for) have a bunch of wheezy boxes on which
a 3.16 wheezy-backports kernel is installed:
ii linux-image-3.16.0-0.bpo.4-amd64 3.16.7-ckt25-2~bpo70+1
ii linux-image-amd64 3.16+63~bpo70+1
Sometimes (not always) a box fails to boot and ends up in the
(initramfs) prompt. We found out that this only happened on boxes with
LVM, and it appears to be a timing issue. If you run the 'lvm2' script
by hand, the system resumes booting:
initramfs) cd scripts/local-top
(initramfs) ls -l
total 8
-rw-r--r-- 1 0 0 70 Apr 6 12:06 ORDER
-rwxr-xr-x 1 0 0 1011 Aug 17 2009 lvm2
(initramfs) . lvm2
Begin: Running /scripts/local-pr[331080.436057] PM: Starting manual
resume from disk
emount ... done.
[331080.453787] EXT4-fs (dm-0): INFO: recovery required on readonly
filesystem
[331080.460802] EXT4-fs (dm-0): write access will be enabled during recovery
[331080.548239] EXT4-fs (dm-0): orphan cleanup on readonly fs
[331080.553916] EXT4-fs (dm-0): 3 orphan inodes deleted
[331080.558928] EXT4-fs (dm-0): recovery complete
[331080.565866] EXT4-fs (dm-0): mounted filesystem with ordered data
mode. Opts: (null)
Begin: Running /scripts/local-bottom ... done.
done.
Begin: Running /scripts/init-bottom ... done.
INIT: version 2.88 booting
One of my collegues found out that by installing the 'udev' from
wheezy-backports, the problem goes away. It seems that a newer kernel
needs a newer udev. Or perhaps it is a heisenbug that gets masked by
installing a newer udev.
So it seems to me that if indeed the 3.16 kernel needs a newer udev,
linux-image-amd64 from wheezy-backports should Depend on udev from
backports, right?
Thanks!
Mike.
Reply to: