Khalid Aziz wrote, on 29/04/10 06:11:
On Tue, 2010-04-27 at 11:33 +0000, Arthur Marsh wrote:
When I tried doing a restart with the official Debian i386 build of
kexec, the machine did a hardware restart.
I then added a line just after the call to do_stop in
/etc/init.d/kexec-load:
ls /boot/vmlinuz*
Then retried selecting restart from the kdm login menu.
Sometimes kexec worked fine and I booted without going to the BIOS
start-up and GRUB menu, on other occasions the ls command revealed that
the newer /boot/vmlinuz* files were absent and only some files normally
hidden underneath the /boot mount-point were visible.
So it appears that /boot was sometimes not mounted, even though the
message about unmount local file systems appeared much later in the
shutdown process.
Can you suggest any further debugging ideas?
Arthur.
The only thing I can think of right now is to insert "/bin/mount"
command before the ls in kexec-load script to see which filesystems are
still mounted when kexec-load runs. I am a little hesitant to suggest
that since mount command might hang depending upon the state of system
at that point. If you are willing to try it, we can find out if /boot
has already been unmounted when you see a kexec failure.
I've verified that /boot was not mounted when kexec-load was being run
when I tried a kexec reboot a short time ago.
As the machine had not been rebooted in several days /boot could have
been un-mounted by means unknown before the shut-down process.
I've added:
mount|grep boot
before and after the call to do_stop in /etc/init.d/kexec-load to see if
/boot gets un-mounted before, during or after the call to do_stop.
Perhaps if /boot is in /etc/fstab but /boot is not mounted at the time
of kexec-load running, kexec-load could issue a warning message?
Arthur.
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
