mountall not detecting terminated fsck?

[Gabriel de Perthuis]

Binary package hint: mountall

My non grub2-migrated grub config contains this:
 kopt=root=/dev/mapper/lvm-whatever ro

mountall --debug appears to wait for / to be mounted, when it already is, except read-only. Once it is remounted rw, or the grub command is changed, mountall doesn't block anymore.

This is with mountall 0.2.0~boot3 .

[Scott James Remnant (Canonical) (canonical-scott)]

This is referring to a mountall package from the ubuntu-boot PPA not the archive

[Gabriel de Perthuis (g2p)]

Is there a more convenient place to report mountall bugs?

This version now got released to karmic with no changes, and the bug is still present.

[Scott James Remnant (Canonical) (canonical-scott)]

Please provide the --debug log output from the version in the archive

[Gabriel de Perthuis (g2p)]

Attached, with the two variations of boot options.

[Gabriel de Perthuis (g2p)]

LANG= apt-cache policy mountall
mountall:
  Installed: 0.2.1
  Candidate: 0.2.1
  Version table:
 *** 0.2.1 0
        500 ftp://ftp.free.fr karmic/main Packages
        100 /var/lib/dpkg/status
     0.2.0~boot5 0
        500 http://ppa.launchpad.net karmic/main Packages

[Scott James Remnant (Canonical) (canonical-scott)]

This looks like it's simply checking your disks, otherwise your log is incomplete. Let it run on longer

[Gabriel de Perthuis (g2p)]

I've just let it run a few hours, and nothing further happens. The system is idling anyway, the tips at http://upstart.ubuntu.com/wiki/OMGBroken get me a console, where pstree and top show nothing else happening, no filesystem checks.

The significant part of the logs, if you compare them, is that in the ro case mountall is waiting for /dev/mapper/raid-rootfs to get mounted on /, when it already is, albeit without the expected rw option.

[Scott James Remnant (Canonical) (canonical-scott)]

On Tue, 2009-10-13 at 10:15 +0000, Gabriel de Perthuis wrote:

> I've just let it run a few hours, and nothing further happens. The
> system is idling anyway, the tips at
> http://upstart.ubuntu.com/wiki/OMGBroken get me a console, where pstree
> and top show nothing else happening, no filesystem checks.
>
> The significant part of the logs, if you compare them, is that in the ro
> case mountall is waiting for /dev/mapper/raid-rootfs to get mounted on
> /, when it already is, albeit without the expected rw option.
>
Could you try 0.2.2 and see what that does?

If that doesn't work, get that shell, stop the running mountall (if
there is one), and then attach "mountall --debug" output (you can run it
from that shell again)

Scott
--
Scott James Remnant
<email address hidden>

[Gabriel de Perthuis (g2p)]

Same with 2.2, here's one more log.

[Scott James Remnant (Canonical) (canonical-scott)]

On Tue, 2009-10-13 at 17:34 +0000, Gabriel de Perthuis wrote:

> Same with 2.2, here's one more log.
>
> ** Attachment added: "mountall 2.2, restarted, ro /"
> http://launchpadlibrarian.net/33600910/mountall.debug.2.2.restarted-ro
>
Thanks, that's the end of the log?

If you don't pipe to a file, is the "fsck clean" line the last you see?

If you could follow the instructions on
http://upstart.ubuntu.com/wiki/OMGBroken to get a root shell alongside,
when it's "hung" like this, could you run "ps ax" to get the process
list.

Thanks,

Scott
--
Scott James Remnant
<email address hidden>

[Gabriel de Perthuis (g2p)]

Yes, it's the last thing I see. ps ax output is attached.

init-+-bash---pstree
     |-mountall
     |-sulogin
     |-udevd---2*[udevd]
     `-upstart-udev-br

[Scott James Remnant (Canonical) (canonical-scott)]

That's all that's running?

This looks like the SIGCHLD for the fsck hasn't been received, which is odd.

[Scott James Remnant (Canonical) (canonical-scott)]

This was caused by not clearing the progress when we actually reap the child; fixed in lucid