Package: autofs
Version: 5.0.8-1
Severity: important

I recently ran into a problem with autofs on our Jessie test machines where
autofs would fail to mount an NFS mount with the following error message

       'Too many levels of symbolic links'

Tests showed that while autofs failed to mount the NFS share, it was still
perfectly possible to mount the share manually and work without any issues.

After some research, I found a thread on LKML where someone else ran into
the exact same issue and after lots of discussion and debugging, he found
the colord daemon being the culprit as killing or restarting the daemon
fixed the issue.

colord was not responsible for the issue in my case, however I used the
hint provided in the aforementioned LKML thread and grepped through
the list of processes and their associated mount files under /proc.

It turned out that, in my case, cgmanager.service still had the mount
point listed in its mount file under /proc. I tried restarting
cgmanager.service through systemctl and the issue was resolved immediately.

During repeated occurences of the problem, I solely tried restarting
cgmanager.service and the issue was always resolved immediately:

> root@devourer:~> lsb_release -c
> Codename:      jessie
> root@devourer:~> cd /net/deb
> -bash: cd: /net/deb: Too many levels of symbolic links
> root@devourer:~> systemctl restart cgmanager.service
> root@devourer:~> cd /net/deb
> root@devourer:/net/deb>

The discussion in [1] suggets that this might be a bug in autofs and
they even come up with a possible solution, quoting:

> Perhaps autofs should use lookup_mnt()  to decide along this pattern:
>
> if ( dentry->d_flags & DCACHE_MOUNTED && lookup_mnt(path)  ) {
>   /* mounted */
> } else {
>   /* not mounted */
> }

But I am not really an expert on autofs and the VFS layer in the kernel,
so I might be completely wrong.

I set the urgency to "important", since the problem will probably
bite all autofs users once they upgrade to Jessie. It might even
be advisable to set the urgency to "grave" to make sure this gets
fixed before we release Jessie.

Cheers,
Adrian

> [1] http://permalink.gmane.org/gmane.linux.kernel.autofs/6915

-- System Information:
Debian Release: jessie/sid
  APT prefers testing
  APT policy: (500, 'testing'), (99, 'unstable'), (1, 'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 3.16-2-amd64 (SMP w/2 CPU cores)
Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) (ignored: LC_ALL 
set to en_US.UTF-8)
Shell: /bin/sh linked to /bin/dash

Versions of packages autofs depends on:
ii  libc6              2.19-12
ii  libxml2            2.9.1+dfsg1-4
ii  multiarch-support  2.19-12
ii  ucf                3.0030

Versions of packages autofs recommends:
ii  kmod               18-3
ii  module-init-tools  18-3
ii  nfs-common         1:1.2.8-9

autofs suggests no packages.

-- no debconf information


-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to