[Samba] msdfs & full_audit issues

[Jeremy Allison]

On Sun, Feb 23, 2014 at 12:09:13PM -0500, Bill Baird wrote:
> I'm having issues enabling VFS full_audit & msdfs. It works great on
> my non-root DFS servers. On my DFS root server, enabled full_audit
> will break DFS links.
> 
> If found some reports of this from searching, but the users appear to
> have resolved the issue from upgrading (to older versions that I am
> using). I'm running Samba 3.6.9-167 on CentOS 6.
> 
> Here is an except from the samba server,
> 
> Feb 23 08:59:01 smbd:  [2014/02/23 17:00:40.578838,  0]
> lib/fault.c:372(dump_core)
> Feb 23 08:59:01 smbd:    dumping core in /var/log/samba/cores/smbd
> Feb 23 08:59:03 smbd:  [2014/02/23 17:00:43.027753,  0]
> lib/fault.c:47(fault_report)
> Feb 23 08:59:03 smbd:
> ===============================================================
> Feb 23 08:59:03 smbd:  [2014/02/23 17:00:43.028276,  0]
> lib/fault.c:48(fault_report)
> Feb 23 08:59:03 smbd:    INTERNAL ERROR: Signal 11 in pid 13946
> (3.6.9-167.el6_5)
> Feb 23 08:59:03 smbd:    Please read the Trouble-Shooting section of
> the Samba3-HOWTO
> Feb 23 08:59:03 smbd:  [2014/02/23 17:00:43.028757,  0]
> lib/fault.c:50(fault_report)
> Feb 23 08:59:03 smbd:    From: http://www.samba.org/samba/docs/Samba3-HOWTO.pdf
> Feb 23 08:59:03 smbd:  [2014/02/23 17:00:43.029244,  0]
> lib/fault.c:51(fault_report)
> Feb 23 08:59:03 smbd:
> ===============================================================
> Feb 23 08:59:03 smbd:  [2014/02/23 17:00:43.029551,  0]
> lib/util.c:1117(smb_panic)
> Feb 23 08:59:03 smbd:    PANIC (pid 13946): internal error
> Feb 23 08:59:03 smbd:  [2014/02/23 17:00:43.034331,  0]
> lib/util.c:1221(log_stack_trace)
> 
> In Windows, the error for the client trying to click a dfs link is:
> 
> X:\'dfs share name' is not accessible
> The specified network name is no longer available.
> 
> A copy of my share configuration is below.
> 
> Any help would be greatly appreciated!

Can you ensure you've built with symbols
then add to your [global] section of
your smb.conf:

panic action = /bin/sleep 99999999

Now when the panic happens you should be able
to attach using gdb to the parent of the sleep
process (which should be an smbd) and get a
full backtrace.

Jeremy.