[Samba] What could cause an "unclean shutdown" of process?

[Frederic Fournier]

Hello,

I added "printcap cache time = 0" to my smb.conf and the printcap related
messages disappeared from my log. Thank you very much for that!!

But the 'unclean shutdown of PID xxx' was not related to printcap. (If you
look in the log, you see that it happens 12 minutes after the last printcap
rescanning and one minutes before the next.)
I made more tests, and the programs that are running on the windows clients
keep crashing. The only two things that I see in samba logs are these
'unclean shutdown of PID xxx' messages in log.smbd and a sudden stop in a
long long list of transactions in log.<client_name>. There are minimal logs
on the client side where I can see: "Error caused by: java.io.IOException:
The specified network name is no longer available".
(The client error seems to be logged 1 second after the server error, but
their clock might not perfectly synchronized, so I can't say for sure what
happens first.)

I need to fix this problem, but I don't know how. So far, Google has
brought me to similar problems (e.g.
http://lists.samba.org/archive/samba/2010-August/157948.html), but to no
solution :(

Any help would be greatly appreciated!

Frederic





On Wed, Aug 29, 2012 at 12:24 PM, David Disseldorp <ddiss at suse.de> wrote:

> Hi Frederic,
>
> On Wed, 29 Aug 2012 11:58:38 -0400
> Frederic Fournier <frederic.bioinfo at gmail.com> wrote:
>
> > Thanks for the tip, I was also looking for a way to get rid of those
> > Printcap messages that pops up every 13 minutes. (especially since there
> is
> > no printer on this server).
> >
> > But I only included those messages to give a bit of context: its the
> three
> > messages in the middle that cause me trouble:
> > smbd/server.c:292 Unclean shutdown of pid 32526
> > Scheduled cleanup of brl and lock database after unclean shutdown*
> >  Cleaning up brl and lock database after unclean shutdown*
> >
> > Do you know how I could find what causes the server to do an unclean
> > shutdown of one of its process (and how to prevent it)? Because when
> there
> > is such an unclean shutdown, there is also a crash on the client side.
> (Or
> > maybe it's the other way around?) How could I find more about it?
>
> The 'cleanup...after unclean shutdown' messages are also caused by the
> printcap process. It does not register it's pid which results in
> the parent smbd process complaining on SIGCHLD when the printcap process
> exits.
>
> Cheers, David
>