smbd crashes with SIGIO

Henrik Nordstrom hno at squid-cache.org
Mon Feb 13 17:14:56 GMT 2006


On Mon, 13 Feb 2006, Jeremy Allison wrote:

> Does this terminate with a core dump ? If so can you
> add a "panic action = /bin/sleep 999999" in the [global]
> section of your smb.conf attach with gdb and post a
> backtrace please ?

Not so sure a backtrace will help much here as SIGIO is an external async 
event.. but maybe.

SIGIO generally indicates there is a filedesriptor opened (or fcntl:ed) 
with the O_ASYNC flag and that I/O is now possible on that filedescriptor.

Another possibility is that you are holding a file lease (fcntl 
F_SETLEASE).

Or that an external process sent you a SIGIO.

Maybe more, but these are the cases I know of.

Regards
Henrik


More information about the samba-technical mailing list