On Mon, Apr 12, 2010 at 07:59:53AM -0500, Jon Schewe wrote:
> /proc/sys/net/ipv4/tcp_keepalive_time:7200
> > netstat -to
> Client:
> tcp        0      0 client:9102       server:54043      ESTABLISHED
> keepalive (7196.36/0/0)

That's strange. It should've been the timeouts you specified in
config files, not 7200 seconds (two hours) which is system default.

It looks like bacula does not use TCP_KEEPIDLE setsockopt(2) on your
system. You might want to report a bug on http://bugs.bacula.org/

IMHO, it should work there. Or if not, it should probably throw a
warning if you try to use it and it is not supported or fails.

Apart from fixing bacula, you can override system default, for
example (on both server and client) do :

echo 60 > /proc/sys/net/ipv4/tcp_keepalive_time

(or edit /etc/sysctl.d/* or /etc/sysctl.conf to retain value across
reboots). Can you try what "netstat -to" says after you lower that
limit and rerun backups ? 

If "netstat -to" then reports smaller timers (60 or less), than it
should fix your problem, so you can try turning accurate back to yes.

Does that help ?

------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users