Gentlemen,

I see duplication jobs continously die with status 26 from one of my media servers. Duplication is going from disk over WAN to another media server to tape.

Source media server is "sitting" behind a firewall. (Firewall in between duplication source media and master server but no firewall in between target media server and master.)

Duplication jobs work fine but always break after around 2 hours (2:15 h exactly).

From the bpdm log on the source media server I can see that the remote host (the master server) has closed the connection.

On the master server I cannot seem to find meaningfull log entries, neither in the bpdbm log nor in the vnetd log. I assume it's a timeout when the source media server is sending meta data to the master when a fragment or backup image has been completed.

Media Servers are on Windows Server 2003, Master is running on Solaris 10 Sparc.

NetBackup version is 6.5.4

I cannot remember or find a sensible timeout for my 7200 seconds (2h) finding. It's not even a sensible multiple of anything, is it?

I cannot see the firewall closing an idle connection (my first guess).

Is there a TCP/IP setting on Solaris that could cause the connection being closed?

Any hint is welcome.

Kind regards,
Thomas