We have a Linux client that occasionally triggers a server error
message claiming that the client attempted to update the same
event record twice. The client logs show messages reporting that
TCP sessions timed out. The client is using Linux kernel level
2.4.20-28. It is our only Linux client at exactly this patch level.
Is this level of Linux known to have problems with TCP sessions
hanging up?

The TSM client code is at level 5.1.6.0. The TSM server is 5.2.2.0
running under mainframe Linux. There is a firewall between the
client and the server. We have a number of other clients working
through the same firewall with no sign of trouble.