[Info-vax] TCPIP RWAST - FIN_WAIT_2

[Jeffrey H. Coffield]

I forked the thread when the topic went from RWAST to FIN_WAIT_2 as the 
problem.

This is an update for those interested.

Since the discovery of the FIN_WAIT_2 state, I have asked others in the 
company to check their servers and the symptoms show up on all servers 
we were able to look at. It's just not quite as bad so I expect they are 
timing out faster.

Apparently since we can't give the network people the exact setting that 
needs to be changed and they don't want to make changes anyway as they 
can't predict the consequences to other services that aren't reporting 
problems, they aren't going to do anything.

We are now looking at what actual services cause the most hung sockets 
and see if we can reduce the number of connections by some sort of 
connection re-use or pooling. With gigabit circuits on an Itanium, the 
time to start a connection has been too small to show any difference in 
performance when connection pooling was used.

Also on 14 systems at 5 sites, this is the only one showing the hung 
sockets.

I still can't find a way to see or change the equivalent of 
tcp_fin_timeout or kill a socket in the FIN_WAIT_2 state.


Jeff