absolutely, yes. We had a horrible time over the last several months
controlling both the socket table and the inbound buffer tables. Several
patches were identified (some new as a result of our issues) and loaded on our
system. We seem to be running better now. These issues have been (mostly) due
to the changes HP made in 6.0 in relation to NS. They made the network services
more Berkeley Sockets compliant. As a result, applications that used to work
fine, now don't close sockets properly. When the socket table fills, you lose
connectivity.
Troy Dodsworth wrote:
> Has any one run into issues with connecting to a socket where the connect
> appears to be failing due the the lack of system resource ? Has any one run
> into problems with sockets connecting due to a number or socket closes which
> have not been flushed by the system yet and cause the system resource to be
> exhausted and the next connection attempt fails ?