On Wed, 2010-05-12 at 16:33 -0400, Brooks, Kenneth S wrote: > Thanks for the responses. > > In the case of the StaleChecking, we also have an IdleConnectionMonitorThread > that will evict connections that are over 20,000ms. > We can do this in our environment with relative confidence because we control > all the servers and those servers are set to close connections at 30,000ms. > That should help to limit the number of 'Connection Resets' we have. > > > So I guess it comes down to: if I wanted to remove my stalechecking and > solely rely on retry handler, what exception would I expect to see if the > connection was stale? > That way I can account for that in the retry handler. >
Beat me with a stick, but I just cant say for sure. I think it can be just a plain IOException, which basically means there is no way of telling I/O errors caused by a stale connection from I/O errors caused by connectivity issues. Oleg --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
