The other issue is that if you try to use a stale connection to make a POST/PUT request (i.e. entity enclosing) then that data will not necessarily be repeatable, so a retry would fail here.
Thanks, Sam 2009/8/4 Oleg Kalnichevski <[email protected]>: > On Mon, Aug 03, 2009 at 03:33:30PM -0700, Ken Krugler wrote: >> Hi all, >> >> On http://hc.apache.org/httpcomponents-client/tutorial/html/ch02.html >> the section titled "Connection keep alive strategy" talks about when to >> implement a custom keep-alive strategy. >> >> But I thought that if a connection gets dropped by the server due to >> inactivity, the next request made using that stale connection will >> trigger an IOException, which HttpClient will automatically handle by >> re-trying with a new connection. >> >> So is the issue that retrying winds up being less efficient, versus >> auto-closing connections? Or is my understanding of what HttpClient 4 >> will do with a stale connection too optimistic? >> >> Thanks, >> >> -- Ken >> >> > > Ken, > > It is not necessarily less efficient but some folks just do not like the idea > of getting an exception. Besides, if combined with a watcher thread evicting > expired connection from the pool, ConnectionKeepAliveStrategy helps ensure > connections get closed as they expire, instead of sitting in the pool > indefinitely waiting for a request to be made. > > Overall, it just makes things cleaner. > > Oleg > > >> -------------------------- >> Ken Krugler >> TransPac Software, Inc. >> <http://www.transpac.com> >> +1 530-210-6378 >> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [email protected] >> For additional commands, e-mail: [email protected] >> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
