On Fri, 2010-01-29 at 14:36 -0800, KARR, DAVID (ATTCINW) wrote: > Using HttpClient 3.0.1 (no, I don't have any choice at this time.) > > I just did a couple experiments, making sure that if I set a non-zero > value for either the socket timeout or the connection timeout, that it > would actually affect the results. If I set my socket timeout to 1 > seconds (1000 millis), several of my slow requests did fail with a > timeout. That's good. I then tried setting my connection timeout to 1 > second, and it didn't make any tests fail. I then set it to 1 > millisecond, and still no tests failed. I find it hard to believe I'm > getting the connection setup done in less than 1 millisecond. > > Is perhaps the "connection manager timeout" something different from the > protocol connection timeout?
Yes, it is. Connection manager timeout defined the maximum period of time a thread can remain blocked waiting for a connection to be leased from the connection pool. Hope this helps Oleg --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
