I am using HttpClient 3.1 server-side to connect to web services on the
intranet.
I am using MultiThreadedHttpConnectionManager and reusing HttpClient instance.
When I am simulating web services and they respond instantly, everything works
as expected and I can stress-test the application.
However, once I start adding delay in some of the services, everything
deteriorates.
This is what is happening in the production and as a result, the whole
application stalls.
I have set the SO and connection timeout to 3 seconds. I simulate the service
that takes more than 3 seconds to respond. I would expect the delayed service
request to time-out allways after 3 seconds (or neighbourhood). In practice,
the
time-out time starts to increment with increased memory consumption and as the
result the server is overwhelmed.
Why is this happening? Why services do not time-out in period set?
Appreciate your help!
Dan
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
