On Thu, 2008-06-05 at 11:30 +0200, Martin Zdila wrote: > > > SocketInputStream.socketRead0(FileDescriptor, byte[], int, int, int) > > > line: not available [native method] [local variables unavailable] > > > > The thread is blocked in the socket read operation. This is perfectly > > normal. If you do not want the operation to block indefinitely you > > should set a timeout for it. > > But that read operation never ends even if the remote http server is normally > responsive.
The client gets blocked in a _native_ method managed by the JVM. It is entirely out of HttpClient control. If the method does not return, that means that the server is not sending any data. > And this occurs only when I see "I/O exception" in the logs = > when the request is retried. > > 1. i don't think that "I/O exception" is normal as network connection to > server is perfect This may be a bug in the JRE but I seriously doubt that > 2. if this "I/O exception" occurs the thread should not halt. > 3. i can easily reproduce that problem. when i use my post form and wait cca > 30 sec then this issue allways appears. > > > Have you tried doing the same test without the proxy? It can well be a > > proxy side problem. > > There is small misunderstanding. That proxy is my application = webserver > receiving HttpServletRequest and forwarding them through HTTP client :-). > > So I think there is bug in HTTP client. I'll try to prepare a testcase ;-) > Sure thing. Oleg > cu > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
