The RESET means that the client and the server disagree on the state of the TCP connection that they set up. It usually means "you sent me a packet that claimed to be part of some connection but as far as I, the server, am concerned, that connection doesn't exist". The next step for me would be to look at the packets that lead up to the RESET to better characterize the disagreement. Perhaps it is actually the proxy that is messing things up.
A 60-second time-out is often an indication that a packet has been dropped. But that only makes sense if a packet was dropped in a way that would cause the client to try to send a packet. Or perhaps you have TCP keep-alives enabled?
To some extend, it should not be possible or at least not easy for the client software to cause a violation in the TCP protocol that would lead to a RESET. The closest scenario I was able to come up with for that would be the service sending the response and then closing the connection followed by the client, upon receiving the response (and perhaps even receiving the EOF in the server-to-client direction), trying to send something to the service. But that wouldn't explain the 60-second time-out before the RESET was received.
But you add a proxy into the mix and it might be:
That's obviously just a guess. It was the only scenario I could come up with that didn't involve some device in the mix violating the TCP protocol. (A simple dropped packet is not a protocol violation, but I couldn't come up with a scenario with a dropped packet that would cause the observed behavior and a dropped packet should not be this predictable anyway.)
- tye
In reply to Re: SOAP::Lite client request over HTTPS hangs until connection is reset by server (dig)
by tye
in thread SOAP::Lite client request over HTTPS hangs until connection is reset by server
by kiteboy
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |