On 01/20/2016 10:48 AM, Rob Godfrey wrote:
What I get from this is that the C++ client gives up retrying when it gets a 320 (connection forced) reply from the broker.
Indeed, the retry gives up when it gets an explicit close rather than the socket failing.
As Keith mentioned earlier, the error message is basically saying "Yes, the broker is up... but it isn't fully configured yet", and the 320 error code (as defined by the AMQP 0-10 spec) says "The client may retry at some later date."... so if the C++ client would continue retrying on receipt of this error then you would establish a connection. I guess this would require a C++ client change, but Gordon would be the one to know here...
As 0-10 did not define a code for authentication failure, 320 has been used for that. It is a rather broad definition 'An operator intervened to close the connection for some reason'. Since it would not always be desired/appropriate to retry on receiving an explicit code, any change would need to be configurable.
Is there a reason the broker wants to accept connections before it is fully configured for use?
--------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@qpid.apache.org For additional commands, e-mail: users-h...@qpid.apache.org
