> I would like to know if really there isn't any liveness timeout control in
> the backend part. Otherwise, it would be very helpful that you describe which
> control method are you using
Once the 'back-end' connection to the proxied stream has been established (as a
result of a successful "DESCRIBE" command), the proxy server periodically (at
an interval randomly chosen between 30 and 60 seconds (usually)) sends a
command to the 'back-end' server, to test whether it's still alive. (This
command will be "OPTIONS", unless the server has specifically reported (in
response to a previous "OPTIONS" command) that it supports "GET_PARAMETER", in
which case the command will be "GET_PARAMETER".)
The proxy server acts, based upon the response to each 'liveness' command. If
the response is "OK", it goes ahead and sends another 'liveness' command later.
If the response is not "OK", or if it detects that the RTSP connection with
the back-end server has failed, then it assumes that the back-end server is
down, and it will then closes the back-end connection, and start again by
sending another "DESCRIBE". (These "DESCRIBE" commands are also repeated - at
increasing random intervals - until the server responds.)
What we don't do, however, is test whether the back-end server actually
responds to each 'liveness' command ("OPTIONS" or "GET_PARAMETER"). So far,
I've assumed that if the back-end server fails, it will do so by closing the
TCP connection, which we (the proxy) will eventually detect. The code
currently does not allow for the TCP connection staying alive, but the back-end
server simply failing to respond at all to a 'liveness' command. Is this
something that you are actually seeing happen??
Ross Finlayson
Live Networks, Inc.
http://www.live555.com/
_______________________________________________
live-devel mailing list
live-devel@lists.live555.com
http://lists.live555.com/mailman/listinfo/live-devel
