Hi,
I just wanted to raise this to see if anyone can see any flaw/problem in my thought processes.
When using outbound it is mandatory to have some form of SIP keep-alive between the UA and the edge proxy. However, in the case where SIP over WebSocket is used the WebSocket transport has its own ping/pong keepalive mechanism.
So to my mind, SIP keep-alives are optional when WebSocket is in use?
In my last email I asked about the flow timer and said that I thought it should be set in conjunction with the tcp_connection_lifetime value, but when WebSocket is in use I believe:
* tcp_connection_lifetime on the edge proxy should be slightly larger than * websocket:keepalive_timeout on the edge proxy, which should be slightly larger than * the flow timer value the registrar sets in 2xx responses to REGISTERs
Regards,
Peter
Peter Dunkley writes:
When using outbound it is mandatory to have some form of SIP keep-alive between the UA and the edge proxy. However, in the case where SIP over WebSocket is used the WebSocket transport has its own ping/pong keepalive mechanism.
the whole section 6 of the websocket draft is a mess. since websocket runs over tcp, it should use normal tcp keepalive like any other outbound tcp ua.
So to my mind, SIP keep-alives are optional when WebSocket is in use?
section 6 is non-normative. if websocket guys want to change what outbound rfc tells, they have to write an rfc about it.
-- juha
Actually, on re-reading RFC 5626 it is perfectly OK to use WebSocket keep-alive instead of SIP keep-alives.
Section 3.5 states, "UAs and Proxies are also free to use native transport keep-alives;"
Regards,
Peter
Peter Dunkley writes:
When using outbound it is mandatory to have some form of SIP keep-alive between the UA and the edge proxy. However, in the case where SIP over WebSocket is used the WebSocket transport has its own ping/pong keepalive mechanism.
the whole section 6 of the websocket draft is a mess. since websocket runs over tcp, it should use normal tcp keepalive like any other outbound tcp ua.
So to my mind, SIP keep-alives are optional when WebSocket is in use?
section 6 is non-normative. if websocket guys want to change what outbound rfc tells, they have to write an rfc about it.
-- juha
sr-dev mailing list sr-dev@lists.sip-router.org http://lists.sip-router.org/cgi-bin/mailman/listinfo/sr-dev
-
Juha Heinanen -
Peter Dunkley