Raphael Coeffic writes:

> Now, it seems that you would like to have the next_hop behavior
> (routing without pre-loaded route), but only for the first
> request. Subsequent requests would be routed just normally (R-URI, or
> Route, if present). Is that correct?

raphael,

yes.  i would like to use outbound_proxy because it can hold domain
name, which in turn allows sems to use more than one outbound proxy when
the domain name maps to more than one A record.  my outbound proxy will
Record-Route the initial request and there is thus no need to force
outbound proxy for subsequent requests.

now that you have clarified the effect of outbound params, the behavior
is ok with me.

i tried to clarify the comments in sems.conf.sample (see below).

-- juha

-------------------------------------------------------------------------

# optional parameter: outbound_proxy=uri
#
# - this sets an outbound proxy for dialogs and registrations initiated 
#   by SEMS.  A preloaded Route header containing the uri is added to
#   each initial request.  The request is then sent to destination obtained
#   by resolving the uri. If outbound_proxy is not set (default setting),
#   no preloaded Route header is added and request is sent to destination
#   obtained by resolving r-uri.  Resolving is done by SIP stack with DNS
#   if uri contains domain name. Warning: If uri can not be resolved, no  
#   requests will be sent out at all!
#  
#   default: empty
#
# Example:
#   outbound_proxy=sip:proxy.mydomain.net

# optional parameter: force_outbound_proxy={yes|no}
#
# - forces SEMS to set outbound_proxy for any requests (not just for
#   registrations and dialog initiating requests). See above what setting
#   of outbound_proxy means.  This option will only have an effect if the
#   outbound_proxy option has been set, and it will break 3261 compatibility
#   in some cases; better use next_hop_ip/next_hop_port.
#
#   default: no
#
# Example:
#   force_outbound_proxy=yes
_______________________________________________
Sems mailing list
[email protected]
http://lists.iptel.org/mailman/listinfo/sems

Reply via email to