Hi,
I'm having problems with OpenSER forwarding calls to a domain where there are only SRV records.
OpenSER seems not to perform any balancing or failover when a target server is unavailable or gives an error.
I wish to forward calls to a number of onward gateways that at times may not respond at all, or may respond with something like "503 Service Unavailable".
The actual destination address is derived from an enum lookup, but it may be for example 123456@example.com
There are SRV records as follows:
_sip._udp.example.com. 600 IN SRV 10 50 5060 gw1.example.com. _sip._udp.example.com. 600 IN SRV 10 50 5060 gw2.example.com.
The behaviour I'd expect/like would be that OpenSER will distribute the calls between the two targets specified in the SRV records, and will fail over to the other target if one fails to respond or gives a fatal error such as a 503.
What actually happens, is that the calls all go to the first one regardless.
Are my expectations reasonable?
Does OpenSER's SRV record implementation work like this or am I misled?
Kind regards,
James
Hi James,
Openser does not do balancing over records with same priority - it will just use the first (as returned by DNS server).
But failover based on NAPTR and SRV records is in place.
Regards, Bogdan
James Holden wrote:
Hi,
I'm having problems with OpenSER forwarding calls to a domain where there are only SRV records.
OpenSER seems not to perform any balancing or failover when a target server is unavailable or gives an error.
I wish to forward calls to a number of onward gateways that at times may not respond at all, or may respond with something like "503 Service Unavailable".
The actual destination address is derived from an enum lookup, but it may be for example 123456@example.com
There are SRV records as follows:
_sip._udp.example.com. 600 IN SRV 10 50 5060 gw1.example.com. _sip._udp.example.com. 600 IN SRV 10 50 5060 gw2.example.com.
The behaviour I'd expect/like would be that OpenSER will distribute the calls between the two targets specified in the SRV records, and will fail over to the other target if one fails to respond or gives a fatal error such as a 503.
What actually happens, is that the calls all go to the first one regardless.
Are my expectations reasonable?
Does OpenSER's SRV record implementation work like this or am I misled?
Kind regards,
James
Users mailing list Users@lists.openser.org http://lists.openser.org/cgi-bin/mailman/listinfo/users
But a DNS server change alternatively the order of DNS SRV answers of same priority! So it should work! For example with bind9:
dig -t SRV _mediaproxy._tcp.private.wifirst.net
;; ANSWER SECTION: _mediaproxy._tcp.private.wifirst.net. 3600 IN SRV 0 30 25060 bob.wifirst.net. _mediaproxy._tcp.private.wifirst.net. 3600 IN SRV 0 30 25060 john.wifirst.net.
and alternatively:
;; ANSWER SECTION: _mediaproxy._tcp.private.wifirst.net. 3600 IN SRV 0 30 25060 john.wifirst.net. _mediaproxy._tcp.private.wifirst.net. 3600 IN SRV 0 30 25060 bob.wifirst.net.
Bogdan-Andrei Iancu a écrit :
Hi James,
Openser does not do balancing over records with same priority - it will just use the first (as returned by DNS server).
But failover based on NAPTR and SRV records is in place.
Regards, Bogdan
James Holden wrote:
Hi,
I'm having problems with OpenSER forwarding calls to a domain where there are only SRV records.
OpenSER seems not to perform any balancing or failover when a target server is unavailable or gives an error.
I wish to forward calls to a number of onward gateways that at times may not respond at all, or may respond with something like "503 Service Unavailable".
The actual destination address is derived from an enum lookup, but it may be for example 123456@example.com
There are SRV records as follows:
_sip._udp.example.com. 600 IN SRV 10 50 5060 gw1.example.com. _sip._udp.example.com. 600 IN SRV 10 50 5060 gw2.example.com.
The behaviour I'd expect/like would be that OpenSER will distribute the calls between the two targets specified in the SRV records, and will fail over to the other target if one fails to respond or gives a fatal error such as a 503.
What actually happens, is that the calls all go to the first one regardless.
Are my expectations reasonable?
Does OpenSER's SRV record implementation work like this or am I misled?
Kind regards,
James
Users mailing list Users@lists.openser.org http://lists.openser.org/cgi-bin/mailman/listinfo/users
Users mailing list Users@lists.openser.org http://lists.openser.org/cgi-bin/mailman/listinfo/users
Hi Marc,
If the DNS server does this, yes indeed, it should work.
regards, bogdan
Marc LEURENT wrote:
But a DNS server change alternatively the order of DNS SRV answers of same priority! So it should work! For example with bind9:
dig -t SRV _mediaproxy._tcp.private.wifirst.net
;; ANSWER SECTION: _mediaproxy._tcp.private.wifirst.net. 3600 IN SRV 0 30 25060 bob.wifirst.net. _mediaproxy._tcp.private.wifirst.net. 3600 IN SRV 0 30 25060 john.wifirst.net.
and alternatively:
;; ANSWER SECTION: _mediaproxy._tcp.private.wifirst.net. 3600 IN SRV 0 30 25060 john.wifirst.net. _mediaproxy._tcp.private.wifirst.net. 3600 IN SRV 0 30 25060 bob.wifirst.net.
Bogdan-Andrei Iancu a écrit :
Hi James,
Openser does not do balancing over records with same priority - it will just use the first (as returned by DNS server).
But failover based on NAPTR and SRV records is in place.
Regards, Bogdan
James Holden wrote:
Hi,
I'm having problems with OpenSER forwarding calls to a domain where there are only SRV records.
OpenSER seems not to perform any balancing or failover when a target server is unavailable or gives an error.
I wish to forward calls to a number of onward gateways that at times may not respond at all, or may respond with something like "503 Service Unavailable".
The actual destination address is derived from an enum lookup, but it may be for example 123456@example.com
There are SRV records as follows:
_sip._udp.example.com. 600 IN SRV 10 50 5060 gw1.example.com. _sip._udp.example.com. 600 IN SRV 10 50 5060 gw2.example.com.
The behaviour I'd expect/like would be that OpenSER will distribute the calls between the two targets specified in the SRV records, and will fail over to the other target if one fails to respond or gives a fatal error such as a 503.
What actually happens, is that the calls all go to the first one regardless.
Are my expectations reasonable?
Does OpenSER's SRV record implementation work like this or am I misled?
Kind regards,
James
Users mailing list Users@lists.openser.org http://lists.openser.org/cgi-bin/mailman/listinfo/users
Users mailing list Users@lists.openser.org http://lists.openser.org/cgi-bin/mailman/listinfo/users
-
Bogdan-Andrei Iancu -
James Holden -
Marc LEURENT