Hi Daniel,

In an ideal world, DNS is, as you said, the easiest of approaches one can think of. However, not all endpoints support [properly] SRV failover. Take for example Asterisk's chan_sip. Is does resolve the hostname by SRV, however the failover is proven/accepted as broken.

https://stackoverflow.com/questions/41487243/dns-srv-failover-not-working-in-asterisk 

Cheers,
--Sergiu

On Mon, Oct 28, 2019 at 8:38 AM Daniel Tryba <d.tryba@pocos.nl> wrote:
On Mon, Oct 28, 2019 at 06:03:09AM +0100, Youssef Boujraf wrote:
> I am looking for failover service of kamailio servers looks like
> haproxy but udp & tcp.
>
> -??one domain name : sip.secure.com pointing my public ip address.
> - two kamailio server pointing same database content (user, whitelist,
> ...) no asterisk or pbx. Kamailio is used Only for signaling (users,
> whitelist & rtp)
>
> Monitoring service : identify who is the kamailio server online,
> offline and the ability to force??all the SIP/RTP/MANAGEMENT
> connection from Kamailio Srv1 to Kamailio Srv2 and vice-versa ?

Why do these kind of failover setups at all? When I create these
failover setups they work perfectly during tests but fail during real
failures. With round robin A records, NAPTR and SRV you can have
multiple live servers. Only the endpoints what actually use a hardcoded
ipaddress will have problems in the case of a failure.

The added benefit is scalability and easy maintenance.

The bad is there isn't a wiki entry on how to do it AFAIK.

BTW the shared database seems to be a bit deprecated (and a common point
of failure), look at DMQ if you have to start from scratch.

_______________________________________________
Kamailio (SER) - Users Mailing List
sr-users@lists.kamailio.org
https://lists.kamailio.org/cgi-bin/mailman/listinfo/sr-users