When you use hash over username kamailio on reload creates the distribution
lists I.e:
1- server5
2- server3
3- server8
Now, when a call from userB arrives, it will distribute it for instance to
item 2, server3. And any other whose hash is about the same goes to the
same.
Now, If server3 fails,
I’m not using those algorithms, I’m using hash over username.
On Mon, 8 Aug 2022 at 22:34, Henning Westerholt wrote:
> Hello,
>
>
>
> I still might not 100% understand your scenario.
>
>
>
> If, on the other hand, we recompute (reshuffle) the list ignoring the
> inactive servers, the traffic wou
Hello,
I still might not 100% understand your scenario.
If, on the other hand, we recompute (reshuffle) the list ignoring the inactive
servers, the traffic would be properly evenly distributed. This is what I’m
looking for.
But does dispatcher e.g. with a hash based distribution or round-robin
Difference is if you are distributing based off the username hash and a
server goes down, it is simply set as “inactive”, but all calls that were
going to that serve will now go to the next in the list, so all those calls
will land on the next server basically doubling the traffic.
If, on the othe
With either module (Dispatcher or Carrierroute) , what would be the functional
difference between 'reloading' and 'recalculating'?
Both modules can be controlled via RPC:
Dispatcher can be set to send OPTIONS messages to the hosts in it's tables to
monitor endpoint status and dynamically bring
Hey Henning, thanks for replying. Quick question: Adding (or removing) in
memory recalculates without reloading?
On Mon, 8 Aug 2022 at 19:47, Henning Westerholt wrote:
> Hello,
>
>
>
> this is possible with carrierroute by RPC command:
> https://kamailio.org/docs/modules/5.5.x/modules/carrierrou
Hello,
this is possible with carrierroute by RPC command:
https://kamailio.org/docs/modules/5.5.x/modules/carrierroute.html#carrierroute.rpc.replace_host
With dispatcher you could of course just adding one host two times by some
script and reload the module, as a simple idea.
Cheers,
Henning
Hello guys,
Is it possible to have dispatcher recompute the distribution list? i.e.
one server goes down, all traffic for that server goes to the next one,
doubling traffic on that one. Is there a way of, when a host becomes
unreachable, recompute the whole list?
Regards,
David Villasmil
email:
