ghrst
commented
5 years ago Hi,
-
Fortunately, your architecture can be implemented using Ansible scripts in this repository. However, to make it work on Debian Buster (I tested it on Debian Stretch), you might need to change a few things
-
This repository uses Corosync, and Pacemaker, which are old, stable, and very well-documented. Corosync provides a heart-beat mechanism to identify nodes in a cluster. On top of this layer Pacemaker can understand your services, e.g VirtualIP, Kamailio, etc. In order to understand how things work you should refer to Pacemaker's documentation at https://clusterlabs.org/. There is also a book about this topic named Pro Linux High Availability Clustering. However, I think the documentation is good enough.
-
If you install a local RTPProxy(or equivalently RTPEngine) instance, then it will become a SPoF. Both by it's capacity to handle concurrent calls, and by stopping. Using a cluster of RTPProxy instances you increase the availability, and capacity of your service.
-
Notice that this ansible repository only sets-up, and installs Kamailio, and RTPProxy. You can record media, but for recording you should refer to RTPEngine's documentation.
yboujraf
Dear,
I got your reference from Mr Daniel-Constantin Mierla.
I was looking for a kamailio failover setup running on Debian 10 Buster & kamailio 5.3 (without docker at this stage) :
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 ?
Questions :
Do you have any diagram how it works ?
A two node Active-Passive Kamailio Cluster with a local RTPProxy instance
A two node Active-Passive Kamailio Cluster with a cluster of RTPProxy servers
What the advantage of the two solutions you are proposing ?
RTPENGINE :
We would like to use RTPENGINE to record audio on file (Mix of caller and callee or seperated)
Best Regards, Youssef