Sasmita,
Dialog replication is not meant to work on such a cluster. Since dialogs
retain more
state than registration, they are more sensitive to the IP of the
current node.
If you want high-availability for dialogs, the basic unit of deployment
shifts from
1 server to 2 servers (an active/backup pair). And in order to have the
same IP on
both servers, you must use VRRP-capable software, such as keepalived.
Deploy as many pairs as needed in each location according to your
traffic volume, and
make sure to give each pair its own Virtual (highly available) IP that
can automatically
move from the active node to the backup in case the former one catches fire.
PS: this information should be included in the dialog replication blog
post, so make
sure to read it closely.
Liviu
On 8/6/19 1:25 AM, Sasmita Panda wrote:
| id | cluster_id | node_id | url | state |
no_ping_retries | priority | sip_addr | flags | description |
| 1 | 1 | 1 | bin:192.168.0.x:5555 | 1 |
3 | 50 | 192.168.0.x | seed | Node A |
| 2 | 1 | 2 | bin:192.168.0.y:5555 | 1 |
3 | 50 | 192.168.0.y | seed | Node B |
| 3 | 1 | 3 | bin:192.168.0.z:5555 | 1 |
3 | 50 | 192.168.0.z | seed | Node C
_______________________________________________
Users mailing list
Users@lists.opensips.org
http://lists.opensips.org/cgi-bin/mailman/listinfo/users
