There is a production Sonic instance that was supposed to have a backup broker. When I checked, I discovered that wasn't the case. There is a backup broker on a different machine that was never added to the primary (there is no replication connection).
The installation and update guide mentions that once I add the backup to the primary I need to restart the primary broker. I need somebody to confirm this fact.
I tryied the scenario, Observed that primary container need to be restared, as any changes(like creating replication connection) to the broker will not affect the running broker until it is being restarted.
-Thanks and Regards,
Adding/removing replication connections are dynamic...hence does not require reloading/restarting the broker/container.
I am going to have to agree with Balraj Dacha. I have created a primary and backup (on a different machine) brokers and went through this scenario multiple times (to create an internal training doc) and I have noticed that the backup broker would go into a Waiting state indefinitely and it will only change states after I restart the primary broker.
I have attached a doc that shows exactly what I did.
I have removed the file; it's not needed
OK, here is the skinny on what I have discovered. My colleagues and I have noticed that sometimes the backup broker fails to synch up with the primary broker. After a little investigation, I discovered the issue was related to the replication connection and specifically with using the hostname vs. the IP address. Any machine that used a hostname name (recommended by the user guides) was always able to correctly negotiate the handshake between the 2 brokers. But for some machines (local dev) for which we had to use the IP addresses (due to some networking issues), the handshake failed every single time and I had to restart the primary broker.