I want to add a failover pair to an existing DMS.
TCP ports 8004 and 9004 are open, but HTTP and HTTPS ports are not.
To my understanding, HTTP and HTTPS are not really needed between the agents in a cluster, but not having it makes clustering a lot harder.
Is my understanding correct and what are the necessary steps to add this new failover pair to the cluster?
Do I need to manually change the DMS.xml file on all agents?
Agents in a DMS cluster communicate over port 8004 (by default).
When setting up that connection, the HTTP or HTTPS ports are used to detect what port the server is running on (typically 8004). If that detection fails, the agent should still try to set up connection using port 8004, so no extra configuration should be required.
However, there is a delay involved with the failing HTTP/HTTPS request, which will cause the connection attempt to be aborted because it's taking over 10 seconds. This is a known issue.
As workaround, you can enlarging the ConnectTimeout option to e.g. 60 seconds instead of the default 10. You can do this via SLNetClientTest > Advanced > Options > SLNet Options > ConnectTimeout. No agent restart is required. Note that the option value is agent-specific.