Summary
When running DSE nodes that use two different network interfaces to separate client traffic from internode traffic, the OpsCenter agent will fail to establish a STOMP connection.
Applies to
OpsCenter 6.0.10, 6.1.2 and 6.1.3
Symptoms
OpsCenter will initially show all agents as healthy, but after a few minutes it will show that the STOMP connections for the agents are down.
Cause
A change was made in OpsCenter 6.0.10/6.1.2 that changes the IP address that the agent uses to identify messages to create unique message ids for dense nodes. In clusters with two network adapters, this causes the agent to identify messages with the wrong IP, resulting in the STOMP connection from the agent being flagged as down.
Workaround
No workaround exists.
Solution
Upgrade to OpsCenter 6.1.4 or 6.0.11.
If you have multiple nodes and agents on the same physical machine, you may start to see issues with backups. This can be resolved by restarting all of the agents on that physical machine.