Clustering Commander

For a high availability deployment, Commander is installed on two application servers behind a load balancer. Each installation connects to the same highly available Microsoft SQL Server, but they can't be connected simultaneously. The passive node activates when one or more failure indicators are observed. These failure indicators are the lack of a recent heartbeat record written to the database by the active node, an inability of the passive node to log in to the active node via the API, or a combination of both.

Microsoft Distributed File System (DFS) replicates key directories between the nodes in real time, to ensure the passive node is fully synchronized with the active node. These directories contain binaries such as scripts, reports, and log files.

Commander Clustering

When the service on the active node fails, the failover follows this process:

  1. The passive node, through its monitoring, determines that the active node is no longer functioning.
  2. The passive node initiates a quarantine of the active node.
  3. The passive node becomes the active node. It remains the active node even when the other node comes back online.
  4. The newly active node executes a script to shut down the failed node.

Commander Clustering

There will always be a small amount of downtime as the passive node becomes active. When the failover is planned, you can use the Service Portal Message of the Day to notify users of an upcoming outage window. For more information, see Display a message of the day in the Service Portal.