This section includes information to help you understand the failover behavior of the Orchestration Server in a high availability environment.
If the primary node in the Orchestration Server cluster fails, you should see a job restart on another Orchestration Server in the cluster. The job must have been flagged as restartable. For more information, see Section 14.7, Testing the Failover of the Orchestration Server in a High Availability Grid.
When the Orchestration Server fails over to a new node, the Orchestration Agents reauthenticate with the new Orchestration Server instance.
If the Orchestration Server fails, any VM builds in progress are canceled and potentially incomplete residual artifacts of the build are cleaned up. When the Orchestration Server restarts or when it fails over in the cluster (the server operates identically in these scenarios), select jobs are run to determine the state of the grid. If the grid was set up with an audit database, the job running at failure time shows as canceled.
The Cloud Manager Monitoring Server and the Monitoring Agent are installed on the same server as the Orchestration Server in the high availability Orchestration Server cluster. The Monitoring Agent reports data to the Cloud Manager Monitoring Server.
The Monitoring Server and the Monitoring Agent services are made highly available along with the Orchestration Server and move between clustered machines as the Orchestration Server does. If a monitoring agent is installed on an Orchestration Server and if that server goes down, the server is displayed as “Down” in the Cloud Manager VM Client (Monitoring view).