New Features › CA Process Automation Release 04.2.00 › Clustering Improvements

Clustering Improvements

The following summary focuses on clustering improvements.

• Removal of primary node dependency in clustered Domain Orchestrator
  • Formerly, dependencies on the primary node (first node installed) represented a potential single point of failure. That is, if the primary node went down, another node in the cluster did not take on the activities being performed by the primary node. This prevented the expected high availability of the Domain Orchestrator. Activities that used to depend on an active Domain Orchestrator primary node included: (1) mirroring newly deployed JAR files, (2) installing agents or other Domain Orchestrator nodes, and (3) deploying files uploaded as User Resources or Reports.
  • Now, there are no dependencies associated with the primary node in a clustered Domain Orchestrator. A clustered Domain Orchestrator now functions with high availability.
• Implications of removal of primary node dependency on F5 iRule
  • Formerly, the iRule included the variables MyPool, PrimaryIP, and PrimaryPort. PrimaryIP and PrimaryPort referred to the primary node of the Domain Orchestrator.
  • Now, MyPool is the only iRule variable.
• Ability to configure a new Domain with existing domain settings and certificates
  • Formerly, the Domain.xml and certificates were present on the file system only. So, if the Domain went down, the configuration and certificates were lost.
  • Now, the Domain.xml and certificates are moved to a central database. This move eliminates the single point of failure if the Domain Orchestrator goes down. It also improves performance when accessing agent data from the UI.