Agent cluster

Current Situation

• Agents can execute jobs on different servers in parallel by use of Subagents for horizontal scalability, see JS-1954. This provides high availability in case that a Subagent or its server fails.
• However, there are no fail-over capabilities in case that an Agent Director fails.

Feature

• See feature description JS-1954
• The Agent Director is added the fail-over capabilities as for the JS7 Controller.
  • The Agent Director implements a passive cluster with an active node and a standby node.
  • Each Agent Director cluster node holds a journal that is synchronized between nodes.
  • The Controller implements a proxy service to access the currently active Agent Director node.
  • An automated fail-over occurs in case that the active Agent Director node becomes unavailable. A fail-back (switch-over) can be initiated by a command forwarded by the Controller to the active Agent Director node.