User confirmation for fail-over in Controller Cluster and Agent Cluster

Current Situation

• A number of users operate their own cluster systems (hardware-based cluster, software-based cluster) that provide their own mechanisms for fail-over and switch-over. In the simplest case, we find Node 1 and Node 2 as an active-passive cluster. The user’s cluster determines which node is active.
• This conflicts with the JS7 cluster, which decides for itself which node should be active. A conflict arises when a user’s applications and JS7 products must run together on either only Node 1 or only Node 2.

Desired Behavior

• Behavior in case of fail-over is configurable: options to automatically fail-over (default) and to require confirmation for fail-over are offered.
• If user confirmations are enabled, then the Proxy Service in JOC Cockpit will request user confirmation for fail-over, similar to a situation when an active Controller instance becomes unavailable without the Proxy (ClusterWatch) being present.
  • Users can use the existing “confirm node loss” operation to confirm fail-over.
  • The functionality is available from the GUI and from the REST API.
• The functionality to confirm fail-over is configured with JOC Cockpit settings and is applied to both Controller Cluster and Director Agent Cluster.