Primus HSM - Performing Maintenance and Configuration Change
In a High Availability (HA) cluster, maintenance on a single device should be performed seamlessly without disrupting connected client applications. To ensure uninterrupted service, client connections must be gracefully transitioned away from the device undergoing maintenance. This is achieved by temporarily refusing new connections, prompting clients to connect to another node in the cluster, and allowing existing sessions to terminate within a five-minute window.
Before proceeding, it is essential to verify that no active client connections remain. This ensures that updates or changes, such as firmware upgrades or license installations, can be performed safely without affecting live operations. The front panel LEDs provide visual confirmation of ongoing client activity.
To prepare a device for maintenance:
- Fade out client connections and pause the client-facing APIs to stop new incoming connections.
Once the maintenance task is complete:
- Resume the Client APIs to allow the device to rejoin the cluster and accept new connections.
After a reboot, the “APIs Pause” setting is reset and the device will again accept client connections automatically, unless synchronization is blocked due to a large number of pending HA objects (see High Availability Cloning).