Processes high availability event.
ha_star [ -C ]
The ha_star command is the generic high availability handling command. It is automatically invoked by the operating system through /etc/rc.ha_star when a CPU predictive failure is reported by the firmware.
If ha_star is invoked without flags, only new events are handled. If ha_star does not find any new event, it exits.
When running, ha_star handles all new events, even those which arrive while ha_star is handling already existing events. Only one instance of ha_star can be running at any given time. Should a second instance of ha_star be launched, it exits.
The operating system invokes ha_star when a high availability event is reported. The event handling may fail or it may be cancelled (for example, by signals). Aborted or cancelled events are held in memory within the kernel. When the cause of the abort has been corrected, then the event handling can be retried. This is when ha_star is invoked manually by the system administrator.
The ha_star command generates error or failure error log entries.
Description by Event Type
The ha_star command is invoked by the operating system to deallocate a CPU when a predictive processor failure event is detected. This deallocation may fail because some threads remain bound to the CPU being deallocated. In some cases, system administrators can fix the condition which led to the failure of the deallocation. For example, they may be able to identify and stop applications with threads bound to the last logical CPU.
The -C flag indicates that the high availability event to be resumed is a CPU deallocation event.
| Item | Description |
|---|---|
| -C | Specifies that the event to be restarted is a CPU deallocation. |
| Item | Description |
|---|---|
| /usr/sbin/ha_star | Contains the ha_star command. |