Your Browser is not longer supported

Please use Google Chrome, Mozilla Firefox or Microsoft Edge to view the page correctly
Loading...

{{viewport.spaceProperty.prod}}

Reconfiguration for multiprocessors

&pagelevel(4)&pagelevel

The system reacts automatically to hardware failures. If one of the components (CPU, channel) fails, it is automatically detached from the operating system. Operation continues using the remaining components.

The operator can attach or detach components by means of reconfiguration commands. If a component is to be disabled in order to be serviced or for other reasons, the operator can detach it by means of the DETACH-DEVICE command. Further operations are then performed on the remaining components. If a component which has been detached is to be made available again, the operator can attach it using the ATTACH-DEVICE command.

Reconfiguration fails if:

  • after a CPU has been detached, there is no longer at least one CPU still operable

  • if inconsistencies arise in the operating system data when a CPU fails and the hardware is unable to make a valid hardware context available

  • if public volumes or the last console are affected by the failure.

Certain devices must be switched on or switched over before they can be addressed by the system once more. Switching over in this case means not only the physical switching of the device to an alternative path, although this is a prerequisite for the operation Switching over incorporates all the actions necessary for the system to be able to address the devices again. The process of switching over or switching on devices is the responsibility of the operator.

If an error occurs in one of the attached CPUs for a Server Unit with more than one CPU, the Server Unit in many cases will be reconfigured automatically. The only time that a reconfiguration cannot be performed is if a valid program context cannot be created for the failed CPU and if the lock mechanism of the operating system is active.