Recovery

Any failure that disrupts ring processing requires recovery. Global resource serialization can both detect the failure and respond to it. Depending on options your installation selects, global resource serialization can automatically rebuild a disrupted ring. It can also automatically issue VARY commands to enable and disable communication links. These actions speed up recovery from a failure and reduce operator intervention in the recovery process. The primary causes of a failure that requires recovery are:

A system fails because of a software problem, which can be either related to ring processing or independent of ring processing.
A system fails because of a hardware problem.
A CTC link fails. The failure can occur in the link itself or in any hardware or software component required in the communication path.
Global resource serialization detects either a temporary problem on a system in the ring (such as when an operator stops a system) as a system failure or a temporary delay in communication as a communication link failure.

Whatever the cause of a break in ring processing, the result is a ring disruption; global resource serialization suspends the processing of requests for global resources until it recovers from the failure.

In designing your complex, plan a complex that can recover from a system or a CTC link failure, regardless of whether the failure is temporary or not.

After a system failure, there should be enough CTC links available to reconfigure a subset ring of n-1 systems, where n is the number of systems in the original ring. That is, you want a complex where a system failure means that only the failed system must withdraw from the ring.

After a CTC link failure, there should be enough CTC links available to reconfigure the original ring. That is, you want a complex where the failure of a single link does not force any system to withdraw from the ring.

To design a complex that can recover effectively from failures, consider both the level of connectivity and the need for alternate CTC links. Alternate links are CTC links that global resource serialization can use for ring acceleration, which improves global resource request response time. Alternate links are available for use when there is a failure on a link that global resource serialization is using to pass the RSA-message around the ring, which improves recovery.