Disaster recovery

Use these disaster recovery solutions for HyperSwap®, Metro Mirror, Global Mirror, and Stretched System, where access to storage is still possible after the failure of a site.

HyperSwap

Active-active volume access is always provided while there is an up-to-date consistent copy. If there is an out-of-date consistent copy, there is not an automatic failover to it, nor is read-only access given to it. Use the stoprcrelationship-access or stoprcconsistgrp-access command to make it accessible. The relationship is then in the Idling state. After you enable access with the stoprcrelationship-access or stoprcconsistgrp-access command, use the startrcrelationship -primary <master/aux> or startrcconsistgrp -primary <master/aux> command to make the relationship leave the Idling state and resume HyperSwap replication. If you previously ran overridequorum, the startrcrelationship or startrcconsistgrp command fails.

When you resume HyperSwap replication, consider whether you want to continue by using the out-of-date consistent copy or revert to the up-to-date copy. To identify whether the master or auxiliary volume has access, look at the primary field that is shown by the lsrcrelationship or lsrcconsistgrp command. To continue to use the out-of-date copy, provide that value as the argument to the -primary parameter of the startrcrelationship or startrcconsistgrp command. To revert to the up-to-date copy, specify the opposite value as the argument to the -primary parameter. For example, if master is shown in the primary field of lsrcconsistgrp for an active-active consistency group in the Idling state, to revert to the up-to-date copy, use startrcconsistgrp -primary aux.

Metro Mirror and Global Mirror

Warning: Inappropriate use of these procedures can allow host systems to make independent modifications to both the primary and secondary copies of data. You are responsible for ensuring that no host systems are continuing to use the primary copy of the data before you enable access to the secondary copy.

In a Metro Mirror or Global Mirror configuration a system is configured at each site. Relationships are configured between the systems to mirror data from storage at the primary site to storage at the secondary site. If an outage occurs at the secondary site the primary site continues operation without any intervention. If an outage occurs at the primary site, then it is necessary to enable access to storage at the secondary site.

Use the stoprcrelationship-access or stoprcconsistgrp-access command to enable access to the storage at the secondary site.

Stretched System

In a stretched system (formerly split-site) configuration, a system is configured with half the nodes at each site and a quorum device at a third location. If an outage occurs at either site, then the other nodes at the other site access the quorum device and continue operation without any intervention. If connectivity between the two sites is lost, then whichever nodes access the quorum device first continues operation. For disaster recovery purposes you might want to enable access to the storage at the site that lost the race to access the quorum device.

Use the satask overridequorum command to enable access to the storage at the secondary site. This feature is only available if the system was configured by assigning sites to nodes and storage controllers, and changing the system topology to stretched.

Important: If you run disaster recovery on one site and then power up the remaining, failed site (which contained the configuration node at the time of the disaster), the cluster asserts itself as designed. This procedure would start a second, identical cluster in parallel, which can cause data corruption.You must follow these steps:
  1. Remove the connectivity of the nodes from the site that is experiencing the outage.
  2. Power up or recover those nodes.
  3. Run satask leavecluster-force or svctask rmnode command for all the nodes in the cluster.
  4. Bring the nodes into candidate state.
  5. Connect them to the site on which the site disaster recovery feature was run.

Other configurations

To recover access to the storage in other configurations, use Recover system procedure.