Recovering from an unplanned CRG Global Mirror failover

This scenario demonstrates what happens in a HyperSwap environment with Global Mirror when an unplanned Global Mirror failover takes place and how to recover from the event.

Before you begin

This scenario uses the configuration created in the Configure HyperSwap with Global Mirror using Copy Services Manager (CSM). Review this configuration before proceeding with the switchover operation.

There are restrictions for the Global Mirror switchover which are the same for all Global Mirror connections:

• the Global Mirror link must be active at the time of the time of the unplanned failover.

About this task

In this scenario, NODE_1 has the primary copy of the data on DS8000 A that is configured for Global Mirror. It replicates data from the primary, DS8000 A to a DR site with DS8000 C and a separate IBM i system, NODE_2. In addition to a Global Mirror connection, DS8000 A has a HyperSwap relationship to DS8000 B. Both storage units are attached to the IBM i system, NODE_1. All disk units in the IASP on DS8000 A correspond to disk units in DS8000 B and are linked across a synchronous connection.

Here, a Global Mirror failover event takes place at the IBM i NODE_1 and PowerHA switches to the IBM i NODE_2. The primary for the Global Mirror moves to NODE_2, the DR system.

Procedure

The diagrams below depict the stages of a Global Mirror failover in an environment configured for HyperSwap replication and Global Mirror.

1. Begin with the environment in normal operation as described above.

2. When PowerHA detects a condition that will trigger a failover, it will attempt to switch from the primary, NODE_1 to the DR system, NODE_2. In this case, the NODE_1, failing and going offline, and the CRG recovery domain listing NODE_2 as the first active backup node in the recovery domain.

3. After the Global Mirror failover takes place, the primary system and storage unit will be the NODE_2 system with
   DS8000 C.

   Important: After the unplanned failover, PowerHA does not resume replication automatically. Global Mirror replication remains detached.

   In order to reattach the Global Mirror, the production system and storage must be brought online. At this time, the user must decide which version of the mirrored data to keep and which version to allow Global Mirror to overwrite when it resumes. When PowerHA is given the command to reattach, the Global Mirror link connects from the current primary system, in this case, the DR system, NODE_2 to the backup, now NODE_1. Data on DS8000 A is overwritten with the Global Mirror copy on DS8000 C.

4. Reattach the Global Mirror session using the Change CSM session (CHGCSMSSN) command. Type into the command line, CHGCSMSSN SSN(MYSSN) OPTION(*REATTACH) and press enter.

5. A confirmation screen will appear asking the user to confirm the direction of replication. If so, enter F16 (Shift + F4) to reattach the Global Mirror.

Results

The Global Mirror connection is reattached from the primary, NODE_2 to the backup, NODE_1.

Example

Figure 1. Stages of a recovery from a CRG failover in a HyperSwap with Global Mirror configuration

Open