AFM recovery improvements for remove and rename operations
The AFM Recovery improvements for remove and rename operations are supported on AFM Single Writer and AFM DR primary mode where Home or Secondary is running IBM Storage Scale file system.
During a failure event, recovery is started by AFM to sync dirty data with home or secondary site. When recovery is triggered on an AFM fileset that contains large data, AFM performs various steps to synchronize the data between Cache and Home. This activity also includes identifying the remove and rename operations.
In case where the AFM fileset has large dirty data, this process might take some time to identify the rename and delete operations.
Finding rename and remove operations on AFM fileset data during recovery can be optimized by setting afmRecoveryVer2 on the AFM Single Writer fileset or AFM DR primary fileset.
By enabling afmRecoveryVer2, AFM optimizes the process of finding IBM Storage Scale rename and remove operations by running policy at the home or secondary. Policy generates the list that is referred to identify the pending rename and remove operations at cache and AFM queues those operations to the Gateway Node.
Running policy scan at the home or secondary helps AFM to quickly identify the rename and recovery operations. This reduces time when directory has large files and various rename and remove operations that are pending for synchronization to the home.
After the afmRecoveryVer2 is set on the AFM fileset, AFM tries to run recovery optimization. If optimized recovery cannot be performed, AFM failback to the older recovery version and completes the recovery.
Before you enable afmRecoveryVer2 parameter on the AFM fileset, cluster and file system must be upgraded to the latest release level.
Example
mmafmctl fs1 stop -j SW1mmchfileset fs1 SW1 -p afmRecoveryVer2=yes#mmafmctl fs1 start -j SW1