AFM synchronous read mount

Creates a separate mount path at the gateway node through which synchronous read calls are sent to the home cluster separately.

When requested data is not available in the cache, AFM retrieves the data from the home cluster, stores the data in the cache, and then completes the read operation.

Sometimes, when data retrieval from home takes time, for example; when the file is migrated to the tape at the home and fetching this file at cache is slow. In cases of late response, EIO error is returned back to the application. This EIO error might cause the queue to drop at the gateway node and AFM put the fileset into stale state.

Enabling afmSyncReadMount creates a separate mount path at the gateway node through which synchronous read calls are sent to the home separately. This arrangement helps other operations that are running on the AFM fileset to synchronize with the home by using the default path. If due to any reason home is slow or busy, the read operation gets killed, and retried without affecting the other operations as this read operation is performed on the separate mount path.

Enabling afmSyncReadMount parameter might avoid dropping the whole queue at the gateway node. As a result, AFM avoids sending EIO error back to the application that performs the read operation and moves AFM (SW/IW) fileset into stale state and avoid recovery in InfoSphere® MDM Server Workbench or IW fileset.

The afmSyncReadMount parameter can be enabled on the AFM Single-Writer (InfoSphere MDM Server Workbench), Independent-Writer (IW), Local-Update (LU), and Read-Only (RO) mode fileset by using mmcrfileset or mmchfileset command.

The valid values for the afmSyncReadMount are yes or no.

To set or unset the afmSyncReadMount parameter, AFM fileset must be stopped or unlinked. The afmSyncReadMount parameter can be enabled either at the time of fileset creation by using mmcrfileset or modify existing fileset by using mmchfilset command.

For more information, see Stop and start replication on a fileset

Example

  1. To create an AFM fileset and enable afmSyncReadMount on the fileset, issue the following command: 
    mmcrfileset fs1 afmRO -p afmMode=ro,afmTarget=home:/gpfs/fs1/target1,afmSyncReadMount=yes --inode-space new
  2. To update an existing AFM fileset and enable afmSyncReadMount on the fileset, AFM fileset must be stopped or unlinked. To stop the fileset, issue the following command:
    mmafmctl fs1 stop -j afmRO
  3. To set the afmSyncReadMount parameter, issue the following command:
    mmchfileset fs1 afmRO -p afmSyncReadMount=yes
    Note: Before you enable afmSyncReadMount parameter, ensure to upgrade the latest IBM Storage Scale cluster version and file system version. For more information, see File system format changes between versions of IBM Spectrum® Scale.
  4. After the parameter is set, the fileset can be started or unlinked. To start the fileset, issue the following command:
    mmafmctl fs1 start -j afmRO