Dell FluidFS NAS Solutions Administrator's Guide

In case of a controller failure, the system automatically migrates each connection from the failed controller to another controller. This causes disconnections to CIFS clients, unless the Migrate Manually policy has been selected for CIFS. However, selection of this option requires you to manually migrate clients. Migration of CIFS clients in any circumstance, will interrupt I/O. Click the Windows Cancel button, and retry the transfer. When the failed controller restarts, the system rebalances the load by migrating clients back to the revived controller automatically. This operation is called fail-back.

Clients that use NFS are stateless and are not affected during fail-back. To optimize the fail-back operation, the system provides you with the following policies for migration on recovery:

• Migrate Immediately — Always keeps the system well balanced, at the cost of possibly disconnecting CIFS clients during work time.
• Migrate Automatically — Always keeps the system well balanced if the controller failure is very short, at the cost of disconnecting CIFS clients. This option causes the system to remain unbalanced for a period of several days, if the failure remains for a long time.

  This mode overcomes short controller failures because clients have not created new material during the short time failure. Therefore, the best practice is to rebalance them as soon as possible. If the failure is longer than 10 minutes, the system remains unbalanced until you rebalance it manually.

• Migrate Manually — Never migrates clients automatically. This requires manual intervention to rebalance the system. If the system requires manual intervention to rebalance it after fail-over, the system sends an appropriate email message to the administrator.

To set the migration policies: