Planning for tape drives as cluster resources

Direct fibre channel tape unit attachments are supported. Management of shared tape drives is simplified by the following PowerHA® SystemMirror® functions:

• Configuration of tape drives using SMIT
• Verification of proper configuration of tape drives
• Automatic management of tape drives during resource group start and stop operations
• Reallocation of tape drives on node failure and node recovery
• Controlled reallocation of tape drives on cluster shutdown
• Controlled reallocation of tape drives during dynamic reconfiguration

Limitations

When you plan to include tape drives as cluster resources remember the following:

• A tape loader or stacker is treated like a simple tape drive by PowerHA SystemMirror.
• No more than two cluster nodes can share the tape resource.
• Tape resources cannot be part of concurrent resource groups.
• The tape drive must have the same name (for example, /dev/rmt0) on both nodes that shares the tape device.
• When a tape special file is closed, the default action is to release the tape drive. PowerHA SystemMirror is not responsible for the state of the tape drive after an application has opened the tape.
• No means of synchronizing tape operations and application controllers is provided. If you decide that a tape reserve and release operation should be done asynchronously, provide a way to notify the application controller to wait until the reserve and release operation is complete.