Tape Pool Starvation and Clone Access
A 4-drive LTO-9 library lib-ams is attached to storage node sn-ams01. Pool ProdTape uses 2 drives for weekly fulls; pool CloneTape uses 2 drives for clones. Daily incrementals write 120 file clients to AFTD aftd-ams (40 TB) and 20 database clients to Data Domain dd-ams. At 01:00 the backup action reports 'media waiting: backup to pool ProdTape waiting for 1 writable volume' although 30 volumes are labeled in ProdTape. Clone jobs fail with 'Cannot read source volume; volume is not mounted on a device accessible to the clone storage node.' One drive remains 'device busy' for 6 hours. mminfo shows many ProdTape volumes with status full and retention dates in the future; two volumes show status read-only.
1. Diagnose the media waiting message. What volume and device conditions would you check, and what immediate and long-term fixes are appropriate?
Read the worked answer
'Media waiting' means the pool has no volume that is both writable and on an available device. The 30 labeled volumes may all be Full or Read-only, or their retention has not expired, so they are not recyclable. Check mminfo for pool, status, written, capacity, and retention/expiration, and use nsrmm -l to inspect volume status and mounts. Check device enabled state and whether a drive is busy with a clone. Immediate fixes: add or expand volumes in ProdTape, mount a recyclable volume, clear an inappropriate read-only status if policy permits, or move clone load off the shared drives. Long-term: size the pool from retention and daily change rate, separate backup and clone drive allocation, and avoid read-only media.
2. Why does the clone fail, and how would you correct clone storage node access? Include AFTD read/write host and tape library control.
Read the worked answer
A clone reads the source volume and writes the target volume, so the clone storage node must be able to access both devices. If the clone action selects a storage node that does not control lib-ams, the source tape may be mounted on sn-ams01's drives but is not accessible to the clone node. Set the clone action's storage node to sn-ams01, or move the clone target to a library that the clone node controls. For AFTD, configure read host and write host correctly when the clone node differs; for tape, the library is normally owned by one storage node, so keep cloning there or export/import volumes. Verify each pool device's storage node assignment.
3. Recommend target sessions and max sessions for the AFTD and tape devices, and explain the effect on the 120-client window.
Read the worked answer
For AFTD, set target sessions to 4-8 and max sessions to 8-16, depending on disk and network capacity; this allows multiple clients to write concurrently while capping load. For tape, set target sessions to 1 and max sessions to 1 per drive to avoid interleaving and improve restore performance. With only 2 ProdTape drives at target 1, only 2 full backups can run at once, so 120 weekly fulls can easily exceed the window. Options are adding tape drives, staging fulls to AFTD and cloning later, or extending the window. For clones, 2 CloneTape drives give only 2 clone streams, so 120 clone jobs need more drives or a separate clone window.