The correct action is to run content router queue processing on the media server that hosts the MSDP storage server. In an MSDP environment, expiring backup images with bpexpdate removes or marks the NetBackup catalog references, but the deduplicated segment references in MSDP are reclaimed through the deduplication transaction queue. Cohesity NetBackup documentation explains that after images are expired, MSDP may not immediately reclaim the freed space and that administrators can accelerate reclamation by manually processing the MSDP transaction queue. The deduplication process writes transaction log data into the queue directory, and that queue data is later processed during the next content router queue processing operation. Official NetBackup Deduplication procedures specify running crcontrol --processqueue on the MSDP storage server, not on the primary server. Image cleanup alone is not the MSDP space-reclamation mechanism after manual expiration, and a storage-pool inventory does not dereference deduplicated fragments. Therefore, the required step is option A.

The correct logs are the NetBackup bpfis logs on the client and the operating system logs on the client. The job details show a Windows snapshot failure: GLOBALROOT\Device\HarddiskVolumeShadowCopy..., status: 156: snapshot error encountered, and status: 1542: An existing snapshot is no longer valid. In Cohesity NetBackup snapshot processing, bpbrm starts bpfis on the client, and bpfis creates the snapshot of the client data and returns snapshot completion status. Therefore, the bpfis client log is the primary NetBackup log for the snapshot creation, mount, and delete phases.

Because this is a Microsoft Windows client snapshot issue, the Windows operating system logs are equally important. Cohesity’s status 156 troubleshooting topic identifies VSS framework failures and notes that VSS-related application events may be written on the affected Windows system; it also references bpfis log content for snapshot failures. The media server bpbrm log can show orchestration status, but it usually reports what the client-side snapshot component returned. The bptm log is for media/data movement and is not the best source for VSS snapshot root cause.

The correct action is to increase the Maximum I/O Streams for the MSDP disk pool. In NetBackup disk storage, the disk pool stream limit is the effective gate for concurrent read/write activity against that pool. Cohesity documentation identifies Limit I/O streams as the disk-pool property that controls read and write streams, and notes that if the limit is reached, NetBackup queues jobs until a volume becomes available. The storage-unit Maximum concurrent jobs setting controls how many jobs NetBackup sends to a disk storage unit, but Cohesity also states that this setting is not the only concurrency limiter and depends on disk-pool and media-server factors. NetBackup performance guidance is more explicit for this scenario: the disk pool Max I/O Streams value sets concurrent read/write operations, while the storage unit concurrent-jobs value addresses backup/restore jobs and does not control secondary operations such as duplications and replications; those are usually dictated by Max I/O Streams. Therefore, increasing storage-unit jobs or SLP priority does not guarantee that AIR replications can run while backups consume all available disk-pool streams. The correct tuning point is the disk pool’s Maximum I/O Streams.