When 4 backup or restore jobs are running concurrently from an EMC Celerra filer, the 5th job will fail with the following NetVault Backup errors:
'Mover has halted due to an internal error'
'[DEVICE]: had a channel error'
The channel error will have the following additional context:
DEVMOVER 658 0 72101 oPost is a 'NDMP_NOTIFY_MOVER_HALTED_POST'
DEVMOVER 659 0 72101 Reason = 3
On the Filer, the datamover will show the following errors:
NDMP: 6: Active NDMP backup/restore streams reach the maximum concurrent streams supported 4. No more NDMP sessions can be started. Wait current backup/restore sessions finish before start new ones.
NDMP: 3: < LOG type: 2, msg_id: 0, entry: Active NDMP backup/restore streams reach the maximum concurrent streams supported 4. No more NDMP sessions can be started. Wait current backup/restore sessions finish before start new ones.
Celerra filers have a maximum number of concurrent NDMP essions for each NDMP-Host Data Mover. The number of concurrent sessions is limited by the concurrentDataStreams server setting.
For all DART OS versions, the default value for concurrentDataStreams is 4.
For EMC Celerra DART OS 5.6 and earlier, the maximum value for concurrentDataStreams is 4, from 6.0.4, the maximum value was increased to 8.
Ensure that backup and restore jobs are configured so that there are never more than concurrentDataStreams jobs running concurrently. If the filer is running 6.0.4 or greater, consider increasing the value of concurrentDataStreams setting. Refer to the EMC document 'Configuring NDMP Backups on Celerra' for details about this setting and how to modify it.
