NDMP restore fails due to conflict in readWriteBlockSize

[harry_85]

Environment: Veritas NetBackup 7.6.0.2 running on windows 2008. NDMP backup/restore is done for EMC VNX.

Issue: While trying to restore the data from the tape, we receive the following error. There were many restores successful in the previous instance. This is the first time we face this issue.

Error ndmpagent(pid=2644) EDC-FS2: Specified record size (262144) is bigger than readWriteBlockSize (65536), please refer to Celerra document on how to increase param PAX.readWriteBlockSize

On the media server, when the backup happened the \netbackup\db\config\SIZE_DATA_BUFFERS_NDMP value was 256KB and 64KB in VNX param PAX.readWriteBlockSize respectively.

When the restore failed, we changed the \netbackup\db\config\SIZE_DATA_BUFFERS_NDMP value to 64KB to match the VNX param. However it still fails.

One option is to change the value on the VNX to 256 KB. However the DataMover will require a reboot and there will be a 10 minutes downtime which is not feasible solution as of now.

Question: Is there any way where we can have this figured out within NetBackup level as the restore is on hold because of this error.

[RamNagalla]

did you try creating the file SIZE_DATA_BUFFERS_RESTORE with 64KB.

if not yet.. i would suggest you to do that and see how it goes..

[harry_85]

We applied that in the media server. However it didnt work. The restore fails with the same error.

[Michael_G_Ander]

Afraid you can not change the block size of a backup image, it will use the block size at the time of the backup. The reason the backup work is probably that the Celerra delivered 4 64KB blocks every time Netbackup requested a 256 KB block.

Could you describe the previous instance where therestores worked ?

A workaround could be to duplicate the backup with a block size of 64 KB and the restore from duplicate tape