Multiplex w/ KMS
NetBackup 7.1.0.4 with the IBM TS3310 tape library.
It would seem I cannot perform Multiplex w/ Encryption (KMS). The backup runs without error (false positive?). The tape is unreadable afterwards.
Anyone else share this experience? How did you fix it?
If you look in bptm for a job of the failed restore you should see lines containing this :
establish_decryption_key: next block encryption
This line should appear after this line:
io_position_for_read: successfully positioned
The question: how many 'establish_decryption' lines are there directly after the postioning line, there should only be 1.
If there are more, you are probably looking at an issue similar to eTrack 3093882, and will need to log a call.
The bptm log, KMS log and nbsu -c -t outputs will be required, as well as a exact description of the issue.
The casue apeears to be firmware related, and an EEB was released to work around the issue.
Martin
It is unlikely that the retention level or timestamps had anything to do with not being able to decrypt your data. With KMS encryption, during a read process, the tape drive will encounter an encrypted block of data on the tape and notify the BPTM process via SCSI commands. BPTM will then communicate with NBKMS on the master server to retrieve the encryption key. Without actually seeing a restore failure from a full BPTM log we can't know if the tape drive is telling us if the data is encrypted or not.
One of the comments earlier mentioned that this is likely a firmware issue, and I would agree to that.
What model of tape drive are you using and what is the firmware version? It wouldn't be the first time that we have encountered defective firmware that would cause this issue. Check out the following technote:- There is a bug that affects certain makes of drive due to a firmware issue. The drive is unable to correctly position after reading the backup header for the restore that is requested. Although this is not a NetBackup fault, there is a EEB that provides a workaround by getting the drive to read trough the headers as opposed to an ' mt -f ' which is what is failing. I'll try and find the details for his issue tomorrow (Fri) Martin