Restore and Verify of multiplexed (MPX) backups to tape fail after upgrading the Solaris SPARC master server to NetBackup 8.3 or 8.3.0.1

Problem

After a Solaris SPARC master server upgrade to NetBackup 8.3 or 8.3.0.1, multiplexed tape backups on any media server succeed, but image-verify and image-restore fail. Non-multiplexed images are not affected. Duplicate copies of these images are corrupted and not-recoverable.

Error Message

May  5,  2020  12:02:26  PM  -  begin  reading
May  5,  2020  12:02:27  PM  -  Error  bptm  (pid=54321)  backup  id  client_1603536789  is  not  a  tar  formatted  image
May  5,  2020  12:02:33  PM  -  Error  bptm  (pid=54321)  The  following  files/folders  were  not  restored:
May  5,  2020  12:02:34  PM  -  Error  bptm  (pid=54321)  UTF  -  /etc/zones/One.xml
May  5,  2020  12:02:34  PM  -  Error  bptm  (pid=54321)  UTF  -  /etc/zones/Two.xml
May  5,  2020  12:02:34  PM  -  Info  tar  (pid=1214)  done.  status:  5
May  5,  2020  12:02:34  PM  -  Info  tar  (pid=1214)  done.  status:  185  -  tar  did  not  find  all  the  files  to  be  restored
May  5,  2020  12:02:34  PM  -  Info  bpbrm  (pid=12439)  media  manager  for  backup  id  sandy_1601535601  exited  with  status  92: media  manager  detected image  that  was  not  in  tar  format

Cause

A defect in NetBackup on Solaris master servers leads to writing the wrong information into the NetBackup image database when a multiplexed backup to a tape storage unit completes.  The multiplexed backup image is written to the tape storage unit correctly, but the NetBackup image database entry is incorrect.  Because of this inconsistency, the image cannot be properly verified or restored.

If the aforementioned multiplexed image (first copy) is duplicated to any other storage unit (second copy), the duplication job  succeeds, but the subsequently created image (second copy) is no longer readable and is lost.

All of the copies of a twinned multiplexed backup image that are written to tape storage units are considered the first copy and are written correctly.  In this case secondary copies are the images that are at risk.

For example: A multiplexed backup is twinned to two storage units resulting in first copies 1 & 2, duplicated to secondary copy 3 -- copy 3 is unusable.

Solution

An Emergency Engineering Binary (EEB) has been created to prevent this problem from occurring with the newly created backup images only. Any existing backup images that were written after upgrade to 8.3/8.3.0.1, but before this EEB is applied, cannot be verified, duplicated, or restored with this fix. Media servers upgraded to 8.3/8.3.0.1 with this EEB applied will fail to duplicate affected images, stopping the data corruption.

*** Any multiplexed images written to tape between upgrade to 8.3/8.3.0.1 to time of EEB application should not be duplicated until the images can be fixed. Without image repair, duplication of the affected images will fail. ***

*** Failure to upgrade all media servers to 8.3/8.3.0.1 and apply EEB will result in further data corruption. ***

Copy 1 (first copy) multiplexed images written to tape can be corrected in the image DB by expiring and importing those images. Once imported, these images can be successfully duplicated.

*** If there are images that had been written to tape using multiplexing and those images were duplicated to another location BEFORE this EEB was applied, and a verify or restore of those duplicated images show the error, please contact NetBackup Technical Support for remediation steps.

If customers have Solaris SPARC master or media servers in the environment, obtain the relevant EEB and apply it based on the version installed.  All media servers in a domain with a Solaris SPARC master must be upgraded to 8.3 or 8.3.0.1 and apply this EEB.

The EEB fix can be obtained from the Download Center links below.

For NetBackup 8.3, Etrack 4019326, EEB Version 2

For NetBackup 8.3.0.1, Etrack 4017052, EEB Version 2