Incremental backups running as full when TIR is enabled

Problem

During backups where TIR option is set, the NetBackup backup will fail to create/set the TIR info file for the drive.  A  *.new is left in the directory.

Error Message

'WRN - old TIR info file {INFO_FILE_PATH.EN_US} ' is missing. Backing up everything in {DRIVE_LETTER.EN_US}.

dtcp_read: TCP - failure: recv socket (360) (TCP 10054: Connection reset by peer)

Cause

Cause #1:  There were BEDS errors in the logs during backups.  These BEDS errors were setting a fail flag that caused NetBackup  to not promote the temporary file *.new file to the standard *.info file. The BEDS errors were in turn cause by insufficient space allowed on the disk drive for Shadow Copies.  This is a OS  level per disk setting, outside of NBU.

Cause #2: For situations where the "TCP - failure" message is noted in the bpbkar debug log, there was a communication error between the client and the media server after all backup activity completed. This is typically caused by the network environment between the the two servers and not associated with any NBU failure. The closed connection will cause an error flag setting that causes 'bpbkar' to not rename or remove the newly created TIR info file. 

Solution

For Cause #1: Using normal Windows actions, choose the disk drive and right click for "Properties".  Select the "Shadow Copies"  tab and then "Settings".  Increase the Maximum size setting or set it to "No limit".

For Cause #2:  Use the "Resilient Network" option on the Master Server and include the failing client name in the list.  It is recommended that the network environment itself be looked at to see if there are problems in the environment that are potentially causing network transmission errors.