We recently encountered a problem where the BizTalk SQL Agent job that executes the backup procedure resulted in a full backup every run in stead of once every day.
We became aware of the issue when disk space events were fired by our monitoring software.
We found the problem when we looked into the table adm_BackupHistory in de BizTalkMgmtDb. For some reason someone had felt the need to manually execute the backup job and somehow managed to enter a date in the future as the last-run entry.
In stead of february 3rd, march 2nd was entered (European dd-mm-yyyy vs US mm-dd-yyyy mix-up.) A classic by al means.
This caused the backup's job logic to determine if it was time for a full backup to fail and always return true.
Once this part of the history was corrected, the Backup Job went back to normal operation.