Incremental job running as Full after failure
Our Incremental backup job that is scheduled to run every day at 10 p.m. failed on 3/31. I tried to restart the job, but it failed with an E00081D9 error code. As far as I could tell, the BE Job Engine service and Management service were both stopped. I started them, and the job seems to have started and then failed again. I right-clicked and chose "Run Now," and it started normally. But it has been running for the past 24 hours, and as far as I can tell it's doing a Full backup, even though the Job Activity monitor says it's running as an Incremental. Our daily Incrementals usually run only a few hundred MB and the jobs finish in a few minutes. Another discussion here suggests running a Full backup on an Exchange server after an Incremental failure, but we're not running Exchange. I guess my question is if this is a software feature and not a bug. Is it running this way because it had to re-set the archive bits after the first job failure?
Edit: Continuing to research this, I came across this page at Veritas: https://www.veritas.com/support/en_US/article.100020511 (Backup using "Differential/Incremental - using modified time" option backs up all the data rather than just the changed data when no remote agent is present)
The main suggestion is that a failed Full job will not add the time of the backup to the Backup Exec database.full backup job completes successfully, and that as result, Incremental (or Diff) jobs will bakc up more data than expeced. In my case it wasn't a full job that failed, however. But reading farther, I come across this information: "The error during the 'Full - Back Up Files - Using modified time' is due to no remote agent detected or present on the remote server during the Backup.
Solution
Differential and Incremental backup methods (by Modified Time or Archive Bit) are supported only if the Agent for Windows Systems (AWS) or Remote Agent for Linux/Unix Systems (RALUS) is installed. For systems unable to have either agent installed, only the backup method FULL is supported."
And, yes, sure enough, we have an outdated agent on the server that's being backed up. But why has it worked for the past six months or whatever, and only now displays this behavior after a failed Incremental job? Has this behavior been addressed in BE 15? (The page is from 2014.) Or are all of the Incrementals that we do kind of shaky and not ultimately reliable? We've been able to restore from Incremental backups when we need to.
Finally the even worse news is that the Incremental job (running as Full) that I started yesterday filled up the drive and asked me to make space before it would continue. I didn't want to delete anything in the BEData folder for fear of rendering unusable what backups we do have. I changed the storage location to 'Any disk storage," since there's space on another drive in the pool. But that was before I came across this info, and now I'm thinking that it would just run another pseudo-Incremental that's really a Full. Does that sound correct?
Once you have a failure, you should always regard the backup chain as suspect and start a new chain with a full backup. Do not attempt to continue with the failed chain. You might not be able to restore from that chain.
The article that you referred to is talking about the situation when the data source is a NAS or some appliance on which you cannot install the BE agent. In those cases, you can only do full backups.
When you upgrade the BE Server, you should always make sure that you update the BE agent on all other servers as soon as possible. Without the most up-to-date agent, you may not get the updated functionalities of the latest update or you might experience unexplanable problems.