Backing up with Veeam Agent for Windows causes USB Controller failure

Not what everyone wants to hear, but you are better off creating a support ticket for this issue since you have already done a lot of troubleshooting and there is not much more I can think of.  They can check the logs and get to the answer or problem easier.

Thanks Chris, I thought I would post here first in case I was missing something obvious. And I also assumed that as a non-paying user I wouldn’t be able to contact support, nice to hear that’s not the case!

Yes you can it is just best effort support.

If your external drive is failing due to merging the incremental into the full, I’m not sure I would trust that drive as a backup device and would consider replacing it, since it might not be able to handle the read activity a full restore would require.

If replacing it is not an option, consider starting a new Active Full backup. This will make it so you don’t need to merge the incremental into the previous full, and after 8 days the previous chain will be deleted. Or you can manually delete the previous chain if you don’t need the restore point history.

Another thing if you can afford to loose what is on the drive format it again as ReFS 64K instead of NTFS.  It would perform better for backups.

9/07/2026 14:50:57 :: Full backup file merge failed Error: Agent: Failed to process method {Transform.Patch}: A device which does not exist was specified.
09/07/2026 17:10:05 :: Error: Agent: Failed to process method {Transform.Patch}: A device which does not exist was specified. Asynchronous request operation has failed. [requestsize

he HDD disconnects because all of the USB controllers seemingly are disabled and restarted. I can tell it’s all of them because my mouse, keyboard, webcam, and USB hub are all unresponsive for around 30 seconds, and then everything returns to normal (note that the external drive is connected directly to the laptop, not via the USB hub).

Very likely, USB hub on the laptop cannot handle the IO load of the merge. 

https://www.veeam.com/kb1932#ffi

Check the animation there to get an idea, and then imagine that this is happening for potentially millions of blocks. I am very doubtful it’s about the drive itself and more likely than naught the load on USB simply craps out.

The external drive or the SSD have had absolutely no issues in the past, and everything was working perfectly with the Veeam Agent up until the merging step was added to the backing up process.

Makes sense, likely merge had not happened yet. 

Few potential options here:

1. Reformat drive and use ReFS if it’s feasible, ReFS supports fast clone which is a light-weight metadata operation instead of physically writing data. You will want to monitor for ReFS events however as with a single external drive, you have no mirroring / repair, so if a required block is lost, you lose all backups depending on those blocks
2. Switch to Forward Incremental (periodic Active or Synthetic fulls). With active fulls, you avoid all synthetic IO operations; if your drive supports ReFS, you can also get fast clone benefits and “space-less” full backups.

I very strongly doubt it’s about faulty / dirty hardware here, very likely the IO just cannot handle it.

If you want to really confirm it, test with Diskspd using the Synthetic Full flags (set the -d value to be ~ same time as the merge occurs until failure -- i.e., if merge starts at 15:30 and failure occurs at 16:00, set -d to at least 1800, probably 1860 just to be sure). Likely you will get the same issue with USB disconnecting and it will more or less confirm the that it’s about the USB not being able to handle the IO load

Hi all,

Thanks a lot for your detailed answers and recommendations. I can’t see ReFS mentioned in my external drive’s documentation and I do have a lot of other data on it besides the Veeam backup, so I will switch the backup type to avoid the merging issue.

If replacing it is not an option, consider starting a new Active Full backup. This will make it so you don’t need to merge the incremental into the previous full, and after 8 days the previous chain will be deleted. Or you can manually delete the previous chain if you don’t need the restore point history.

Switch to Forward Incremental (periodic Active or Synthetic fulls). With active fulls, you avoid all synthetic IO operations; if your drive supports ReFS, you can also get fast clone benefits and “space-less” full backups.

So if I understand correctly, currently the backup is running as Forever Forward Incremental, and since I have more than 8 .vib files, Veeam is attempting to merge the older incrememntal backups to the full backup. So If I start a new active full backup and do that again every 6 months for example , deleting the .vbk and and .vib files of the previous chain, I will only be doing a full backup twice a year and would avoid the merging completely with an 8-day retention policy (since I backup monthly). Does that sound right?

Thanks again for all your help.