Backup Fails to Download Chunk, but Check Passes

Support
#1

Good Morning,
I had one of my backups randomly fail last week and I’ve been trying to figure it out. This is the log…

Running backup command from /cache/localhost/0 to back up /backuproot/Canon M50 Injest
Options: [-log backup -storage Remote_Server -threads 1 -stats]
2025-11-07 02:00:01.342 INFO REPOSITORY_SET Repository set to /backuproot/Canon M50 Injest
2025-11-07 02:00:01.343 INFO STORAGE_SET Storage set to /backupremote/Duplicacy
2025-11-07 02:00:02.556 INFO BACKUP_START Last backup at revision 476 found
2025-11-07 02:00:02.598 INFO BACKUP_INDEXING Indexing /backuproot/Canon M50 Injest
2025-11-07 02:00:02.598 INFO SNAPSHOT_FILTER Parsing filter file /cache/localhost/0/.duplicacy/filters
2025-11-07 02:00:02.598 INFO SNAPSHOT_FILTER Loaded 0 include/exclude pattern(s)
2025-11-07 02:00:06.616 ERROR DOWNLOAD_CHUNK Failed to download the chunk 03a95bf3a5e81f1c49b4ced18dac86ec66cfb51ba06c5491d510176ef830b4e3: read /backupremote/Duplicacy/chunks/03/a95bf3a5e81f1c49b4ced18dac86ec66cfb51ba06c5491d510176ef830b4e3: resource temporarily unavailable
Failed to download the chunk 03a95bf3a5e81f1c49b4ced18dac86ec66cfb51ba06c5491d510176ef830b4e3: read /backupremote/Duplicacy/chunks/03/a95bf3a5e81f1c49b4ced18dac86ec66cfb51ba06c5491d510176ef830b4e3: resource temporarily unavailable

The chuck does exist on the storage when checked manually, and I have 5 other backups that still work to the same storage. A check on the storage passes fine. This backup had been working for more than a year before it failed.

I’m not sure how to proceed. Most of the topics I’ve found about missing chunks have failed the check, so I’m not sure if the solutions listed would work here, and I don’t want to screw anything up.

Any help would be appreciated.

#2

What is /backupremote mounted to? There some isisss with your mount, connection, server, or target media.

What OS, and how duplicacy is setup?

#3

Both source and target are unraid servers. They are connected via tail scale, and the target share is mounted via unassigned devices on the source server. Duplicacy is running in docker.

I thought of the mount, but the other backups that still work correctly use the same tunnel, share and folder. Is it possible that there is a corrupted chunk? Would a check return as clear with a corrupted chunk? Or maybe it’s a file permission issue with that specific file?

#4

Yes, there could be corrupted chunk. Check only checks for the presence of the chunks, not for their integrity. To test chunks integrity you need to delete verified chunks file and run check with -chunks argument.

However, since the error you saw is “resource unavailable”, it’s more likely that your SMB server is getting overwhelmed; I’d assume your unraid data pool is running zfs and is periodically scrubbed, so data corruption is impossible. Speaking of ZFS — heavy load, especially during scrub, can also cause such issues on improperly configured devices.

I strongly suggest not backing up to a mounted share when direct connection is available (such as SMB:// or SFTP://); this will help exclude the whole host of issues and risks, especially if you have a container engine in between, especially if that container engine is docker.

Edit. Tailscale relay latency (if direct connection is not possible) can contribute to SMB stability as well, even if you use a direct endpoint, let alone mount. Don’t use SMB over high latency links. Use SFTP.