Immich data folders wiped / missing assets
❓Question☑️Ready
I’m seeing a strange issue where my Immich data folders appear to have been wiped.
I didn’t notice anything wrong in the app at first because thumbnails were still loading, but when I tried to download an image it failed. Checking the logs showed missing asset paths, and when I inspected my library directory, the data was gone. Based on timestamps, this seems to have happened around 1/17 at midnight, and Immich continued running normally afterward.
I’m running Immich v2.4.1 via Docker with a custom storage layout:
- library, backups, and profile are on NFS (TrueNAS)
- thumbs, uploads, and encoded-video are local on NVMe
- I use a storage template, so uploads land locally first and are then offloaded to the library
Even the .immich files were missing on the NFS-backed folders. However, new media uploaded after 1/17 has been importing fine, which makes it feel like Immich lost access to the existing data and just kept going.
Possibilities I’m considering:
- A cron job or scheduled task around midnight
- An NFS disconnect that caused Immich to behave badly when the mount returned
I did restructure my data layout a couple weeks ago (moving videos and backups to TrueNAS as well), but everything appeared to be working correctly afterward.
Fortunately, I have frequent TrueNAS snapshots, so I’m currently restoring the library data from snapshots and merging in post-incident uploads so nothing is lost. Once the data copy finishes, I plan to run the library check script (the one shared by skatsubo in a recent thread) to reconcile DB vs disk.
Logs: https://gist.github.com/palanjam/5e021716b0fddda9523c0a8497a7e82c
Docker Compose: https://gist.github.com/palanjam/bb78eae177a9e310354581c9d19fdee3
I didn’t notice anything wrong in the app at first because thumbnails were still loading, but when I tried to download an image it failed. Checking the logs showed missing asset paths, and when I inspected my library directory, the data was gone. Based on timestamps, this seems to have happened around 1/17 at midnight, and Immich continued running normally afterward.
I’m running Immich v2.4.1 via Docker with a custom storage layout:
- library, backups, and profile are on NFS (TrueNAS)
- thumbs, uploads, and encoded-video are local on NVMe
- I use a storage template, so uploads land locally first and are then offloaded to the library
Even the .immich files were missing on the NFS-backed folders. However, new media uploaded after 1/17 has been importing fine, which makes it feel like Immich lost access to the existing data and just kept going.
Possibilities I’m considering:
- A cron job or scheduled task around midnight
- An NFS disconnect that caused Immich to behave badly when the mount returned
I did restructure my data layout a couple weeks ago (moving videos and backups to TrueNAS as well), but everything appeared to be working correctly afterward.
Fortunately, I have frequent TrueNAS snapshots, so I’m currently restoring the library data from snapshots and merging in post-incident uploads so nothing is lost. Once the data copy finishes, I plan to run the library check script (the one shared by skatsubo in a recent thread) to reconcile DB vs disk.
Logs: https://gist.github.com/palanjam/5e021716b0fddda9523c0a8497a7e82c
Docker Compose: https://gist.github.com/palanjam/bb78eae177a9e310354581c9d19fdee3
