Troubleshooting

Troubleshooting

Failure Center

The Activity > Failures page collects items that need attention and provides Retry and Details actions.

Logs

Main Docker/Synology log:

/config/censorarr.log

Native Windows launcher log:

C:\ProgramData\Censorarr\logs\windows-launcher.log

Docker logs:

docker logs --tail 200 censorarr

Follow live:

docker logs -f censorarr

Permission denied on a movie

Symptoms:

ffprobe ... Permission denied

or:

SKIPPED media file because Censorarr does not have read permission

Check:

echo "PUID=$PUID PGID=$PGID"
gosu "$PUID:$PGID" id

Then test a real file:

gosu "$PUID:$PGID" head -c 1 "/media/Movie/Movie.mkv" >/dev/null && echo READ_OK || echo READ_FAILED

On Synology, also test as root:

head -c 1 "/media/Movie/Movie.mkv" >/dev/null && echo ROOT_READ_OK || echo ROOT_READ_FAILED

If PUID/GID fails but root succeeds, use:

CENSORARR_SYNOLOGY_COMPAT_MODE: "auto"

Censorarr 1.6.5 checks nested ACLs and can fall back automatically.

Windows folder or network share is unavailable

If a folder works in File Explorer but not in Censorarr:

1. Confirm Censorarr is running under the same signed-in Windows account.
2. Try a UNC path such as \\SERVER\Share\Movies if a mapped drive is not visible.
3. Confirm the account has read permission, plus write permission when Apply mode will modify media.
4. Restart Censorarr after changing credentials or drive mappings.

See Windows Installation.

GPU Worker says Invalid worker token

Current protocol uses:

X-Censorarr-Token

Check that:

1. main Censorarr and worker use the same literal token
2. the worker is current and not an old pre-rename build expecting the legacy protocol
3. no GUI-saved token is overriding the environment value

Test the worker directly:

curl -H 'X-Censorarr-Token: YOUR_TOKEN' http://GPU-IP:9000/health

Token looks correct but still fails

Censorarr secret precedence is:

GUI-saved secret
then environment variable
then legacy config value

So a stale GUI token can override ASR_WORKER_TOKEN.

See Security & Secrets.

GPU Worker cannot be reached

Check:

docker ps
docker logs --tail 100 censorarr-gpu-worker

Verify:

TCP 9000
correct worker IP
9000:9000 port mapping
firewall/routing

Worker says no CUDA device

This is normally a Docker/NVIDIA passthrough issue.

Verify the host sees the GPU and that NVIDIA Container Toolkit is configured before troubleshooting Censorarr.

Container keeps restarting

Check Docker logs first.

Censorarr 1.6.5 handles per-file ffprobe/read failures without terminating the whole daemon, so a restart loop usually points to a startup/configuration problem rather than one bad movie.

/config or /work bind mount fails on Synology

Make sure the project contains:

config/
work/

The public repository preserves them with .gitkeep.

Use relative mounts:

- ./config:/config
- ./work:/work

rather than hard-coded /volume1/docker/censorarr/... project-internal paths.

Censorarr sees no media

Inside the container:

ls -la /media
ls -la /tv

If empty, verify the left side of the compose volume mapping points to the correct host/NAS folder.

Radarr/Sonarr connects but media does not match

Check path mappings.

Example:

Radarr reports: /movies/...
Censorarr sees: /media/...

Mapping:

- from: /movies
  to: /media

Bazarr connects but subtitles do not associate

Check both movie and TV path mappings.

Browser cannot open Censorarr

Docker/Synology: check compose:

ports:
  - "8087:8787"

Then use:

http://SERVER-IP:8087

Native Windows normally uses:

http://127.0.0.1:8087

Self-test

docker exec -it censorarr python /app/selftest.py

This is useful after upgrades and before reporting a reproducible application bug.