Troubleshooting¶
Most issues come down to mounts, secrets, or indexing failures. Start with logs and the status page.
Check logs¶
For startup problems, look for database migration errors, missing environment variables, or Meilisearch connection warnings.
Container will not start¶
Check that .env exists and has at least:
Then check the compose service:
If port 8000 is already in use, change the left side of the port mapping:
Then open http://localhost:8080.
Source path does not exist¶
In Docker, source paths are container paths.
If compose has:
then the source path is /data/documents.
After changing mounts, restart:
No search results¶
Check these in order:
- Is the source added?
- Did you run Reindex?
- Does Admin → Status show successful files?
- Are include patterns too narrow?
- Are you filtering by the wrong source or type?
- Are files failing because of size limits or extractor errors?
From the CLI:
Indexing is slow¶
First indexing runs can be slow for PDFs, Office files, RAW photos, media files, and network mounts.
Things that help:
- exclude dependency/build folders
- use includes for sources that only need a few file types
- keep app/index volumes on SSD storage
- lower file size limits for huge files
- disable RAW/media metadata probing if you do not need it
See Performance Tuning.
Failed files stay in the status page¶
If the files were deleted or moved after scanning, use Clear stale on the status page for that source.
For deeper drift, run a full reindex:
Login stopped working¶
Tokens expire. Run:
If every token becomes invalid after restart, check that SESSION_SECRET is set and stable in .env.
Health endpoint is degraded¶
A degraded response usually means Meilisearch is not reachable. In the default managed setup, check the OneSearch container logs. In legacy external mode, also check the Meilisearch container.