Question 1

How do I fix Docspell v0.42.0 full-text content search returning no results with Solr 9?

Accepted Answer

The usual cause is that Solr is running without the required `analysis-extras` module. Since Docspell v0.41.0, Solr language field types such as `text_sk` and `text_kh` need this module. Without it, schema initialization can silently fail, leaving fields like `content_sk` missing; Solr then returns HTTP 400 for content searches while the UI shows no results. 1. Enable the module when starting Solr: ```bash # Docker -e SOLR_MODULES=analysis-extras # Non-Docker Solr solr start -Dsolr.modules=analysis-extras ``` For Docker Compose, add it to the Solr service: ```yaml services: solr: environment: SOLR_MODULES: analysis-extras ``` 2. Restart Solr so the module is loaded. 3. Verify the missing field now exists, replacing `` with your Docspell Solr core name: ```bash curl 'http://localhost:8983/solr//schema/field/content_sk' ``` 4. If the field is still missing on an already-broken installation, add it manually using the Schema API: ```bash curl -X POST 'http://localhost:8983/solr//schema' \ -H 'Content-Type: application/json' \ -d '{ "add-field": { "name": "content_sk", "type": "text_general", "indexed": true, "stored": true, "multiValued": false } }' ``` Add `content_kh` the same way if it is also missing. 5. Cancel any stuck `reIndexAll` job before retrying. A failed reindex job can repeatedly clear the Solr index with `DELETE *:*`. After the module is enabled and the schema is correct, run reindexing again. The permanent upstream fix is for Docspell's `SolrSetup` to check whether field types exist before creating them and to include `content_sk` and `content_kh` in schema initialization.

Question 2

Why does unoconv fail with 'Failed to connect to soffice.bin in 6 seconds' when converting XLSX to PDF?

Accepted Answer

This error occurs because unoconv times out trying to connect to a LibreOffice listener, often due to stale or accumulated soffice.bin processes. Workaround: kill all existing soffice processes and start unoconv in permanent listener mode. Run: pkill -f soffice.bin && unoconv --listener &. Then retry your conversion. Ensure the listener is started before processing. For persistent environments (like Docker), consider moving to unoserver as a more reliable replacement for unoconv.

Question 3

How to pass custom metadata when uploading documents to Mayan EDMS?

Accepted Answer

The upload API currently does not accept arbitrary custom metadata. As a workaround, encode the metadata in the document filename (e.g., using an absolute path with embedded info) during upload. In your addon or job-done hook, extract and parse the filename to retrieve the original metadata. Note that filenames may not be unique and could be exposed in the DMS. A feature request to support a dedicated metadata field exists in issue #2334.

docspell

Core Features

What It Can't Do

Use Cases

Tags

Getting Started

Similar Projects