[Question]: After adding meta_fields,the chat with assistant fails to retrieve any text blocks,even when metadata_condition is added,how can this problem be solved?

[wangjuu]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (Language Policy).
• Non-english title submitions will be closed directly ( 非英文标题的提交将会被直接关闭 ) (Language Policy).
• Please do not modify this template :) and fill in all the required fields.

Describe your problem

After i pulled the nightly version of Docker image(2026.1.27),i added meta_fields to documents in the knowledge base.When using the chat with assistant interface for question and answer,i added the metadata_condition parameter.Even though the documents contained the answer to the questions,the assistant did not retrieve any relevant knowledge for the responses.
The parameters are as follows:
meta_fields = {
"doc_type": "common document"
}

"metadata_condition": {
"logic": "and",
"conditions": [
{
"name": "doc_type",
"comparison_operator": "is",
"value": "common document"
}
]
}

[dosubot]

This is a known issue with recent versions: metadata filtering will fail if meta_fields are missing or mismatched, or if the metadata_condition field name doesn't exactly match the key in meta_fields. Batch uploads do not set meta_fields by default, so you need to update metadata for each document after upload. After updating, verify that every document has the correct meta_fields and that the field name in metadata_condition matches exactly (e.g., "doc_type") source.

If retrieval still fails, test without metadata_condition to confirm if the issue is with filtering or document indexing. Check backend logs for warnings during metadata update or retrieval—silent failures are common if metadata isn't indexed properly source.

If the problem persists, re-create the knowledge base and re-upload documents to fix possible schema or index mismatches after upgrading source. For reliable metadata filtering, upgrade to a version newer than 2026-01-27, as several fixes have been merged since then source.

As a workaround, you can use tags for filtering if strict metadata filtering is unreliable in your current version.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[Magicbook1108]

what's your ragflow version? check by git log