infiniflow
[Feature Request]: Can the ES index be created based on the knowledge base instead of the user ID, so that it is more convenient to maintain?
Is there an existing issue for the same feature request?
- [x] I have checked the existing issues.
Is your feature request related to a problem?
Describe the feature you'd like
es 索引是否可以根据知识库创建，而不是根据用户id创建，这样更方便维护
Describe implementation you've considered
No response
Documentation, adoption, use case
Additional information
No response
Using Ragflow can encounter various problems, sometimes requiring the recovery of ES indexes. However, indexes are created by users, and restoring a single index will restore all the knowledge bases under the user. In fact, I just want to recover a certain knowledge base
Hi @KevinHuSh , I've seen the create index func has consider kb id, see code es_conn.py#L95 , however, it seems like we didn't actually use knowledgebaseId , do we consider to support this, for example configuring this by the end user, like doc_index_naming: ragflow_{uid}_{kbid} so that user could choose how to create the index.
Checked dify implementation for elasticsearch vector search integration, it would directly use collection id (kb id in ragflow) to create the index.
Looks like the infinity implementation used the kb id as part of the table naming, see code , @KevinHuSh do you have any suggested way to do this index customization ? I'm afraid this would be a break-change for the document index name, however, all kbs stored in a same index would also bring issues like @baiyi-os mentioned.