How to correctly handle deleted or updated documents?

[patng323]

I am using LightRAG server. I uploaded some docs and it all works well. However, my documents will be updated regularly (and some docs will be removed). After I updated a file (e.g. doc123.txt), I see that a duplicate doc entry (also for doc123.txt) is in the Documents list (in the Web UI). How can I remove the old ones?

In lightrag.py I see that the method adelete_by_doc_id was deprecated. So what is the correct way to handle deleted or updated docs? I assume it's a common scenarios in real production environment.

[onestardao]

yep — what you're running into is a document shadowing issue.
LightRAG doesn't consistently clean up vector store + metadata when a doc is updated, especially if it reuses the same filename (like doc123.txt).

this isn't just an edge case — it's actually Problem No.12 in our failure map:

"update doesn't overwrite old vector / stale UI still lists both"

you’re right: adelete_by_doc_id is deprecated, and there's currently no full-path way to clear all traces (index, vector, UI, cache) in one go.

we’ve open-sourced a full diagnostic & solution library (MIT License):

👉 https://github.com/onestardao/WFGY/blob/main/ProblemMap/README.md

if you’re deploying LightRAG in prod, happy to show you how to force hard-sync or shadow-purge behavior — we built it for exactly these headaches.

just reply here if interested, I’ll walk you through the module.