You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@lucene.apache.org by "onestardao (via GitHub)" <gi...@apache.org> on 2025/08/07 15:32:30 UTC
Re: [I] How to catch potential issues with the new Faiss vector format? [lucene]
onestardao commented on issue #14875:
URL: https://github.com/apache/lucene/issues/14875#issuecomment-3164722322
You're absolutely right about the risks in native FAISS-based vector formats. We've actually documented multiple failure patterns in OpenSearch clusters that stem from semantic bootstrapping mismatches and untracked FAISS state transitions.
In our diagnostic map, this falls under **Problem No.14** and **No.15**, especially when the system silently proceeds with misconfigured engines and later crashes at runtime. (Sometimes even deletes the data directory.)
We’ve built a full test suite and detection framework for these edge cases. Happy to share details or help if you’re exploring structured mitigations.
Let us know if this is useful.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@lucene.apache.org
For additional commands, e-mail: issues-help@lucene.apache.org