Concurrency model for updating data

I want to store information in the rows of the metadata. For example metadata= metadata={"count": 5} I want concurrent processes to be able to increment the count. Does pinecone expose any consistency mechanism to implement a reader/writer lock here? Checking data freshness with describe_index_stats() doesn’t help here if the vector count isn’t changing.

Hi @eds228, and welcome the Pinecone community forums!

Thank you for your question.

Would you mind sharing more about your use case and application setup?

The Pinecone SDKs don’t expose any locking mechanisms as you’re describing here, because their focus is on performant upserts and queries.

You’d have to implement locks at the application level.

These docs don’t correlate exactly to what you’re asking about, but I thought they might be of interest to you to review if you’re thinking of going down this path as they demonstrate how to implement parallel upserts.

You may be able to combine some of these patterns with application-level locking to achieve your end goal.