I recently started using Pinecone Serverless and encountered an issue with deleting vectors. I’ve noticed that I’m consistently hitting a rate limit with a 429 error, even when I’m making only around 20 asynchronous API calls. I suspect that the rate limit number is set too low for my use case.
Is there a way to optimize the deletion process to reduce the number of API calls? Specifically, can I provide multiple prefixes in a single request that would return all the associated IDs, helping me avoid hitting the rate limit excessively?
Here’s the example CURL request I’m currently using:
curl -X GET “https://$INDEX_HOST/vectors/list?namespace=ns1&prefix=doc1#”
-H “Api-Key: $PINECONE_API_KEY”
curl -X GET “https://$INDEX_HOST/vectors/delete?ids=doc1#chunk1&ids=doc1#chunk2&ids=doc1#chunk3&ids=doc1#chunk4&namespace=example-namespace”
-H “Api-Key: $PINECONE_API_KEY”