[Bug] `client.list_indexes()` kicks out an unrelated error

[jul-carras]

• I believe this is a new bug
• I have searched the existing Github issues and Community Forum, and I could not find an existing post for this bug

Describe the bug
After getting the client object using our API Key, using client.list_indexes() returns an odd error.

Error information

pinecone.core.client.exceptions.PineconeApiTypeError: 
Invalid type for variable 'dimension'. 
Required value type is int and passed type was NoneType at ['received_data']['indexes'][1]['dimension']

Steps to reproduce the issue locally

import os
from pinecone import Pinecone

pc = Pinecone(api_key=os.environ.get("PINE_API_KEY"))
pc.list_indexes()

Environment

• OS Version: Sequoia 15.2
• Python version: 3.11.3
• Python SDK version: 4.1.2

Additional context
This post suggests it could be a pod vs serverless issue, but the error message is really strange. If this is the case, the message should reflect that better.

Hitting the API directly seems to work:

curl -i -X GET "https://api.pinecone.io/indexes" \
-H "Api-Key: $PINECONE_API_KEY" \
-H "X-Pinecone-API-Version: 2024-10"

[skittlejohn]

Also experiencing this issue with pinecone-client@5.0.1. The .list_indexes call is embedded into a handful of other methods which all break as well.

Would love to vouch for using Pinecone in production; hard to do so with this bug.

For instance, the .delete_index call fails but only after the index is deleted, leading to very inconsistent behavior. Index gets deleted successfully but then errors out on internal confirmation of the delete operation with no notice given in the stack trace to the delete operation's success.

[skittlejohn]

I see that this will be fixed when version 6 is stable. Coding around it for now.

[anawishnoff]

Hi all, the issue here is likely that there's an index in the project that doesn't have a dimension -- this should be handled in the latest version of the SDK, but let us know if theres still an issue.