As an alternative to the AIStore Python SDK for accessing AIStore, you might prefer using other popular object storage client libraries. For example, you can use Amazon's Boto3 library, or its underlying botocore library.

This package aistore.botocore_patch.botocore exposes an interface to access AIStore as if it were Amazon S3, allowing developers to utilize AIStore object storage without changing their existing S3 client code.

By default, botocore doesn't handle HTTP redirects, which prevents you from using it with AIStore.

To resolve this, install aistore with the botocore extra, and then import aistore.botocore_patch.botocore in your code. This will dynamically patch HTTP redirect support into botocore, via monkey patch.

$ pip install aistore[botocore]

import boto3
from aistore.botocore_patch import botocore

When Authentication (AuthN) is enabled on the AIStore server, AIStore expects a JWT authorization token with each request to grant the required permissions. Normally, boto3 signs requests by adding complex signature and signing information to the Authorization header. However, AIStore doesn’t need this extra information—just the token.

To handle this in boto3, you can modify the Authorization header to include your JWT token before the request is made to AIStore.

# Create the S3 client
s3 = boto3.client(
    "s3",
    endpoint_url="http://localhost:8080/s3",
)

# Custom request hook to inject the Authorization header
def add_auth_header(request, **kwargs):
    request.headers['Authorization'] = "Bearer <token>"

# Attach the request hook to modify headers
s3.meta.events.register('before-send.s3.*', add_auth_header)

This setup replaces the Authorization header with the correct token before the request is sent to the AIStore server.

• AIStore GitHub
• Documentation
• Blog
• AIStore pip package
• KubeCon 2024 presentation