@Marvin

AwsCredentials(region_name=None, profile_name=None, aws_access_key_id=None, aws_session_token=None, **aws_client_parameters={'config': None, 'verify': True, 'use_ssl': True, 'api_version': None, 'endpoint_url': None, 'verify_cert_path': None}**, aws_secret_access_key=None)

Which leads to

TypeError: unhashable type: 'dict'

whenever I try to work with the S3 bucket

The issue you're encountering seems to revolve around how the

AwsCredentials

and

aws_client_parameters

are being handled within the Prefect framework, specifically leading to a

TypeError: unhashable type: 'dict'

. ### Key Points: 1. AWSCredentials Block: - The

AwsCredentials

class can be used within an

S3Bucket

block to provide authentication when interacting with an S3 bucket. - Example usage:

my_s3_bucket = S3Bucket(
         bucket_name="my_s3_bucket",
         credentials=AWSCredentials.load("my_aws_credentials")
     )

2. Reported Issue: - The error

TypeError: unhashable type: 'dict'

occurs due to the way

aws_client_parameters

are stored. - This is often due to incorrect storage or handling of the parameters, especially when assigned through the UI. 3. Possible Solution: - Ensure that

aws_client_parameters

is properly initialized and not directly passed as a dictionary. - Use the helper methods to initialize

AwsCredentials

properly. ### GitHub Issues: - Issue #14147: Discusses the

TypeError

with similar configurations. - Issue #13349: Highlights how

aws_client_parameters

being stored as a dictionary leads to errors. References: - For more details on configuring and using

AwsCredentials

, you can refer to the Prefect documentation on Blocks and AWS integrations. - GitHub Issue: AWS Client Parameters Issue Try ensuring the

AwsCredentials

are properly instantiated according to the documentation, possibly avoiding direct dictionary assignment in the parameters. If the error persists, consult the open issues for updates or consider opening a new issue if your scenario slightly differs.

hi @Blake - is this an older version of prefect aws?

Thanks @Nate

hrm thats interesting - can you show the code that gets you to this error? ideally an MRE would be super helpful

Updating to latest prefect-aws + prefect 3.0 has resolved thid

👍