@Marvin I'm running the following flow in a deployment:

from prefect import flow

@flow(name = "Test")
def test():
    print("Hello world!")

if __name__ == "__main__":
    test()

When the agent picks it up, I get this warning message:

snowflake.connector.cursor - Failed to import ArrowResult. No Apache Arrow result set format can be used. ImportError: DLL load failed while importing nanoarrow_arrow_iterator: The specified module could not be found.

It seems there are no open issues on GitHub that match your query. This warning message suggests there is a problem with loading Apache Arrow's shared libraries (DLLs) which are used by the

snowflake.connector.cursor

module of the Snowflake Connector for Python. These kinds of errors typically occur when: 1. The necessary binaries are not installed. 2. They are not on the system path where the loader expects to find them. 3. There is a version mismatch between the DLLs installed and the ones required by the module. Considering Prefect 2.x is out of my direct experience, I couldn't be more cheerful to suggest a few steps you might take: - Ensure that you have

pyarrow

installed in your environment as it provides the Apache Arrow library needed by Snowflake Connector. Use

pip install pyarrow

. - Verify that all environment variables are correctly set if required by the Snowflake Connector. - Review the setup of the deployment environment as these issues can be specific to the environment configuration. - Look into the deployment logs or system reports for additional details on the missing module or missing path. - If possible, try reinstalling the Snowflake Connector for Python. - Check the compatibility of the versions of the Snowflake Connector and pyarrow to ensure they work together as expected. If the issue persists, you may consider filing an issue on the prefecthq/prefect GitHub repository with details of your environment and the error, or reaching out to Prefect support for help. Would you like me to help with anything else?