Configuring logging

[pcoccoli]

IGNORE

[dosubot[bot]]

Hey @pcoccoli! I'm here to assist you with any bugs, questions, or contributions you have. Let's tackle this logging configuration together!

I couldn't find specific documentation on configuring logging levels for the langchain_aws module in Langfuse. However, you can generally control logging in Python by using the logging module. Here's a basic example of how you might suppress log messages from a specific logger:

import logging

# Suppress log messages from the 'langchain_aws' logger
logging.getLogger('langchain_aws').setLevel(logging.WARNING)

This code sets the logging level for the langchain_aws logger to WARNING, which means it will ignore INFO and DEBUG messages. You can adjust the level to ERROR or CRITICAL if you want to suppress more messages.

Additionally, you can enable debug mode in the Langfuse Python SDK to get verbose logs by setting the debug parameter to True when initializing the Langfuse client or by setting the environment variable LANGFUSE_DEBUG to True ^[1][2].

To continue talking to Dosu, mention @dosu.

[pcoccoli]

Wrong project!

[pcoccoli]

In case anyone actually looks at this, when I use any langfuse version past 2.51.5, I get tons of unwanted messages printed to stdout. Something in langfuse seems to mess with python logging and some loggerMap nonsense floods stdout. Unfortunately I didn't save that output so I can show it here; I will come back to it next time I need to upgrade, I'm sure!

[pcoccoli]

Here's the example from this discussion's revision history (before I deleted it):

{'loggerMap': {<Logger langchain_aws.vectorstores.inmemorydb.base (INFO)>: None}}

It's a langchain message but that's just a coincidence; downgrading to langfuse 2.51.5 makes it stop.

[hassiebp]

@pcoccoli What kind of Langchain app are you running? It would be great to have some minimal repro example.