How Can I Write Logs Directly to AWS S3 from Memory Without First Writing to stdout? (Python, boto3)

Question

I'm trying to write Python log files directly to S3 without first saving them to stdout. I want the log files to be written to S3 automatically when the program is done running. I'd like to use the boto3 put_object method:

import atexit
import logging

import boto3  


def write_logs(body, bucket, key):
    s3 = boto3.client("s3")
    s3.put_object(Body=body, Bucket=bucket, Key=key)

log = logging.getLogger("some_log_name")
log.info("Hello S3")

atexit.register(write_logs, body=log, bucket="bucket_name", key="key_name")

quit()

This throws an error when uploading to S3. If i remember correctly, it requires that the object uploaded to S3 must be bytes-like. I'll update the question with the exact error once I have time to recreate the issue.

Answer 1

You need to add a couple things here. First, create a StringIO object. Then, write the logs to the StringIO object using a logging StreamHandler. Add the handler to your logger. Finally, call the getvalue() method on the StringIO object. You can write that to S3.

import atexit
import io
import logging

import boto3  


def write_logs(body, bucket, key):
    s3 = boto3.client("s3")
    s3.put_object(Body=body, Bucket=bucket, Key=key)  


log = logging.getLogger("some_log_name")
log_stringio = io.StringIO()
handler = logging.StreamHandler(log_stringio)
log.addHandler(handler)

atexit.register(write_logs, body=log_stringio.getvalue(), bucket="bucket_name", key="key_name")

log.info("Hello S3")

quit()