Closed ChristinaLast closed 4 months ago
Also, it says that is is logging to S3 but the bucket hasnt been updated since January 2023: https://s3.console.aws.amazon.com/s3/buckets/openaq-mlflow-bucket?region=us-east-1&tab=objects
openaq-engine run-pipeline local_data models --pollutant pm25 --source openaq-api --country US
use this command to test.
Also, for local env variables change MLFLOW_TRACKING_URI=https://localhost:5000
closed in #53
So there is an MLFLOW instance running on http://3.82.115.10:5000/
however there is nothing being logged to the online mlflow server. Ideally we have:
However, there is evidence some of it is working. e.g.
/home/ec2-user/openaq-engine/3.82.115.10:5000/201015484053809956/139d9896058446dfb8213e5aa7b44915/meta.yaml
is a path of metrics fromm modmel runs that are being logged in the correct directory (just that server isnt "picking up on that"The place we want mlflow to complete (e.g. when we are in the more "classical" mlflow pipeline area.