In the local metaflow deployment: https://github.com/outerbounds/local_metaflow_deployment What AWS creds should be specified in the .env? Also, should the flows be saved locally or in s3 bucket?. Thank you very much 🙏

it works with S3-based datastore. You can set up your AWS credentials as usual, making sure the credentials allow you to access an S3 bucket of your choice

Ok. Where should I place python flows code, for it to be visible inside of the metaservice_ui?

you can run it anywhere as long as your configuration points at the right service

Thank you very much 😊 Would this be sufficient? { "METAFLOW_DATASTORE_SYSROOT_S3": "s3://sparkhudi/metaflow", "METAFLOW_DATATOOLS_SYSROOT_S3": "s3://sparkhudi/metaflow/data", "METAFLOW_DEFAULT_DATASTORE": "s3", "METAFLOW_DEFAULT_METADATA": "service", "METAFLOW_SERVICE_URL": "http://localhost:3000", "METAFLOW_SERVICE_INTERNAL_URL": "http://localhost:8080" }

looks pretty good

UI is up an running on localhost:3000, but get requests fail. Service is up and running with no errors, postgres is up and running - no errors, mfdeploy-ui (nginx) shows receiving get requests in logs. And nothing has been created in the s3 bucket yet. Any hints?)

@cold-ability-47914 are you able to run a flow successfully? Can you run one successfully with cards? (‘python my flow.py—with cards’) if that works then we can say that the S3 connection is ok

I deployed to AWS with Argo terraform example instead - works fine. Will try custom AMI with docker image on AWS batch next as it seems easier than Argo.