Hi, I was facing a problem with metaflow DAG preview and artifact inspection which I asked here. The solutions proposed and answered kindly by @square-wire-39606 was not feasible due to the tools that I had access to. I found a way which I found by going through metaflow docs and questions for this problem, and shared it in this article. https://medium.com/@rezashabrang.m/configuring-metaflow-for-on-prem-deployment-f8c4da24a097 The way I needed to deploy this S3 minio bucket was by only using docker and I did not want to use ingress, helm and kubernetes solution here. Hope this article helps anyone else going through this problem like myself and wanted a straightforward way for deploying metaflow with full functionalities locally. Thanks!

Eeeeey! Very cool 😎

Cheers! An excellent writeup on the setup, should prove useful to many 🙂 I had a quick question regarding the images you used, specifically about the custom built

metaflow-ui-backend:latest

. Was there some secret sauce required to get the backend working? The

netflixoss/metaflow_metadata_service:v2.4.3

image should work for all the services as it bundles all of them in the same release. Varying the CMD for the container is the way to choose which service to launch, and for the ui_backend it should be

/opt/latest/bin/python3 -m services.ui_backend_service.ui_server

Hi, There is no secret sauce about the ui backend, no. I just forked the metaflow service repo and built the images on github actions due to some restrictions that I had. And thank you for pointing out that ui backend is also bundled in netflixoss image I was not aware of that.