Johnpaul

This isn’t an MLflow bug it’s a tracing provider conflict in Databricks 17.3. Databricks already installs and initializes its own OpenTelemetry TracerProvider, so your custom block: trace._TRACER_PROVIDER = None trace.set_tracer_provider(tracer_provider)...

I ran into this same issue last week, it happens when Docker appends the folder suffix and Kong loses the route to Storage. I fixed it by overriding the docker...

@MarcAntoineCarrier28 Yep, I hit this too, Docker folder suffix breaks Kong’s routing to Storage. I have a patch that locks container names and gets everything running reliably. It’s a few...