mlflow health endpoint is doutbful

[pebeto]

mlflow local instances provide /health and /ping endpoints to check if it is running. Platforms like DagsHub don't have it (well, as far as I have been able to test), making users being unable to use this package. Should we remove that check?

[pebeto]

If we are not going to remove this check, we must search a way to detect if the provided URI is a valid mlflow instance.

[ablaom]

Just to be clear, are you saying the current health checks don't use public API?

And this has gotten us into trouble, because our checks are crashing sessions via DagsHub?

If the answer to both questions is "yes" then I agree we need to remove the check. In any case, this ought to be something we implement at MLFlowClient.jl, yes? I don't how to officially check instance health, but we could post a question at mljflow.

[pebeto]

The current healthcheck is using a non REST endpoint to show mlflow instance status. This endpoint is enabled by default with the common distribution. However, platforms like DagsHub removed it by unknown reasons. So, yes to both questions.

We need to request a healthcheck inside the REST API to have an official way to do this. If that's not the case, we can play with the webpage to extract certain parts to ensure an instance (but that's awful).