mlflow server #11

Supports: kubernetes

Deploy this charm on Kubernetes with the CLI. Find out how by reading the docs.

Description

MLflow


mlflow

Description

A charm which provides a Kubernetes installation of the MLflow server.

Optionally, it knows how to interface with the Kubeflow charms to configure Kubeflow notebook clients with access to the MLflow server.

Usage

TODO: Provide high-level usage, such as required config or relations

Developing

Create and activate a virtualenv with the development requirements:

virtualenv -p python3 venv
source venv/bin/activate
pip install -r requirements-dev.txt

Testing

The Python operator framework includes a very nice harness for testing operator behaviour without full deployment. Just run_tests:

./run_tests

Configuration

kubeflow_nodeport
(int) The nodeport for Kubeflow (istio-ingressgateway)
30600
kubeflow_port
(int) The port istio-ingressgateway will be listening on
80
minio_nodeport
(int) The nodeport for Minio
30650
minio_port
(int) The port Minio will be listening on
9000
mlflow_nodeport
(int) The nodeport for MLFlow
31380
mlflow_port
(int) The port MLFlow will be listening on
5000
mlflow_scrape_interval
(string) How frequently to scrape targets from MLFlow.
None
mlflow_scrape_timeout
(string) Per-scrape timeout when scraping MLFlow.
None