There is an instance in Data Science Pipelines code where the Tekton templater expects to find a Secret named exactly `mlpipeline-minio-artifact`.   This makes deployments inflexible and in general is bad practice, fix to source the secret values from environment vars instead.