Metadata for custom ML Runtimes

This topic addresses the metadata for custom ML Runtimes.

All new custom ML Runtimes must override the Edition metadata of existing ML Runtimes. The rest of the metadata can be overridden to communicate the expectations for the users of the image. Both the Docker label and the environment variable match in a custom ML Runtimes image. In order to add or register a custom ML Runtimes to a deployment, the user facing metadata combination must be unique in that deployment. For example, among Editor, Edition, Kernel, Version, and Maintenance Version, at least the Maintenance Version must be incremented when publising a new iteration of the same image.

See the following reference table for more details on ML Runtime metadata.


Environment variable	Docker Label	Description	Override in custom non-PBJ ML Runtimes	Value in PBJ ML Runtimes
`ML_RUNTIME_METADATA_VERSION`	`com.cloudera.ml.runtime.runtime-metadata-version`	Metadata version	Not allowed	Must be set to: 2 for Ubuntu-based ML Runtimes 3 for Chainguard-based ML Runtimes.
`ML_RUNTIME_EDITOR`	`com.cloudera.ml.runtime.editor`	Cloudera AI Editor installed in the image.	Allowed	Required
`ML_RUNTIME_EDITION`	`com.cloudera.ml.runtime.edition`	Edition of the image, a notion of the ML Runtimes capabilities.	Required	Required
`ML_RUNTIME_DESCRIPTION`	`com.cloudera.ml.runtime.description`	Longer description of the ML Runtimes image capabilities.	Recommended	Required
`ML_RUNTIME_KERNEL`	`com.cloudera.ml.runtime.kernel`	Main kernel included in the image, for example, Python 3.8.	Not allowed	Required
`ML_RUNTIME_SHORT_VERSION`	`com.cloudera.ml.runtime.short-version`	Main version of the image, for example, 1.0. This shows up as Version in the selection screen.	Recommended	Required
`ML_RUNTIME_FULL_VERSION`	`com.cloudera.ml.runtime.full-version`	The complete version consists of the short version and the maintenance version, for example, 1.0.1	Optional	Optional
`ML_RUNTIME_MAINTENANCE_VERSION`	`com.cloudera.ml.runtime.maintenance-version`	The maintenance version must be an integer, for example, 1. Only the largest maintenance version of the set of the same short version images are visible for users to select. Increment this number if you create a drop in replacement of an existing ML Runtimes. Start with 1 with a new version or edition.	Recommended	Required
`ML_RUNTIME_CUDA_VERSION`	`com.cloudera.ml.runtime.cuda-version`	CUDA version installed in the image.	Not allowed	Do not set
`ML_RUNTIME_GIT_HASH`	`com.cloudera.ml.runtime.git-hash`	Git hash of ML Runtimes source	Not allowed	Do not set
`ML_RUNTIME_GBN`	`com.cloudera.ml.runtime.gbn`	Cloudera internal build number	Not allowed	Do not set