riseml/cronjobs repository overview

⁠About

This is the central API server of RiseML. It provides endpoints for clients and all access to the database which contains user and experiment information. The API server can be used, for example, to:

• Start and stop experiments
• List experiments and their status
• Get information about the cluster, its nodes and GPUs

The API server communicates to other components like the scheduler⁠ via a message queue. Only the API server directly accesses the database. Information is pushed to the database via the API by other components, e.g., by the scheduler and the monitor⁠.

⁠Contributing

RiseML welcomes contributions from all developers. The high level process for development matches many other open source projects. See below for an outline.

• Fork this repository
• Make your changes
• Submit a pull request (PR) to this repository with your changes, and unit tests whenever possible.
• The RiseML core contributors will review your code. After they sign off on your code, they'll label your PR with LGTM. Once that happens, the contributors will merge it.

⁠Environment Variables

The following environment variables are available for this component:

⁠Required

• CLUSTER_ID: UUID of this cluster
• VERSION: Code version
• DATABASE_URL: URL for database, e.g., postgresql://username:password@localhost:31543/riseml
• AMQP_URL: URL for amqp, e.g., amqp://localhost:30567
• GITWEB_URL: URL for gitweb component, e.g., http://localhost:31213/git

⁠Optional

• ENVIRONMENT: whether this is a production, staging or development instance, default: development
• AMQP_STATE_EXCHANGE: name of the exchange where experiment state updates are sent, default: states
• AMQP_SCHEDULER_QUEUE: name of the queue where scheduler notifications are sent, default: scheduler
• IMAGE_REGISTRY_INTERNAL: the image registry when accessed from inside Kubernetes, i.e. the overlay network, e.g.: riseml-registry:5000
• IMAGE_REGISTRY_EXTERNAL: the image registry when accessed a bare host/node in the cluster, e.g.: localhost:31500
• IMAGE_REPOSITORY: name of the build repository in the image registry, default: build
• IMAGEBUILDER_IMAGE: the image name of the dockerbuild image, default: riseml/imagebuilder
• IMAGEBUILDER_NODE_SELECTORS: the Kubernetes node selectors on where to run the imagebuilder, default: empty
• GPU_NODE_SELECTORS: the Kubernetes node selectors that identify nodes that have a GPU, default: empty
• BACKEND_URL: URL for the backend, default http://riseml.com/backend⁠
• OUTPUT_ROOT: directory where the output volume is mounted, default: /output