Instantiation

The instantiation block describes how a Transformer is created by the Pipeline Engine. Two options are available: job_image (instantiate as a Kubernetes Job) and rdp_workflow (instantiate via the SDL workflow runtime).

`job_image`

job_image tells the Pipeline Engine to instantiate the Transformer as a Kubernetes Job. Available fields are as follows:

"job_image": {
    "image": "asdf",
    "pull_policy": "IfNotPresent", // Alternatives: Always, Never
    "image_pull_secret": "pat-name",
    "default_replicas": 2,
    "ports": [
        {
            "name": "http-port",
            "containerPort": 8080
        }
    ],
    "volumes": [ // A list of Kubernetes Volumes, just in JSON format instead of YAML
        {
            "name": "volume-1",
            "configMap": {
                "name": "volume-configmap",
                "items": {
                    "key": "config",
                    "path": "my-config.conf"
                }
            }
        }
    ],
    "volume_mounts": [ // A list of Kubernetes Volume Mounts, just in JSON format instead of YAML
        {
            "name": "volume-1",
            "mountPath": "/etc/config"
        }
    ],
    "args": [ // Alternative arguments passed to the main container at startup
        "tail", "-f", "/dev/null"
    ]
}

`rdp_workflow`

rdp_workflow tells the Pipeline Engine to instantiate the Transformer via the SDL workflow runtime in a one-shot configuration. This is used for transformers with the rdp_workflow type.

"rdp_workflow": {
    "definition_id": "my-custom-definition-id"
}

definition_id: Optional. If present, overrides the Transformer template’s UID when searching for the appropriate node definition in the workflow runtime. If omitted, the template’s own UID is used.