Job

A representation of a Kubernetes Job in Piceli.

Kubernetes Jobs create one or more pods and ensure that a specified number of them successfully terminate. When the specified number of pods terminates successfully, the job is complete. The Job class in Piceli allows for defining these jobs, including their behavior, lifecycle, and cleanup strategy.

Properties

• cleanup_after_seconds: (Optional) Time in seconds after the job finishes to automatically clean up the job’s pods. If not specified, pods are not automatically cleaned up.

• backoff_limit: (Optional) Specifies the number of retries before considering the job as failed. Defaults to Kubernetes’ default if not specified.

• labels: (Optional) Labels to be applied to the job for identification and selection.

Job Cleanup and Retry Policy

One of the key features of a Kubernetes Job is controlling the job’s lifecycle, including cleanup after completion and the strategy for retrying failed jobs. Piceli’s Job class encapsulates these functionalities, providing an easy way to define and manage them.

• cleanup_after_seconds: Automatically cleans up job pods after the specified duration since job completion.

• backoff_limit: Controls the number of retries for a failed job. A job is considered failed if the pod exits with a non-zero status.

Restrictions

• Restart Policy: Kubernetes Jobs do not support a restart policy of Always. The only valid values are OnFailure and Never. This is enforced by the Piceli Job class to align with Kubernetes API requirements.

Example

from piceli.k8s import templates

job = templates.Job(
        name="test-job",
        image_pull_secrets=["docker-registry-credentials"],
        backoff_limit=1,
        containers=[
            templates.Container(
                name="test-job",
                command=["python", "--version"],
                image="docker-image",
                env={"K0": "V0"},
                liveness_command=[
                    "sh",
                    "-c",
                    "test $(expr $(date +%s) - $(cat /tmp/health_check)) -lt 60",
                ],
                resources=templates.Resources(
                    cpu="100m", memory="250Mi", ephemeral_storage="11Mi"
                ),
            )
        ],
        template_labels={"pod_name": "test-job"},
        labels={"job_name": "test-job"},
    )

In this example, a Job is defined to perform batch processing, with automatic cleanup of its pods 1 hour after the job finishes and a backoff limit of 3 retries.