Argo runner

• Overview
• Prerequisites
• Usage
• Configuration
• Development

Overview

Argo runner is a runner, which executes Argo workflows. It is used as a built-in runner for Capact Kubernetes implementation.

Prerequisites

• Go
• kubectl
• running Kubernetes cluster with Argo installed

Before starting the runner you need to:

1. Ensure the Service Account used by the Argo workflow has proper RBAC permissions. Here is an example command to add permissions for using default service account in default namespace:

kubectl create clusterrolebinding default-default-admin --clusterrole admin --serviceaccount default:default

2. Create a job to use for ownerReference for the workflow and a secret for status reporting:

kubectl apply -f cmd/argo-runner/setup.yml

Usage

1. Create the runner input YAML file:

cat <<EOF > /tmp/argo-runner-context.yaml
name: argo-runner-job
dryRun: false
platform:
  namespace: default
  ownerRef:
    apiVersion: batch/v1
    kind: Job
    name: argo-runner-owner
    uid: $(kubectl get jobs argo-runner-owner -ojsonpath='{.metadata.uid}')
EOF

2. Start the runner type:

RUNNER_CONTEXT_PATH=/tmp/argo-runner-context.yaml RUNNER_ARGS_PATH=cmd/argo-runner/example-args.yaml RUNNER_LOGGER_DEV_MODE=true go run cmd/argo-runner/main.go

You can check the workflow status in Argo UI on http://localhost:2746 after setting port-forwarding:

kubectl port-forward -n capact-system svc/argo-server 2746

Configuration

The following environment variables can be set:

Name	Required	Default	Description
RUNNER_CONTEXT_PATH	yes		Path to the YAML file with runner context
RUNNER_ARGS_PATH	yes		Path to the YAML file with input arguments
RUNNER_LOGGER_DEV_MODE	no	false	Enable additional log messages
KUBECONFIG	no	~/.kube/config	Path to kubeconfig file

Development

To read more about development, see the Development guide.