[Argo Workflow] Workflow 작성하기

💡 Argo Workflow에서 다양한 형태의 workflow를 작성하는 방법을 알아보자

1. Single Job

# single-job.yaml
apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  generateName: hello-world-
  namespace: default
spec:
  entrypoint: whalesay
  templates:
  - name: whalesay
    container:
      image: docker/whalesay
      command: [cowsay]
      args: ["hello world"]
      resources:
        limits:
          memory: 32Mi
          cpu: 100m

파라미터전달

# param.yaml
apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  generateName: hello-world-parameters-
  namespace: default
spec:
  entrypoint: whalesay
  arguments:
    parameters: ###### job에 아래 args를 넘길게요
    - name: message
      value: hello world through param

  templates:
  ###############
  # entrypoint
  ###############
  - name: whalesay
    inputs:
      parameters: ###### 나는 이런 녀석들을 파라미터로 받아요
      - name: message 
    container:
      image: docker/whalesay
      command: [cowsay]
      args: [""] ###### message라는 파라미터를 사용할게요

2. Serial Step

간단한 순차실행은 step 으로, double dash - - 를 사용해 나타낸다.

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  generateName: serial-step-
  namespace: default
spec:
  entrypoint: hello-step
  templates:

  ###############
  # template job 템플릿
  ###############
  - name: whalesay
    inputs:
      parameters:
      - name: message
    container:
      image: docker/whalesay
      command: [cowsay]
      args: [""]

  ###############
  # entrypoint 엔트리포인트 
  ###############
  - name: hello-step
    # 순차 실행
    steps:  ##### step 으로 구분해요 
    - - name: hello1
        template: whalesay
        arguments:
          parameters:
          - name: message
            value: "hello1"
    - - name: hello2
        template: whalesay
        arguments:
          parameters:
          - name: message
            value: "hello2"
    - - name: hello3
        template: whalesay
        arguments:
          parameters:
          - name: message
            value: "hello3"

이렇게 정의하면

hello1 → hello2 → hello3

위 순서로 순차실행한다

3. Parallel Step

병렬실행은 single dash - 를 사용해 나타낸다

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  generateName: parallel-steps-
  namespace: default
spec:
  entrypoint: hello-step
  templates:

  ###############
  # template job
  ###############
  - name: whalesay
    inputs:
      parameters:
      - name: message
    container:
      image: docker/whalesay
      command: [cowsay]
      args: [""]

  ###############
  # entrypoint
  ###############
  - name: hello-step
    # 병렬 실행
    steps:  ##### step 으로 구분해요 
    - - name: hello1
        template: whalesay
        arguments:
          parameters:
          - name: message
            value: "hello1"
    - - name: hello2
        template: whalesay
        arguments:
          parameters:
          - name: message
            value: "hello2"
      - name: hello3        # 기존 double dash에서 single dash로 변경
        template: whalesay
        arguments:
          parameters:
          - name: message
            value: "hello3"

hello1 → hello2, hello3

hello 1 → hello2 는 더블대쉬로 순차실행

hello2, hello3 은 싱글대쉬로 병렬실행

4. DAG

좀더 복잡한 순서는 DAG(Directed asyclic graph : 방향성 비순환 그래프)를 통해 나타낸다

간단한녀석들은 step으로 나타냈지만, 좀더 복잡한 (ex. 다이아몬드) workflow는 dag로 나타낸다

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  generateName: dag-diamond-
  namespace: default
spec:
  entrypoint: diamond
  templates:
  
  ###############
  # template job
  ###############
  - name: echo
    inputs:
      parameters:
      - name: message
    container:
      image: alpine:3.7
      command: [echo, ""]

  ###############
  # entrypoint
  ###############
  - name: diamond
    # DAG 구성
    dag:  ##### dag로 구분해요 
      tasks:
      - name: A
        template: echo
        arguments:
          parameters: [{name: message, value: A}]
      - name: B
	        dependencies: [A]  ##### dependency 를 나타내요 
        template: echo
        arguments:
          parameters: [{name: message, value: B}]
      - name: C
        dependencies: [A]
        template: echo
        arguments:
          parameters: [{name: message, value: C}]
      - name: D
        dependencies: [B, C]
        template: echo
        arguments:
          parameters: [{name: message, value: D}]

선행조건을 dependencies 에 표기한다. 디펜던시가 걸린 녀석들은 선행조건이 만족되고 나서야 실행된다

A → (B, C) → D (다이아몬드모양..;) workflow로 진행된다

6. Exit Handling

종료시점에 특정작업을 수행하도록 종료작업 핸들링이 가능하다

apiVersion: argoproj.io/v1alpha1
kind: Workflow
metadata:
  generateName: error-handlers-
  namespace: default
spec:
  entrypoint: intentional-fail
  # 에러 핸들러 작업 지정
  onExit: error-handler ##### entrypoint 작업이 끝나면 자연히 이녀석이 실행돼요 

  templates:
  
  ###############
  # template job
  ###############
  - name: send-email
    container:
      image: alpine:latest
      command: [sh, -c]
      args: ["echo send e-mail:  "]
  
  ###############
  # 종료 핸들러
  ###############
  - name: error-handler
    steps:
    - - name: notify
        template: send-email

  ###############
  # entrypoint
  ###############
  - name: intentional-fail
    container:
      image: alpine:latest
      command: [sh, -c]
      args: ["echo intentional failure; exit 1"]

7. Cron Workflow

workflow와 기본적으로는 동일한데, 스케쥴정보를 넣어 cron job을 돌릴수있다

apiVersion: argoproj.io/v1alpha1
kind: CronWorkflow
metadata:
  name: test-cron-wf
spec:
  schedule: "* * * * *" #### 원하는 스케쥴(cron)을 작성
  concurrencyPolicy: "Replace"
  startingDeadlineSeconds: 0
  workflowSpec:
    entrypoint: whalesay
    templates:
    - name: whalesay
      container:
        image: alpine:3.6
        command: [sh, -c]
        args: ["date; sleep 90"]

cronWorkflow 전용 옵션은 다음과같이 있다

option name	default value	desc
schedule	none	Schedule at which the Workflow will be run. E.g. 5 4 * * *
timezone	Machine timezone	Timezone during which the Workflow will be run from the IANA timezone standard, e.g. America/Los_Angeles, Asia/Seoul
suspend	false	if true Workflow scheduling will not occur. Can be set from the CLI, GitOps, or directly
concurrencyPolicy	Allow	Policy that determines what to do if multiple Workflow are scheduled at the same time. Available options: Allow, Replace, Forbid
startingDeadlineSeconds	0	Number of seconds after the last successful run during which a missed Workflowwill be run
successfulJobsHistoryLimit	3	Number of successful Workflows that will be persisted at a time
failedJobsHistoryLimit	1	Number of failed Workflow that will be persisted at a time

https://github.com/bjpublic/core_kubernetes/tree/master/chapters/17 핵심만 콕! 쿠버네티스 https://argoproj.github.io/argo-workflows/cron-workflows/