AWS Step Functions で再実行ができるようになったので、前回のシンプル編に引き続き、ジョブネットのエラー対応をどれだけ実現できるか試してみます。

想定するジョブネット

今回はジョブが選択と分岐の形状と想定をして実験してみます。今回はジョブ２－２とジョブ４－２がそれぞれ異常終了したときの挙動を確認します。また、異常終了したジョブと並行して実行されるジョブ（ジョブ２－１や４－１）の完了タイミングが、異常終了の前になるか後になるかで多少の挙動の違いがあるためそこも実験していきます。

実装の概要

前回と同じくaws-samples の amazon-stepfunctions-ssm-waitfortasktoken を利用した SfnRunCommandByInstanceIds を用いて、Steps Functions に以下のようなステートマシンを定義しました。並行稼働部は Parallel を使用しています。

{
  "Comment": "This is a test state machine by yki",
  "StartAt": "StartAutomationWaitForCallBack (Job 1)",
  "States": {
    "StartAutomationWaitForCallBack (Job 1)": {
      "Type": "Task",
      "Resource": "arn:aws:states:::aws-sdk:ssm:startAutomationExecution.waitForTaskToken",
      "Parameters": {
        "DocumentName": "SfnRunCommandByInstanceIds",
        "Parameters": {
          "InstanceIds": [
            "i-xxxxxxxxxxxxxxxxx"
          ],
          "taskToken.$": "States.Array($$.Task.Token)",
          "workingDirectory": [
            "/home/ec2-user/sample-batch/"
          ],
          "Commands": [
            "./do-some1.sh"
          ]
        }
      },
      "Next": "Parallel"
    },
    "Parallel": {
      "Type": "Parallel",
      "Next": "StartAutomationWaitForCallBack (Job 3)",
      "Branches": [
        {
          "StartAt": "StartAutomationWaitForCallBack (Job 2-1)",
          "States": {
            "StartAutomationWaitForCallBack (Job 2-1)": {
              "Type": "Task",
              "Resource": "arn:aws:states:::aws-sdk:ssm:startAutomationExecution.waitForTaskToken",
              "Parameters": {
                "DocumentName": "SfnRunCommandByInstanceIds",
                "Parameters": {
                  "InstanceIds": [
                    "i-xxxxxxxxxxxxxxxxx"
                  ],
                  "taskToken.$": "States.Array($$.Task.Token)",
                  "workingDirectory": [
                    "/home/ec2-user/sample-batch/"
                  ],
                  "Commands": [
                    "./do-some2-1.sh"
                  ]
                }
              },
              "End": true
            }
          }
        },
        {
          "StartAt": "StartAutomationWaitForCallBack (Job 2-2)",
          "States": {
            "StartAutomationWaitForCallBack (Job 2-2)": {
              "Type": "Task",
              "Resource": "arn:aws:states:::aws-sdk:ssm:startAutomationExecution.waitForTaskToken",
              "Parameters": {
                "DocumentName": "SfnRunCommandByInstanceIds",
                "Parameters": {
                  "InstanceIds": [
                    "i-xxxxxxxxxxxxxxxxx"
                  ],
                  "taskToken.$": "States.Array($$.Task.Token)",
                  "workingDirectory": [
                    "/home/ec2-user/sample-batch/"
                  ],
                  "Commands": [
                    "./do-some2-2.sh"
                  ]
                }
              },
              "End": true
            }
          }
        }
      ]
    },
    "StartAutomationWaitForCallBack (Job 3)": {
      "Type": "Task",
      "Resource": "arn:aws:states:::aws-sdk:ssm:startAutomationExecution.waitForTaskToken",
      "Parameters": {
        "DocumentName": "SfnRunCommandByInstanceIds",
        "Parameters": {
          "InstanceIds": [
            "i-xxxxxxxxxxxxxxxxx"
          ],
          "taskToken.$": "States.Array($$.Task.Token)",
          "workingDirectory": [
            "/home/ec2-user/sample-batch/"
          ],
          "Commands": [
            "./do-some3.sh"
          ]
        }
      },
      "Next": "Parallel (1)"
    },
    "Parallel (1)": {
      "Type": "Parallel",
      "End": true,
      "Branches": [
        {
          "StartAt": "StartAutomationWaitForCallBack (Job 4-1)",
          "States": {
            "StartAutomationWaitForCallBack (Job 4-1)": {
              "Type": "Task",
              "Resource": "arn:aws:states:::aws-sdk:ssm:startAutomationExecution.waitForTaskToken",
              "Parameters": {
                "DocumentName": "SfnRunCommandByInstanceIds",
                "Parameters": {
                  "InstanceIds": [
                    "i-xxxxxxxxxxxxxxxxx"
                  ],
                  "taskToken.$": "States.Array($$.Task.Token)",
                  "workingDirectory": [
                    "/home/ec2-user/sample-batch/"
                  ],
                  "Commands": [
                    "./do-some4-1.sh"
                  ]
                }
              },
              "End": true
            }
          }
        },
        {
          "StartAt": "StartAutomationWaitForCallBack (Job 4-2)",
          "States": {
            "StartAutomationWaitForCallBack (Job 4-2)": {
              "Type": "Task",
              "Resource": "arn:aws:states:::aws-sdk:ssm:startAutomationExecution.waitForTaskToken",
              "Parameters": {
                "DocumentName": "SfnRunCommandByInstanceIds",
                "Parameters": {
                  "InstanceIds": [
                    "i-xxxxxxxxxxxxxxxxx"
                  ],
                  "taskToken.$": "States.Array($$.Task.Token)",
                  "workingDirectory": [
                    "/home/ec2-user/sample-batch/"
                  ],
                  "Commands": [
                    "./do-some4-2.sh"
                  ]
                }
              },
              "End": true
            }
          }
        }
      ]
    }
  }
}

エラーなく実行できることも確認済です。