Version: Next

5.4.4 工作流插件

在OAM（Open Application Model）模型的部署过程中，其默认情况是无状态的部署方式。然而，随着应用架构的日益复杂化，我们有时会面临着一些需要逐步进行的、多阶段的部署需求。这种情况下，传统的一次性部署方法可能无法满足所有需求，从而带来诸多挑战。因此，为了解决这个问题并满足用户更多样化、复杂化的部署需求，我们参考OAM应用标准，引入了强大的workflow能力：此项功能能够对部署过程进行细致的管理和控制，使得复杂的部署需求能够按照预设的步骤和流程逐步执行，大大提升了部署的灵活性和可控性。
在SREWorks中，我们内置了5个工作流插件，也允许用户自定义工作流插件：

apply-components 按需部署组件代码地址
deploy 部署应用全量组件代码地址
remoteDeploy 部署应用全量组件至远端集群代码地址
rollout-promotion 灰度控制器，可以决定继续向下部署还是回滚代码地址
suspend 部署暂停代码地址

工作流案例（灰度发布）

通过在发布过程中不断调整 Helm / Kustomize 模板中对应 Workload spec 中的 partition 字段，来满足灰度发布的诉求。
如下，是一个典型的发布 Application 中的单个组件描述：

revisionName: "K8S_MICROSERVICE|clustermanage|_"
scopes:
  - scopeRef:
      apiVersion: apps.abm.io/v1
      kind: Namespace
      name: "{{ Global.deployNamespace }}"
dataInputs:
  - valueFrom:
      dataOutputName: "Global.clusterId"
    toFieldPaths:
      - spec.name
parameterValues:
  - name: values
    value:
      deployNamespace: "{{ Global.deployNamespace }}"
    toFieldPaths:
      - spec.values

在实现灰度方案后，原 Application 描述调整为：
需要朝 POST /workflow/launch发起下面的内容（和原来的 /deployments/launch 发起内容的区别在于多加了 workflow Section，其他 Section 的内容等价，不用改）。
注意发起 launch 的时候要加 category=DEPLOY (URL 参数)，否则不能使用回滚功能！(本质是给 workflow 打个分类)

components:
  - revisionName: "K8S_MICROSERVICE|clustermanage-test|_"
  - revisionName: "K8S_MICROSERVICE|clustermanage|_"
    scopes:
      - scopeRef:
          apiVersion: apps.abm.io/v1
          kind: Namespace
          name: "{{ Global.deployNamespace }}"
    dataInputs:
      - valueFrom:
          dataOutputName: "Global.clusterId"
        toFieldPaths:
          - spec.name
    parameterValues:
      - name: values
        value:
          deployNamespace: "{{ Global.deployNamespace }}"
        toFieldPaths:
          - spec.values
workflow:
  steps:
    - type: apply-components
      properties:
        components: 
        - type: K8S_MICROSERVICE
          name: clustermanage-test
    - type: apply-components
      properties:
        components: 
        - type: K8S_MICROSERVICE
          name: clustermanage
        rollout:  # 不提供 rollout 则直接原样部署
          type: partition  # 目前支持 partition(滚动更新) , 待支持 bluegreen / canary
          properties:
            partition: "1"
            name: "abmRolloutPartition"  # 可选，如果不填，默认是 abmRolloutPartition，否则可以自定义传入的参数 Name
    - type: suspend  # 此处将整个 workflow 置为暂停状态，这个时候需要外部通过 PUT /workflow/{instanceId}/context 来进一步设置变量，参考下面的说明
    - type: rollout-promotion  # 此处根据外界设置的 context 中的 continueRollout / rollback 决定是怎么走 (三种类型)
    - type: apply-components  # 如果上一步决定了继续向下执行的话，那么会继续正常部署
      properties:
        components: 
        - type: K8S_MICROSERVICE
          name: clustermanage

当 workflow 处于 suspend 节点暂停的时候，可以通过 PUT /workflow/{instanceId}/context 来设置下一步 rollout 使用的变量，允许传入的 context 内容如下：

灰度成功，确认并继续向下执行：{"continueRollout": true, "rollback": false}
灰度失败，决定回滚：{"continueRollout": false, "rollback": true}
灰度失败，就此结束，不执行操作：{"continueRollout": false, "rollback": false}

通过上述 context API 设置变量成功后，需要请求 PUT /workflow/{instanceId}/resume恢复 workflow 的执行
注意：当使用 suspend+rollout-promotion 类型的时候，务必保证 Workflow 发起的时候，使用的 options 中设置的 category 为 DEPLOY，否则无法使用回滚功能 (回滚即找到当前应用的上一个 category=DEPLOY 的 Application 重新全量部署)

工作流接口

发起 Workflow API

POST /workflow/launch
URL Parameters:

appId: 应用 ID
category: 分类 (固定为 DEPLOY 即可)

Body: YAML 字符串
Body 示例如下：

apiVersion: core.oam.dev/v1alpha2
kind: ApplicationConfiguration
metadata:
  annotations:
    appId: testapp
    appInstanceName: "test-app-instance"
  name: deploy-workflow
spec:
  parameterValues:
  components:
    - revisionName: K8S_MICROSERVICE|clustermanage|_
  policies:
    - name: topology-e2e-env
      type: topology
      properties:
        namespace: "abm-daily"
    - type: override
      name: override-e2e-env
      properties:
        components:
          - revisionName: K8S_MICROSERVICE|clustermanage|_
            dataInputs:
              - valueFrom:
                  dataOutputName: "Global.KUBECONFIG"
                toFieldPaths:
                  - spec.base64Kubeconfig
            parameterValues:
              - name: name
                value: "appmanager-helm-e2e-rollout"
                toFieldPaths:
                  - spec.name
              - name: ignoreCreateNamespace
                value: true
                toFieldPaths:
                  - spec.ignoreCreateNamespace
              - name: values
                value:
                  resources:
                    limits:
                      cpu: "0.5"
                      memory: "1Gi"
                  image:
                    repository: "nginx"
                    tag: "latest"
                  replicaCount: 3
                toFieldPaths:
                  - spec.values
  workflow:
    steps:
      - type: apply-components
        properties:
          policies:
            - topology-e2e-env
            - override-e2e-env
          components:
            - type: K8S_MICROSERVICE
              name: clustermanage
          rollout:
            type: partition
            properties:
              partition: "1"
              name: "abmRolloutPartition"
      - type: suspend
      - type: rollout-promotion
      - type: apply-components
        properties:
          policies:
            - topology-e2e-env
            - override-e2e-env
          components:
            - type: K8S_MICROSERVICE
              name: clustermanage

返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc1eed516690483758985249e69af",
    "timestamp": 1669048376027,
    "data": {
        "id": 1,
        "gmtCreate": 1669048376010,
        "gmtModified": 1669048376010,
        "appId": "testapp",
        "category": "DEPLOY",
        "gmtStart": null,
        "gmtEnd": null,
        "workflowStatus": "PENDING",
        "workflowErrorMessage": null,
        "workflowConfiguration": "yaml str here",
        "workflowSha256": "a71a1dcd0ac86e291f4442fe7477316f346e1275f91b1cd087b8e05b28fc9890",
        "workflowOptions": {
            "creator": "admin",
            "category": "DEPLOY"
        },
        "workflowCreator": "admin",
        "workflowContext": null,
        "lockVersion": null
    }
}

查询 Workflow Instance 列表 API

GET /workflow
URL Parameters (可选)

instanceId：Workflow 实例 ID
appId：应用 ID
workflowStatus：Workflow 状态 (可选PENDING/PREPROCESSING/RUNNING/SUSPEND/SUCCESS/FAILURE/EXCEPTION/TERMINATED)
workflowCreator：创建者

返回示例

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc1eed216690487987592481e34f2",
    "timestamp": 1669048798778,
    "data": {
        "page": 1,
        "pageSize": 1,
        "total": 1,
        "items": [
            {
                "id": 1,
                "gmtCreate": 1669048376000,
                "gmtModified": 1669048381000,
                "appId": "testapp",
                "category": "DEPLOY",
                "gmtStart": null,
                "gmtEnd": null,
                "workflowStatus": "SUSPEND",
                "workflowErrorMessage": null,
                "workflowConfiguration": null,
                "workflowSha256": "a71a1dcd0ac86e291f4442fe7477316f346e1275f91b1cd087b8e05b28fc9890",
                "workflowOptions": {},
                "workflowCreator": "admin",
                "workflowContext": null,
                "lockVersion": 3
            }
        ],
        "empty": false
    }
}

查询单个 Workflow Instance API

GET /workflow/{instanceId}
注意查询单个 WorkflowInstance API 会返回 workflowConfiguration和 workflowContext字段。
返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc13a6816690488656747929e6cf8",
    "timestamp": 1669048865710,
    "data": {
        "id": 1,
        "gmtCreate": 1669048376000,
        "gmtModified": 1669048381000,
        "appId": "testapp",
        "category": "DEPLOY",
        "gmtStart": null,
        "gmtEnd": null,
        "workflowStatus": "SUSPEND",
        "workflowErrorMessage": null,
        "workflowConfiguration": "your workflow yaml",
        "workflowOptions": {
            "creator": "admin",
            "category": "DEPLOY"
        },
        "workflowCreator": "admin",
        "workflowContext": {
            "_internal_overwrite_parameters": {
                "Global": {
                    "KUBECONFIG": "xxxx"
                }
            }
        },
        "lockVersion": 3
    }
}

设置 Workflow Instance Context

PUT /workflow/{instanceId}/context
Request Body (JSON)：

{
    "context": {
        "continueRollout": true,
        "p": false
    }
}

返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc1eed416690490952012305e2fb9",
    "timestamp": 1669049095228,
    "data": {}
}

唤醒 Workflow Instance

PUT /workflow/{instanceId}/resume
返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc1eed316690491570732464e6293",
    "timestamp": 1669049157125,
    "data": {
        "command": "RESUME",
        "workflowInstanceId": 1,
        "clientHost": "10.219.*.*:7001",
        "success": true,
        "message": ""
    }
}

上面的返回说明当前已向集群下发命令 RESUME 指定 workflowInstanceId，接收者为 10.219.113.192:7001 实例，已开始继续运行。

终止 Workflow Instance

PUT /workflow/{instanceId}/terminate
返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc1eed316690493056674000e6294",
    "timestamp": 1669049305709,
    "data": {
        "command": "TERMINATE",
        "workflowInstanceId": 1,
        "clientHost": "10.219.*.*:7001",
        "success": true,
        "message": ""
    }
}

含义同 Resume。注意如果终止一个已经到达终态的 Workflow Instance 是不会产生任何效果的。

重试 Workflow Instance

PUT /workflow/{instanceId}/retry
重试当前已经到达终态的 Workflow 实例 (SUCCESS/FAILURE/EXCEPTION/TERMINATED)，注意该方法将会从第一个节点开始，使用原始参数重新运行一遍当前 Workflow 实例。
返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc13a5716690494195578320e6ed6",
    "timestamp": 1669049419592,
    "data": {
        "command": "RETRY",
        "workflowInstanceId": 1,
        "clientHost": "10.219.*.*:7001",
        "success": true,
        "message": ""
    }
}

重试 Workflow Instance (自定义起始 Task)

PUT /workflow/{instanceId}/retryFromTask
重试当前已经到达终态的 Workflow 实例 (SUCCESS/FAILURE/EXCEPTION/TERMINATED)，注意该方法从指定 taskId 开始进行重试，即重新运行 taskId 及之后的所有 WorkflowInstance 任务，该方法会获取 taskId 对应的快照内容，以此为输入进行重试。
URL Parameters:

taskId: Workflow Task ID

返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc1eed216690495799026690e34f4",
    "timestamp": 1669049579909,
    "data": null
}

查询 Workflow Task 列表

GET /workflow/{instanceId}/tasks
URL Parameters:

taskId: Workflow Task ID
instanceId: Workflow Instance ID
appId: 应用 ID
taskType: 任务类型 (即 ApplicationConfiguration 中 workflow type)
taskStatus: 任务状态 (可选 PENDING/RUNNING/WAITING/RUNNING_SUSPEND/WAITING_SUSPEND/SUCCESS/FAILURE/EXCEPTION/TERMINATED)
deployAppId: 通过 Workflow Task 发起的二次应用部署单进行反向查询
deployWorkflowInstanceId: 通过 Workflow Task 发起的二次 Workflow 实例 ID 进行反向查询

返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc13a5716690498584322587e6ed7",
    "timestamp": 1669049858444,
    "data": {
        "page": 1,
        "pageSize": 4,
        "total": 4,
        "items": [
            {
                "id": 1,
                "gmtCreate": 1669048376000,
                "gmtModified": 1669048381000,
                "batchId": 1,
                "workflowInstanceId": 1,
                "appId": "testapp",
                "gmtStart": null,
                "gmtEnd": null,
                "taskType": "apply-components",
                "taskStage": "pre-render",
                "taskStatus": "SUCCESS",
                "clientHostname": null,
                "deployAppId": 25446,
                "deployAppUnitId": "internal",
                "deployAppNamespaceId": "",
                "deployAppStageId": "",
                "lockVersion": 4,
                "taskProperties": {},
                "taskErrorMessage": null
            },
            {
                "id": 2,
                "gmtCreate": 1669048376000,
                "gmtModified": 1669049157000,
                "batchId": 2,
                "workflowInstanceId": 1,
                "appId": "testapp",
                "gmtStart": null,
                "gmtEnd": null,
                "taskType": "suspend",
                "taskStage": "pre-render",
                "taskStatus": "SUCCESS",
                "clientHostname": null,
                "deployAppId": 0,
                "deployAppUnitId": "internal",
                "deployAppNamespaceId": "",
                "deployAppStageId": "",
                "lockVersion": 4,
                "taskProperties": {},
                "taskErrorMessage": null
            },
            {
                "id": 3,
                "gmtCreate": 1669048376000,
                "gmtModified": 1669049157000,
                "batchId": 3,
                "workflowInstanceId": 1,
                "appId": "testapp",
                "gmtStart": null,
                "gmtEnd": null,
                "taskType": "rollout-promotion",
                "taskStage": "pre-render",
                "taskStatus": "SUCCESS",
                "clientHostname": null,
                "deployAppId": 0,
                "deployAppUnitId": "internal",
                "deployAppNamespaceId": "",
                "deployAppStageId": "",
                "lockVersion": 3,
                "taskProperties": {},
                "taskErrorMessage": null
            },
            {
                "id": 4,
                "gmtCreate": 1669048376000,
                "gmtModified": 1669049160000,
                "batchId": 4,
                "workflowInstanceId": 1,
                "appId": "testapp",
                "gmtStart": null,
                "gmtEnd": null,
                "taskType": "apply-components",
                "taskStage": "pre-render",
                "taskStatus": "SUCCESS",
                "clientHostname": null,
                "deployAppId": 25447,
                "deployAppUnitId": "internal",
                "deployAppNamespaceId": "",
                "deployAppStageId": "",
                "lockVersion": 4,
                "taskProperties": {},
                "taskErrorMessage": null
            }
        ],
        "empty": false
    }
}

查询单个 Workflow Task 详情

GET /workflow/{instanceId}/tasks/{taskId}
返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc1eed516690499557175596e69ac",
    "timestamp": 1669049955736,
    "data": {
        "id": 1,
        "gmtCreate": 1669048376000,
        "gmtModified": 1669048381000,
        "batchId": null,
        "workflowInstanceId": 1,
        "appId": "testapp",
        "gmtStart": null,
        "gmtEnd": null,
        "taskType": "apply-components",
        "taskStage": "pre-render",
        "taskStatus": "SUCCESS",
        "clientHostname": null,
        "deployAppId": 25446,
        "deployAppUnitId": "",
        "deployAppNamespaceId": "",
        "deployAppStageId": "",
        "lockVersion": 4,
        "taskProperties": {
            "components": [
                {
                    "name": "helm-demo-chart@0.0.3@test",
                    "type": "ABM_HELM"
                }
            ],
            "rollout": {
                "type": "partition",
                "properties": {
                    "partition": "1",
                    "name": "abmRolloutPartition"
                }
            },
            "policies": [
                "topology-e2e-env",
                "override-e2e-env"
            ]
        },
        "taskErrorMessage": ""
    }
}

终止 Workflow Task

PUT /workflow/{instanceId}/tasks/{taskId}/terminate
RequestBody (JSON)

{
    "extMessage": "附加信息，为什么终止的原因，可选"
}

返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "0bc13a5b16690500991053780e40d8",
    "timestamp": 1669050099118,
    "data": {}
}

注意如果终止一个已经到达终态的 Workflow Task 是不会产生任何效果的。

暂停 Workflow Task

PUT /workflow/{instanceId}/tasks/{taskId}/suspend
暂停指定 Workflow 任务 (RUNNING -> RUNNING_SUSPEND / WAITING -> WAITING_SUSPEND)，其他状态下没有实际意义。
RequestBody (JSON)

{
    "extMessage": "附加信息，为什么暂停，可选"
}

返回示例：

{
    "code": 200,
    "success": true,
    "message": "SUCCESS",
    "requestId": "1bc13a5b16690500991053780e41d8",
    "timestamp": 1669050199118,
    "data": {}
}

注意如果暂停一个已经到达终态的 Workflow Task 是不会产生任何效果的。

工作流案例（灰度发布）

工作流接口

发起 Workflow API​

查询 Workflow Instance 列表 API​

查询单个 Workflow Instance API​

设置 Workflow Instance Context​

唤醒 Workflow Instance​

终止 Workflow Instance​

重试 Workflow Instance​

重试 Workflow Instance (自定义起始 Task)​

查询 Workflow Task 列表​

查询单个 Workflow Task 详情​

终止 Workflow Task​

暂停 Workflow Task​

发起 Workflow API

查询 Workflow Instance 列表 API

查询单个 Workflow Instance API

设置 Workflow Instance Context

唤醒 Workflow Instance

终止 Workflow Instance

重试 Workflow Instance

重试 Workflow Instance (自定义起始 Task)

查询 Workflow Task 列表

查询单个 Workflow Task 详情

终止 Workflow Task

暂停 Workflow Task