Argo 是一个基于 Kubernetes 的工作流引擎，同时也支持 CI、CD 等丰富的功能。Argo 开源在 https://github.com/argoproj。

安装 Argo

使用 argo install

  1. # Download Argo.
  2. curl -sSL -o argo https://github.com/argoproj/argo/releases/download/v2.1.0/argo-linux-amd64
  3. chmod +x argo
  4. sudo mv argo /usr/local/bin/argo
  6. # Deploy to kubernetes
  7. kubectl create namespace argo
  8. argo install -n argo
  1. ACCESS_KEY=AKIAIOSFODNN7EXAMPLE
  2. ACCESS_SECRET_KEY=wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY
  4. helm install --namespace argo --name argo-artifacts --set accessKey=$ACCESS_KEY,secretKey=$ACCESS_SECRET_KEY,service.type=LoadBalancer stable/minio

创建名为 argo-bucket 的 Bucket（可以通过 kubectl port-forward service/argo-artifacts-minio :9000 访问 Minio UI 来操作）：

  1. # download mc client
  2. sudo wget https://dl.minio.io/client/mc/release/linux-amd64/mc -O /usr/local/bin/mc
  3. sudo chmod +x /usr/local/bin/mc
  5. # create argo-bucket
  6. EXTERNAL_IP=$(kubectl -n argo get service argo-artifacts-minio -o jsonpath='{.status.loadBalancer.ingress[0].ip}')
  7. mc config host add argo-artifacts-minio-local http://$EXTERNAL_IP:9000 $ACCESS_KEY $ACCESS_SECRET_KEY --api=s3v4
  8. mc mb argo-artifacts-minio-local/argo-bucket

然后修改 Argo 工作流控制器使用 Minio：

  1. $ kubectl -n argo create secret generic argo-artifacts-minio --from-literal=accesskey=$ACCESS_KEY --from-literal=secretkey=$ACCESS_SECRET_KEY
  2. $ kubectl edit configmap workflow-controller-configmap -n argo
  3. ...
  4.     executorImage: argoproj/argoexec:v2.0.0
  5.     artifactRepository:
  6.       s3:
  7.         bucket: argo-bucket
  8.         endpoint: argo-artifacts-minio.argo:9000
  9.         insecure: true
  10.         # accessKeySecret and secretKeySecret are secret selectors.
  11.         # It references the k8s secret named 'argo-artifacts-minio'
  12.         # which was created during the minio helm install. The keys,
  13.         # 'accesskey' and 'secretkey', inside that secret are where the
  14.         # actual minio credentials are stored.
  15.         accessKeySecret:
  16.           name: argo-artifacts-minio
  17.           key: accesskey
  18.         secretKeySecret:
  19.           name: argo-artifacts-minio
  20.           key: secretkey

使用 Helm

注意：当前 Helm Charts 使用的 Minio 版本较老，部署有可能会失败。

  1. # Download Argo.
  2. curl -sSL -o /usr/local/bin/argo https://github.com/argoproj/argo/releases/download/v2.0.0/argo-linux-amd64
  3. chmod +x /usr/local/bin/argo
  5. # Deploy to kubernetes
  6. helm repo add argo https://argoproj.github.io/argo-helm/
  7. kubectl create clusterrolebinding default-admin --clusterrole=cluster-admin --serviceaccount=kube-system:default
  8. helm install argo/argo-ci --name argo-ci --namespace=kube-system

访问 Argo UI

  1. $ kubectl -n argo port-forward service/argo-ui :80
  2. Forwarding from 127.0.0.1:52592 -> 8001
  3. Forwarding from [::1]:52592 -> 8001
  5. # 使用浏览器打开 127.0.0.1:52592

工作流

首先，给默认的 ServiceAccount 授予集群管理权限

  1. # Authz yourself if you are not admin.
  2. kubectl create clusterrolebinding default-admin --clusterrole=cluster-admin --serviceaccount=argo:default

示例1： 最简单的工作流

  1. apiVersion: argoproj.io/v1alpha1
  2. kind: Workflow
  3. metadata:
  4.   generateName: hello-world-
  5. spec:
  6.   entrypoint: whalesay
  7.   templates:
  8.   - name: whalesay
  9.     container:
  10.       image: docker/whalesay:latest
  11.       command: [cowsay]
  12.       args: ["hello world"]
  1. argo -n argo submit https://raw.githubusercontent.com/argoproj/argo/master/examples/hello-world.yaml

示例2：包含多个容器的工作流

  1. # This example demonstrates the ability to pass artifacts
  2. # from one step to the next.
  3. apiVersion: argoproj.io/v1alpha1
  4. kind: Workflow
  5. metadata:
  6.   generateName: artifact-passing-
  7. spec:
  8.   entrypoint: artifact-example
  9.   templates:
  10.   - name: artifact-example
  11.     steps:
  12.     - - name: generate-artifact
  13.         template: whalesay
  14.     - - name: consume-artifact
  15.         template: print-message
  16.         arguments:
  17.           artifacts:
  18.           - name: message
  19.             from: "{{steps.generate-artifact.outputs.artifacts.hello-art}}"
  21.   - name: whalesay
  22.     container:
  23.       image: docker/whalesay:latest
  24.       command: [sh, -c]
  25.       args: ["cowsay hello world | tee /tmp/hello_world.txt"]
  26.     outputs:
  27.       artifacts:
  28.       - name: hello-art
  29.         path: /tmp/hello_world.txt
  31.   - name: print-message
  32.     inputs:
  33.       artifacts:
  34.       - name: message
  35.         path: /tmp/message
  36.     container:
  37.       image: alpine:latest
  38.       command: [sh, -c]
  39.       args: ["cat /tmp/message"]
  1. argo -n argo submit https://raw.githubusercontent.com/argoproj/argo/master/examples/artifact-passing.yaml

工作流创建完成后，可以查询它们的状态和日志，并在不需要时删除：

  1. $ argo list
  2. NAME                     STATUS    AGE   DURATION
  3. artifact-passing-65p6g   Running   6s    4s
  4. hello-world-cdnpq        Running   8s    6s
  6. $ argo -n argo logs hello-world-4dhg8
  7.  _____________
  8. < hello world >
  9.  -------------
  10.     \
  11.      \
  12.       \
  13.                     ##        .
  14.               ## ## ##       ==
  15.            ## ## ## ##      ===
  16.        /""""""""""""""""___/ ===
  17.   ~~~ {~~ ~~~~ ~~~ ~~~~ ~~ ~ /  ===- ~~~
  18.        \______ o          __/
  19.         \    \        __/
  20.           \____\______/
  22. $ argo -n argo delete hello-world-4dhg8
  23. Workflow 'hello-world-4dhg8' deleted

更多工作流 YAML 的格式见官方文档和工作流示例。