从触发器管道传递 CICD 变量 - Gitlab

Question

从一项作业触发管道时如何传递 CICD 变量？

我有一项工作需要触发另一项工作。 触发管道的作业是这样的

deploy.with-modelmesh-crd-upgrades:
  image: my-image
  environment: $GCP_PROJECT
  stage: deploy
  when: manual
  variables:
    ZONE: europe-west1-b
  script:
    - !reference [ .deploy_auth, script ]
    - gcloud container clusters get-credentials ewx-k8s --zone $ZONE --project ${CI_ENVIRONMENT_NAME}
 
    # Triggering deploy.uninstall-ml-serving-helm-release-and-deploy
    - |
      response_body=$(mktemp)  # Create a temporary file and store its name in $response_body
      response_code=$(curl --request POST \
                           --fail \
                           --form token=$TRIGGER_TOKEN \
                           --form ref=$CI_COMMIT_REF_NAME \
                           --form "variables[TRIGGERED_FROM]=deploy.with-modelmesh-crd-upgrades" \
                           --write-out "%{http_code}" \
                           --verbose \
                           --output "$response_body" \
                           "https://gitlab.com/api/v4/projects/47198141/trigger/pipeline")
      echo "Checking the response:"
      cat $response_body
      echo $TRIGGERED_FROM
      if [[ "$response_code" -ne 201 ]]; then
        cat $response_body
        echo "Failed to trigger deploy.uninstall-ml-serving-helm-release-and-deploy job with response code: $response_code."
        rm $response_body
        exit 1
      fi
      rm $response_body

如果我们基本上检查我的curl命令是

curl --request POST \
    --fail \
    --form token=$TRIGGER_TOKEN \
    --form ref=$CI_COMMIT_REF_NAME \
    --form "variables[TRIGGERED_FROM]=deploy.with-modelmesh-crd-upgrades" \
    --write-out "%{http_code}" \
    --verbose \
    --output "$response_body" \
    "https://gitlab.com/api/v4/projects/project-id/trigger/pipeline"

这就是我想要从上面触发的工作。我有以下配置：

deploy.uninstall-ml-serving-helm-release-and-deploy:
  # only for modelmesh module
  image:
    name: hashicorp/terraform:1.5.5
    entrypoint: [ "" ]
  environment: $GCP_PROJECT
  stage: deploy
  script:
    - echo "TRIGGERED_FROM = $TRIGGERED_FROM"
    - env
    - export GOOGLE_CREDENTIALS=$(base64 -d ${GCP_SA_KEY_B64})
    - cd terraform
    - terraform init
    - terraform workspace select ${CI_ENVIRONMENT_NAME} || terraform workspace new ${CI_ENVIRONMENT_NAME}
    # Destroying resources
    - |
      destroy_cmd="terraform destroy -input=false -auto-approve --var-file=env/${CI_ENVIRONMENT_NAME}.tfvars --var=ml_deployer_image_tag=${CI_COMMIT_REF_SLUG}-${CI_COMMIT_SHA::7} --var=ray_image_tag=${CI_COMMIT_REF_SLUG}-${CI_COMMIT_SHA::7}"
      for target in \
        "module.modelmesh[0].helm_release.modelmesh" \
        "google_compute_disk.modelmesh-disc" \
      do
        destroy_cmd="$destroy_cmd -target=\"$target\""
      done
      eval $destroy_cmd
    # Apply changes
    - terraform apply -input=false -auto-approve --var-file=env/${CI_ENVIRONMENT_NAME}.tfvars --var=ml_deployer_image_tag=${CI_COMMIT_REF_SLUG}-${CI_COMMIT_SHA::7} --var=ray_image_tag=${CI_COMMIT_REF_SLUG}-${CI_COMMIT_SHA::7}
  rules:
    - if: '$TRIGGERED_FROM == "deploy.with-modelmesh-crd-upgrades"'
      when: always
    - when: manual

请注意，这里的想法是传递正在设置的

TRIGGERED_FROM

变量

deploy.with-modelmesh-crd-upgrades

作业，在curl命令运行时，并在

deploy.uninstall-ml-serving-helm-release-and-deploy

作业上请求该变量，因为我有这个定义的规则：

rules:
    - if: '$TRIGGERED_FROM == "deploy.with-modelmesh-crd-upgrades"'
      when: always
    - when: manual

问题是，当我在curl命令中声明

TRIGGERED_FROM

环境变量时，它没有被设置，在

deploy.with-modelmesh-crd-upgrades

作业中，执行该命令时似乎存在问题。 我还尝试简化curl命令来做到这一点：

- 'curl --fail --request POST --form token=$TRIGGER_TOKEN --form ref=$CI_COMMIT_REF_NAME --form "variables[TRIGGERED_FROM]=deploy.with-modelmesh-crd-upgrades" "https://gitlab.com/api/v4/projects/47198141/trigger/pipeline"'

但不幸的是，

TRIGGERED_FROM

环境变量根本没有被设置。

检查文档，看来我的两种curl命令方法似乎与这里所说的方式一致：https://docs.gitlab.com/ee/ci/triggers/index.html#pass-cicd-variables -在 API 调用中

可能出了什么问题？为什么

TRIGGERED_FROM

环境变量根本没有被设置。？

Answer 1

我遇到了同样的问题，解决方案是在curl命令中=号后面的变量周围加上引号，如下所示：

--from variables[TRIGGERED_FROM]="deploy.with-modelmesh-crd-upgrades"

很烦人，但这是使其发挥作用的唯一方法。