无法在Jenkins Pipeline外部Groovy脚本中使用环境变量

• env.test_USR在Pipeline内部可正常设置和使用，在gerritCheck中能正常生效，但调用自定义的slackNotify函数发送Slack消息时，变量值为null，导致SonarQube链接显示?id=null，预期应为?id=test_12345
• 该变量无法在Jenkins Pipeline代码块外部的函数中读取

错误输出 #

https://sonar.company.com/dashboard?id=null

预期输出 #

https://sonar.company.com/dashboard?id=test_12345

相关代码片段 #

notifyChannel = "#sw-test"

sonarStageLink = "SonarQube: <https://sonar.company.com/dashboard?id=${env.test_USR}>

def slackNotify(String status, String color) {
    slackSend (
        channel: notifyChannel,
        color: color,
        message: "Analysis: \n${sonarStageLink}\n"
    )
}

pipeline {
    agent {
        label 'test'
    }

    environment {
        RUN_SONAR_URL = 'https://sonar.company.com'
        test = credentials('test-pullbranchanalysis')
    post {
        success {
            gerritCheck checks: ['jenkins:code-analysis': 'SUCCESSFUL'], url: "id=${env.test_USR}"
            script {
                if (!params.Refresh) {
                    echo "success"
                    slackNotify("succeeded", "good")

原因分析 #

sonarStageLink是在Pipeline代码块外部定义的，此时env.test_USR还未被初始化，所以变量值为null。而gerritCheck是在Pipeline内部的post块中执行，能正常访问到已初始化的env变量。

修复方案 #

方式1：将链接生成逻辑移到函数内部并传参 #

把SonarQube链接的构建放到slackNotify函数里，调用时传入env.test_USR的值：

notifyChannel = "#sw-test"

def slackNotify(String status, String color, String sonarId) {
    def sonarStageLink = "SonarQube: <https://sonar.company.com/dashboard?id=${sonarId}>"
    slackSend (
        channel: notifyChannel,
        color: color,
        message: "Analysis: \n${sonarStageLink}\n"
    )
}

pipeline {
    agent {
        label 'test'
    }

    environment {
        RUN_SONAR_URL = 'https://sonar.company.com'
        test = credentials('test-pullbranchanalysis')
    }
    post {
        success {
            gerritCheck checks: ['jenkins:code-analysis': 'SUCCESSFUL'], url: "id=${env.test_USR}"
            script {
                if (!params.Refresh) {
                    echo "success"
                    slackNotify("succeeded", "good", env.test_USR)
                }
            }
        }
    }
}

方式2：在Pipeline内部动态初始化变量 #

先声明全局变量，再在Pipeline内部的合适阶段赋值，确保能访问到已初始化的env变量：

notifyChannel = "#sw-test"
def sonarStageLink // 先声明全局变量

def slackNotify(String status, String color) {
    slackSend (
        channel: notifyChannel,
        color: color,
        message: "Analysis: \n${sonarStageLink}\n"
    )
}

pipeline {
    agent {
        label 'test'
    }

    environment {
        RUN_SONAR_URL = 'https://sonar.company.com'
        test = credentials('test-pullbranchanalysis')
    }
    stages {
        stage('Init') {
            steps {
                script {
                    sonarStageLink = "SonarQube: <https://sonar.company.com/dashboard?id=${env.test_USR}>"
                }
            }
        }
        // 其他业务阶段...
    }
    post {
        success {
            gerritCheck checks: ['jenkins:code-analysis': 'SUCCESSFUL'], url: "id=${env.test_USR}"
            script {
                if (!params.Refresh) {
                    echo "success"
                    slackNotify("succeeded", "good")
                }
            }
        }
    }
}

关键说明 #

Jenkins Pipeline的变量作用域分为全局脚本作用域和Pipeline块作用域：

• 全局脚本作用域的代码在Pipeline启动时就执行，此时environment块中定义的变量还未加载
• Pipeline块（包括agent、stages、post等）内部的代码会在环境变量初始化后执行，能正常访问env对象

内容的提问来源于stack exchange，提问作者AhmFM