Managed to successfully run integration tests in parallel on the same node and I now I would like to distribute them across different nodes. In the example below I want stages IT 1 and IT 2 to run on different nodes leaving IT 3 to run on the original node. Tried several combinations by using node as a parent of stage('IT 1') and node as a child but I get syntax errors for both. What is the proper syntax to achieve this?
pipeline {
  agent { label '!master' }
  stages {
    stage('Integration Tests') {
      parallel {
        stage('IT 1 (slow)') {
          steps {
            sh 'run-it-1.sh'
          }
        }
        stage('IT 2 (slow)') {
          steps {
            sh 'run-it-2.sh'
          }
        }
        stage('IT 3 (quick)') {
          steps {
            sh 'run-it-3.sh'
          }
        }
      }
    }
  }
}
Edit: Using label instead of node works for declarative pipelines. Example below:
stage('IT 1 (slow)') {
  agent { label '!master' }
  steps {
    sh 'run-it-1.sh'
  }
}