A pipeline like this will never actually retry - it'll just run once and fail.

pipeline {
  agent any
  options {
    retry (3)
  }
  stages {
    stage("foo") {
      steps {
        echo "hello"
        error "Failing - retry me!"
      }
    }
  }
}

This is because the error thrown in the stage is caught and not rethrown again until the end of the build, bypassing the retry step's execution completely.