Jenkins Pipeline:闭包不应该和外部平行地解决变量吗？

Question

我在尝试在Jenkins管道脚本中执行代码时遇到了一些令人困惑的行为.闭包内的值带有一些意外的值.我已经看到了对局部变量技巧的引用,据说可以解决这个问题,但它对我不起作用.

简化示例:创建3个作业'a','b','c'打印其参数 - 传入和复制到本地.首先并行执行作业; 然后执行外部并行以比较结果.定义因为println在我的Jenkins中提供访问异常(我不是管理员).

def say(s) {println(s)} // println gives exception inside create_jobs

def create_jobs() {
  def map = [:]                // needed for parallel
  def jobrunfn = { jobid ->    // return a closure that prints hello from job
    def pid = "$jobid"
    return {
      def xsay = { s -> say("[$pid] $s") }
      xsay "HELLO from $pid"
    }
  }

  map['a'] = jobrunfn('a')
  map['b'] = jobrunfn('b')
  map['c'] = jobrunfn('c')
  return map
}

def jobs = create_jobs()
parallel(jobs)              // for Jenkins pipeline only - not groovy interp
for (j in jobs) { jobs[j.key]() }   // groovy interp - parallel not available

输出 - 内部并行,'pid'有时是'c',当它应该是'a'或'b'时:

[Pipeline] parallel
[Pipeline] [a] { (Branch: a)
[Pipeline] [b] { (Branch: b)
[Pipeline] [c] { (Branch: c)
[Pipeline] [a] echo
[a] [a] HELLO from c
[Pipeline] [a] }
[Pipeline] [b] echo
[b] [b] HELLO from c
[Pipeline] [b] }
[Pipeline] [c] echo
[c] [c] HELLO from c
[Pipeline] [c] }
[Pipeline] // parallel

[Pipeline] echo
[a] HELLO from a
[Pipeline] echo
[b] HELLO from b
[Pipeline] echo
[c] HELLO from c
[Pipeline] End of Pipeline
Finished: SUCCESS

参数总是传入的最后一个值(不确定原因,但它是一致的,我知道你需要局部变量技巧.)

但是,局部变量技巧在并行中并不一致:即使本地有时也是'c'(对于嵌套'说'它是正确的,但是主体错了); 在平行之外它按预期得到'a'和'b'.

我误会了吗？

Linux上的Jenkins 2.32.2,Pipeline:Grooby 2.27

Answer 1

这是一个管道的错误.看票.