Cluster API: worker process can't get 'SIGTERM' signal when being killed

[baoshan]

cluster = require "cluster"

if cluster.isMaster

  for i in [0...4]
    cluster.fork()

  setTimeout ->
    worker.kill() for id, worker of cluster.workers
  , 2000

else
  process.on 'SIGTERM', ->
    console.log 'SIGTERM'
  setTimeout (->), 100000

Outputs nothing. I think it should log SIGTERM 4 times.

What's wrong with the above code?

[baoshan]

I noticed line 420 @ lib/cluster.js

process.once('disconnect', process.exit.bind(null, 0));

Is that line the reason for above behavior?

If worker exits immediately once disconnected, what's the intended use for signal of api worker.kill(sign)?

[trevnorris]

This belongs on the mailing list. Once an actual bug has been identified open a new issue with said information. Also, post example code in javascript.

[baoshan]

Got it!

[trevnorris]

Thanks. :-)

[hunterloftis]

@baoshan did you ever figure this out? I'm seeing the exact same thing (and would expect the exact same output as you did).

[hunterloftis]

Aha: #5832