TypeError: Cannot read property 'remove' of undefined at close (cluster.js:497:15)

[nullivex]

I am seeing these blowups on Travis.

I will add some portions of the build log.

Setting environment variables from .travis.yml
$ export TRAVIS=travis
$ export CXX=g++-4.8
6.56s$ nvm install 5.0
######################################################################## 100.0%
Checksums empty
Now using node v5.0.0
$ node --version
v5.0.0
$ npm --version
3.3.6
$ nvm --version
0.23.3

During the testing I am seeing the following.

Beginning prism shutdown
Beginning store shutdown
Beginning store shutdown
cluster.js:497
    if (handle.remove(worker)) delete handles[key];
              ^
TypeError: Cannot read property 'remove' of undefined
    at close (cluster.js:497:15)
    at Worker.onmessage (cluster.js:431:7)
    at ChildProcess.<anonymous> (cluster.js:713:8)
    at emitTwo (events.js:92:20)
    at ChildProcess.emit (events.js:172:7)
    at handleMessage (internal/child_process.js:686:10)
    at Pipe.channel.onread (internal/child_process.js:440:11)
  1) "before all" hook for "should still download content"

And another

cluster.js:497
    if (handle.remove(worker)) delete handles[key];
              ^
TypeError: Cannot read property 'remove' of undefined
    at close (cluster.js:497:15)
    at Worker.onmessage (cluster.js:431:7)
    at ChildProcess.<anonymous> (cluster.js:713:8)
    at emitTwo (events.js:92:20)
    at ChildProcess.emit (events.js:172:7)
    at handleMessage (internal/child_process.js:686:10)
    at Pipe.channel.onread (internal/child_process.js:440:11)
  2) "after all" hook for "should still download content"

I am not sure what the internals are of what are causing this. However, I can note that on my Windows machines with Node 5.0 I am not getting these errors.

These crashes are happening randomly for me. If I restart my tests they happen in a different place each time.

It only happens during cluster shutdown.

I am doing a build on multiple node versions to see if I can figure out where this is spawning from.

[nullivex]

Okay, I will try to add some environmental states here.

I am using the cluster module through a module called https://www.npmjs.com/package/infant which I have been using since Node 0.10 without issue.

I am unsure if there is something in the cluster handling causing this error. I googled for it and couldn't find anyone else seeing this blow up.

I dont see these errors in the unit tests for infant so I curious about what exactly is causing these strange errors.

[nullivex]

Okay nevermind.

I found the fix against this error here: 33827e3

I guess it just hasnt been released yet.

[nullivex]

Any idea when this fix will be released?

[yosuke-furukawa]

You would be better to wait for release v5.1.0.
#3736

According to the PR, we will release the version soon.

[Fishrock123]

v5.1.0 will release this week, probably tomorrow. It has been delayed due to some jenkins issues: #3709

Closing this since the fix itself has already landed.