AIX: parallel/test-cluster-dgram-1 failures

[mhdawson]

• Version: master
• Platform: AIX
• Subsystem: cluster
  This test seems to have started failing recently:

not ok 112 parallel/test-cluster-dgram-1
# Mismatched <anonymous> function calls. Expected 1, actual 0.
# at setupWorker (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:69:36)
# at master (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:59:7)
# at Object.<anonymous> (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:18:3)
# at Module._compile (module.js:556:32)
# at Object.Module._extensions..js (module.js:565:10)
# at Module.load (module.js:473:32)
# at tryModuleLoad (module.js:432:12)
# at Function.Module._load (module.js:424:3)
# at Module.runMain (module.js:590:10)
# Mismatched <anonymous> function calls. Expected 1, actual 0.
# at setupWorker (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:69:36)
# at master (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:59:7)
# at Object.<anonymous> (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:18:3)
# at Module._compile (module.js:556:32)
# at Object.Module._extensions..js (module.js:565:10)
# at Module.load (module.js:473:32)
# at tryModuleLoad (module.js:432:12)
# at Function.Module._load (module.js:424:3)
# at Module.runMain (module.js:590:10)
# Mismatched <anonymous> function calls. Expected 1, actual 0.
# at setupWorker (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:69:36)
# at master (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:59:7)
# at Object.<anonymous> (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:18:3)
# at Module._compile (module.js:556:32)
# at Object.Module._extensions..js (module.js:565:10)
# at Module.load (module.js:473:32)
# at tryModuleLoad (module.js:432:12)
# at Function.Module._load (module.js:424:3)
# at Module.runMain (module.js:590:10)
# Mismatched <anonymous> function calls. Expected 1, actual 0.
# at setupWorker (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:69:36)
# at master (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:59:7)
# at Object.<anonymous> (/home/iojs/build/workspace/node-test-commit-aix/nodes/aix61-ppc64/test/parallel/test-cluster-dgram-1.js:18:3)
# at Module._compile (module.js:556:32)
# at Object.Module._extensions..js (module.js:565:10)
# at Module.load (module.js:473:32)
# at tryModuleLoad (module.js:432:12)
# at Function.Module._load (module.js:424:3)
# at Module.runMain (module.js:590:10)
 ---
 duration_ms: 0.833
 ...

[mhdawson]

May be related to this refactoring: #8261 which landed recently.

[mhdawson]

@jasnell. Opened issue to cover and has info printed from last failure.

[mhdawson]

Can we either revert the refactor of this specific test case until we can figure out the right way to go ?

[santigimeno]

#8383 fixes the issue for me in FreeBSD.

[jasnell]

@mhdawson ... can you test if #8383 fixes the issue on AIX. If it does not, then reverting the change for this one test is just fine.

[mhdawson]

Ok starting tests now.

[mhdawson]

Tested results are in #8383

Net is that is helps but I still see failure 3/150 versus 0 with the old version.

[jasnell]

Ok, let's revert this one then. I'll get the PR together for that shortly.

[mhdawson]

PR to revert that one test if we decide its the right way to go: https://github.com/nodejs/node/pull/8384/files

[mhdawson]

Ok updated fix in #8383 resolves the issues. :)

[jasnell]

It looks like the failure here may have been happening before the refactor but just hadn't been caught. Adding the common.mustCall() added the additional constraint that highlighted the flaky failure. I'd say that's a win ;-) Thanks for looking into it @mhdawson