Now that node 4.2 is released, we should make sure that service-runner is also tested against it. There are some small behavior changes in relevant interfaces. I'm not sure if those are actually covered by tests yet. If they aren't, then it might be worth adding tests for those.
Description
Event Timeline
This change was needed (had any effect) only when the worker got into an infinite cpu-bound loop, and killed externally by SIGKILL before master detects worker's dead. In that case, master crashed too. The corresponding node.js bug is here. Although it's not marked as fixed, I can't reproduce the problem any more on 0.10.40, 0.12.7 or 4.2.0, but people on the ticket claim it's still an issue for them.
Creating an automatic test for the problem is very complicated, as (at least for me) an original problem is not reproducible any more. So, I propose to wait until it's fixed upstream (there was a PR for that), and then remove the workaround.
As for now, I've tested the code manually on 4.2.0, and it works fine, so I think we can close this ticket. @GWicke, @mobrovac, what do you think?
We're running 4.2 in production, so this ticket doesn't seem to be relevant any more. Closing.