问题
One day, out of nothing, my app decided not to die.
After pressing ctrl + c when I inspect port it's still there and not only on my machine but also on server that is managed by PM2 so I need to go there each time and kill the process by hand. I tried to look for issues in my code and when I couldn't I just though that it's simply one of the dependencies fault and soon there will be a fix. It was slightly annoying but I could kill the process with kill -9 PID and I had most of the tasks I had on front-end side so it wasn't that big issue. Today, more than a week later problem it's still here. I went back in history, picked a commit that I made weeks ago, where everything worked perfectly fine, switched NodeJS from 5.1.0 to 4.2.1, cleared npm cache, reinstalled all dependencies and I still see the problem. I'm using LoopbackJS, but normally I start the app simply with "node server/server.js" and then the problem described above happens, but if I use "slc run" and then try to kill the app with ctrl + c it just hangs forever, I mean I can press ctrl+c as many times as I want and it's still running in console foreground.
If instead of pressing ctrl+c I kill the tab in console, app dies without problems.
This is what I see after running "lsof -i tcp:4000" when the app is supposed to be dead but is not
Edit: Running and killing it with Strongloop process manager - slc start/slc stop works fine but it would be more convenient to use normal way of running NodeJS app (node server.js) during development and it doesn't change the fact that there is some issue and it would be better to not hide it under the carpet.
回答1:
Out of anger I stripped out my app to the most basic elements. File by file, piece by piece and I found the answer. The culprit was phantomJS - https://github.com/sgentle/phantomjs-node. PATCH version 0.8.2 introduced this bug. Fix was created almost a month ago, merged a week ago but not published on npm yet.
来源：https://stackoverflow.com/questions/33879801/cant-kill-nodejs-app