node has this great option:
--trace (trace function calls)
which gives you tons of output of exactly how your program is running.
when you have an error that causes node to crash without any stacktrace, this option will show you the stacktrace before it dies.
it worked great for me when I had an error that happened right after my program started. I got about 5 mins of output and then the error.
but now I have an error that doesn't show unless I stress test my app making the --trace option generate way way way too much output. Even if I pipe the output to a file or to grep it doesn't help. Node just takes way to long to reach my error because of all the IO it's doing. It would literally take days (months?) to reach my error from all the IO.
Is there a way to limit --trace output and just get the stacktrace at the end when it crashes?