Implement a request timeout.

Add a feature to abort a request if it doesn't complete within a certain period.

This is hard to do. The first problem is that the process can be multithreaded 
and so one can't just 
kill the process. The second problem is that Python doesn't allow you to kill 
threads.

The only way to achieve this is for a separate monitor thread to monitor how 
long requests are 
taking and if a request takes longer than the timeout, to inject a Python 
exception into the context 
of the thread for the request handler which is taking to long.

Although one can do this, it doesn't interrupt the thread. If that thread is 
blocked in C code it will 
remain blocked and the exception will only be noted once it returns back into 
Python code.

Further, because it is a Python exception, even though one might use the 
SystemExit exception, an 
application still technically can catch it and possibly ignore it. So, it may 
be necessary for that 
monitor thread to keep injecting the signal into the thread on a regular basis 
until the request code 
finally allows it to propagate up to top most caller. If it doesn't allow this 
after certain time, then 
process might be killed off, although can only do this if using daemon mode.

Original issue reported on code.google.com by Graham.Dumpleton@gmail.com on 4 Mar 2010 at 11:14

ramkrishanbhatt / modwsgi

Implement a request timeout. #183