Re: OOM in 2.2.14

From: Warren Young (tangent@cyberport.com)
Date: Thu Jul 13 2000 - 03:09:03 EST


jmcmullan@linuxcare.com wrote:
>
> Just to be a frick'n idiot, why not just have the kernel
> issue messages to the kernel log when swap space becomes low,
> (configurable via /proc), and have a user-land daemon monitor
> for that, and Do The Right Thing(TM)?

The problem isn't whether it's good or bad to have the kernel do
something about OOM conditions, but what "the Right Thing" is.

If your solution involves killing the "guilty" program, that's awfully
hard to define in a generic way. A human admin can figure it out, but
it's a much harder task to teach the computer to make that same
judgment. Since we want the computer to never make mistakes, (i.e.
killing something important) our job is doubly hard.

I suspect that once we get a good algorithm for determining what the
"guilty" process is, it'll become a matter of mere implementation
details. Implementation details like "kernel or userland", for
example. :)

-- 
= Warren -- ICBM Address: 36.8274040 N, 108.0204086 W, alt. 1714m

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sat Jul 15 2000 - 21:00:16 EST