Re: [PATCH] oom killer (Core)

[Thomas Gleixner]

On Thu, 2004-12-02 at 10:29 -0800, Andrew Morton wrote:
> Could be, I dunno.  I've never done any work on the oom-killer and I tend
> to regard it as a stupid thing which is only there so you can get back into
> the machine to shut down and restart everything.
> 
> I mean, if you ran the machine out of memory then it is underprovisioned
> and it *will* become unreliable whatever we do.  The answer is to Not Do
> That.  As long as the oom-killer allows you to get in and admin the machine
> later on then that's good enough as far as I'm concerned.  Others probably
> disagree ;)

I agree, but the current situation made me drive 150km, because sshd or
even init was killed.

I hit this problem, when a forking server application got out of control
because there were suddenly connecting hundreds of clients due to a
different problem.

As long as I can log into the machine and fix the crap it's ok. There's
no need for anything else than accessability and a half way
deterministic behaviour.

tglx


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