Program Killers, beancounters, etcetera [Regarding linux memory DOS
Alex K. (kolex@erols.com)
Thu, 02 Sep 1999 19:59:13 -0400
Hello,
I don't mean to butt in, but I must interject on the kernel-based
program killer changes that people are talking about. Currently, the
program killer that Linux has is simple and brutal. Although it's not
very accurate, it gets the job done. I am not saying that there is no
better solution, I just think that a userspace tool could perform the
more complicated emergency memory management algorithms. A program could
poll /proc every few milliseconds (overhead would be minimal) and if it
detects a problem (say memory usage going above a certain percent) it
would perform whatever it is designed to do (I don't want to start
rehashing all the ideas everyone has had, so I'll just say the
advantages of implementing it in user-space:
1: Kernel will be simpler and a bit smaller - a good thing, though the
size difference will be tiny.
2: People who don't know how to program Linux kernel or are afraid to
(like me) mess up their only computer could still make a userspace
program.
3: A running program could be replaced with a different one on the fly
without having to reboot or recompile the kernel.
4: The kernel memory/process management should be, to my understanding,
as universal as possible, while a particular management program could be
made to avoid killing X, etc..., which is a detail that the kernel
shouldn't have to deal with.
The idea of using memory quotas is interesting, but personally, I think
it would be too much of an administrative hassle. Plus, a lot of things
that users can start programs that su to root, like X. X runs as root,
so the memory it allocates would be unlimited - but a user could still
select startx options that would make X consume a vast amount of memory,
unhindered.
-
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/