On Tue, 14 Mar 2000, Alex Belits wrote:
> On Tue, 14 Mar 2000, Jason Gunthorpe wrote:
>
> > > Killing Oracle, or any other server that depends on some process being
> > > alive and keeping a valuable, complex, hard to recover data on disk and
> > > in memory, is in some cases not any better than just blowing up the box.
> >
> > It might not be better, but it isn't worse. At least your filesystem is
> > nice and consistent and you can telnet in and recover.
> >
> > I think you have to accept that it is impossible to always kill the 'right
> > process' because that definition is too vauge. Killing the biggest process
> > however is an iterative thing, eventually the thing actually sucking up
> > ram will become the biggest process and be killed [barring any
> > maliciousness].
>
> If those were the only choices, it would be reasonable, but why some
> userspace program can't decide, what should be the policy, and tell the
> kernel? It can be done the same way, priority is handled.
You can indeed do this. The kernel's OOM handler is simply a last gasp
ultimate defence. Ideally, many other measures should have kicked in long
before this point - per-process rlimits, per-user rlimits, the system-wide
resource protection daemon you suggest.
We have the first, we will have the second one day, and someone needs to
write the third :-)
James.
-
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 : Thu Mar 23 2000 - 21:00:17 EST