On Mon, 6 Mar 2000, Rik van Riel wrote:
> It's quite possible (about a 3% chance) that the PID of the
> hog (that we _want_ to get killed) is in the <1000 range...
>
> Also, "simply reverse-ordering chronologically" doesn't work
> if it means killing two dozen innocent processes before we
> get to the memory hog.
>
> Killing processes when we're out of memory is _bad_ (but
> unavoidable) so we'd better make sure that we kill the hog
> and keep all other processes alive...
>
> regards,
>
> Rik
Another option to consider. Linux currently has a freepage.min below
which only the kernel can alloc memory.
However this doesn't help the admin to telnet/log on and clean up
manually. Would it be possible to add a freepage.root-min. Below which
only euid==0 processes could allocate memory? So that
getty/login/inetd/telnetd could still get memory in the case that a user
process is causing trouble.
(or this is impractical for some reason?)
regards,
-
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 : Tue Mar 07 2000 - 21:00:23 EST