lkml.org 
[lkml]   [1997]   [May]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Out of memory kernel deat
On Mon, 12 May 1997, Jon Tombs wrote:

-> My previous reply doesn't seem to have made it to the list.
->
-> I proposed the following algorithm for oem situation.
->
-> Igonoring process 1, give "death points" to each process, for the following
-> attributes.
->
-> X points for every page rss
-> -Y points for every second of execution time
-> Z points for not having uid=0
-> Z/2 points for not having euid=0

Why Z/2? Why not a configurable factor (i.e. have 4 weights instead of three)?

-> The process with the biggest score gets killed.
->
-> The important processes, tend to be:
-> owned by root
-> old (started at boot).
Yes, but old doesn't mean a lot of execution time. For example my named,
syslogd and klogd use very little CPU, they would be bait. The starttime
of the process is IMHO a much better criterium.


-> There might me other factors to include, but they don't occur to me at the
-> moment.

Try disk I/O. A lot of system processes only use disk I/O at startup, so
little disk I/O might be a good indicator.

You could also think of a process being registered as vital through /proc in
some way (by root only of course) and applying the criteria to all other
processes.

-> I think the above is a solution in the spirit of linux. If we use a fixed
-> set of rules to kill a process, somebody will always be able to produce a
-> situation that kills an important process. Let them change the weighting.

I agree.

--------------------------------------
Hugo Van den Berg - hbe@cypres.nl
Phone - +31 (0)30 - 60 25 400
Fax - +31 (0)30 - 60 50 799
--------------------------------------



\
 
 \ /
  Last update: 2005-03-22 13:39    [W:0.044 / U:1.132 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site