lkml.org 
[lkml]   [2000]   [Mar]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: Some questions about linux kernel.
In <Pine.LNX.4.04.10003121537280.19554-100000@eax.student.umd.edu> Adam (adam@eax.com) wrote:

>> Rik's OOM patch:
>> You have a file with holes or other compression. When you try to
>> write to this, the kernel searches for junk files. It finds core
>> files, /tmp files, emacs backup files, *.o files...

> Problem is that one's thing junk is other person's valuable thing.

Partially - yes. For 1% cases or so.

> If I run some web server, the 'httpd's are the most imporant thing,
> and it is thing which I least want to get killed.

> On the other hand if I run some important simulation, I'll be willing
> to let daemons like sendmail or http die in hopes the simulation
> finish.

> Going back to files analogy the intermediate data from simulation
> could be in tmp, that case I would want to save them (of course
> one could argue that they should be in home dir, but still) even
> in espense of removing stuff elsewhere, which I can later restore
> from distribution cd.

> Bottom line is that I don't belive any kind of AI in OMM will do the
> right job.

It does. In 99% cases. If you system constantly overflows mem+swap in
normal situations then you need to add more swap. And unusual situations
is when PHP script (or Netscape or whatever) suddenly goes nuts and eats
tons of memory in a second. In such case it's better to kill this PHP script
(or Netscape) then anything else.

P.S. Yes, better to debvelop system with unswappable userspace AI or something.
Do it is you really need to cover remaining 1% -- you have sources ... Better
to do right thing in 99% cases (with OOM patch) then to do wrong thing in 99%
cases. Without this patch usually first killed thing is syslogd, then klogd
and slightly after that -- sshd (when you are trying to log on and fix problem).
After that point system is unusable...




-
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/

\
 
 \ /
  Last update: 2005-03-22 13:57    [W:1.337 / U:0.044 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site