> From: [email protected] [mailto:discuss-
> [email protected]] On Behalf Of Jonathan Bayer
>  
> The problem is that all of a
> sudden the load average skyrockets to 40-50 or even higher. This kills
> the server and makes it totally unresponsive.

Remember what "load average" means.  The number of processes that are running 
or waiting.  To answer the question of why this is happening on your system, 
run top.  Look at which processes are running or waiting.

If it just so happens the offense is a million httpd processes spawning, then 
you've either got a misconfiguration in your httpd config (should lower the 
number of concurrent threads, perhaps) or as others have pointed out, maybe 
you're swapping (out of memory.)

This brings up a completely separate flame war:

I always say, and sometimes inadvertently start arguments over this (google 
"swap is religion") that it is good to have some swap available in your system 
- you're giving the kernel an extra degree of freedom, to swap out idle 
processes in order to keep ram available for cache & buffer, if the kernel 
happens to feel that's a more valuable resource at the time.  This is a 
performance enhancer.  But if you have too much active memory consumption by 
your active processes, you *absolutely* never want to be swapping active 
memory.  (Where "absolute" is used as a non-absolute term, a couple of fringe 
exceptions here and there, but not applicable to present case.)  I recommend 
giving a system generally 1G of swap space, just to limit the damage if some 
processes runaway and start swapping active memory.  Before too long, you want 
your system to fail the offending processes with "out of memory error."

_______________________________________________
Discuss mailing list
[email protected]
https://lists.lopsa.org/cgi-bin/mailman/listinfo/discuss
This list provided by the League of Professional System Administrators
 http://lopsa.org/

Reply via email to