Re: [cobalt-users] HELP HELP HELP RAQ4r CPU slowly creeps up till server is un-usable

[Larry Smith <lesmith@xxxxxxxxx>]

INRE RE: [cobalt-users] HELP HELP HELP RAQ4r CPU slowly creeps up till server 
is un-usable:
> -----Original Message-----
> From: cobalt-users-admin@xxxxxxxxxxxxxxx
> [mailto:cobalt-users-admin@xxxxxxxxxxxxxxx] On Behalf Of Larry Smith
> Sent: Friday, March 21, 2003 7:25 AM
> To: cobalt-users@xxxxxxxxxxxxxxx
> Subject: Re: [cobalt-users] HELP HELP HELP RAQ4r CPU slowly creeps up till
> server is un-usable
>
>
> INRE RE: [cobalt-users] HELP HELP HELP RAQ4r CPU slowly creeps up till
> server
>
> is un-usable:
> > Hello,
> >
> > Alright this is a really weird issue and I am hoping that someone may
> > have some insite on how to fix this- it has become to frustrating for
> > words...
> >
> > What Happened:
> > -Tuesday night around midnight my RAQ4r decided to max out the CPU
> > rendering SMTP useless and web pretty much the same. Every since then
> > I have had to pretty much reboot the server ever hour... After that
> > the server usually takes an hour for the CPU to ramp up to a critical
> > state.
> >
> > What I have done.
> > - First I though it was the mirroring between my two drive (I have the
> > latest kernel installed---slow, fixed with echo         ETC... command)
> > that wasn't it..
> > - Second I thought it was Tomkat- Removed it from my httpd.conf file.. It
> > doesn't start-- Which is good-- cause it would   take forever to get the
> > thing started.. I though that was eating my CPU... Nope! Wasn't it..
> > - Third I have read every log file I can think of.. /var/log-->
> > messages, secure, maillog, xferlog-- CD'd  to httpd read access, error
> > logs and I have found nothing suspicious other that the usual CPU load
> > to high,, POP login rejected cause Sendmail has shut it self off due
> > to the 20% process limit (Which I have gone into the Sendmail.cf file
> > and upped it to 80 just get users mail to send,, ;-( This is not good
> > and I know that but,, I am in a bad situation..
> > - Fourth I have tailed all of the logs above, mainly the access log in
> > httpd and watch the CPU usage- Hoping to find the ONE site that is
> > killing my CPU.. But I have yet to find it.. I have also shut down the
> > sites that I
> > suspected- but this did not help either.. It's just a gradual process-
>
> that
>
> > the naked eye's cannot see..
> >
> > If anyone has experienced this, or heard about IT through someone's
> > second cousins grandpa's brothers best friend ;-)) and can tell me
> > about it or how they fixed it-- Or another way to track it down IT
> > WOULD BE SO GREATLY APPRECIATED!! I have put two day's into this
> > problem and it's driving me nutz!
> >
> > Thank You,
>
> Matt,
>
>   How about running "top" on the system and watch which processes are
> "eating" the cpu ???
>
> Yes, it my last email to the list I stated that I did this-
>
> "Also, I forgot to mention that I have been watching TOP and have found no
> one process that is using more than 1% or 1.0 of the CPU. Oh,, and I have
> been using the command ps -aux to try and find a runaway process that way
> and it pretty much showed me the same thing as TOP."
>
> How about something else?
>
> Thank You,
>
> Matt
>

 Hmmm, interesting, but what does TOP say about memory, swap, CPU (idle) and 
so forth.  If you are saying "my RAQ4r decided to max out the CPU" then 
something has to show up either in the "ps fax" or top.  Top breaks down the 
utilization figures into user, system, nice, idle.  Which one is "using" all 
the cpu??  It may not be a "single" process, but may be the number of 
processes that a particular program is spawning.

EG:  
try "ps faux | grep mail" to see how many processes doing mail are running
try "ps faux | grep http" to see how many http processes are running

-- 
Larry Smith
SysAd ECSIS.NET
sysad@xxxxxxxxx