Load Question
From: Matt (MattNOChildersSPAM_at_msn.com)
Date: 05/27/04
- Next message: Tommy Reynolds: "Re: vmstat bi disk read or write?"
- Previous message: Thomas Evensen: "Re: Have anyone installed the linux-wlan driver on FC2?"
- Next in thread: Robert W.: "Re: Load Question"
- Reply: Robert W.: "Re: Load Question"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Thu, 27 May 2004 08:45:31 -0500
I am running Fedora Core 1 with Squid as a Caching Server for about 2000
users. It is an older Dell Server and I wanted to monitor the load on the
machine to see if I need to upgrade it. My question is concerning
/proc/loadavg. I am running top, and notice that one of the two processors
stays between 30-50% in normal operation, while the other will stay between
5-20%. During the peak hours, one will run maxed at around 99% while the
other will bounce around between 20-50%. I have yet to see the total cpu
idle amount get below 50%. Now, /proc/loadavg will stick around 0.20 - 0.60
on the 1 minute avg during normal usage, but in peak hours it has gotten as
high as 4.50. Running free shows that during peak and non-peak times I
still am ok on mem usage. So, what exactly is /proc/loadavg monitoring? I
have done some research and all I can find is that the numbers displayed is
an avg of the runnable processes on a machine for 1, 5, and 15 minutes. So
I guess my question is, how should I interprit the output of loadavg. If I
have an average of 4.50 for a 1 minute interval, which I understand is high,
what should I be looking at in order to help get this number lower? Or is
it a problem at all? Any help with this would be much appreciated.
Thanks in advance,
Matt
- Next message: Tommy Reynolds: "Re: vmstat bi disk read or write?"
- Previous message: Thomas Evensen: "Re: Have anyone installed the linux-wlan driver on FC2?"
- Next in thread: Robert W.: "Re: Load Question"
- Reply: Robert W.: "Re: Load Question"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|