Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 19 Oct 2001 06:26:36 -0700
From:      Cy Schubert - ITSD Open Systems Group <Cy.Schubert@uumail.gov.bc.ca>
To:        Ryan Thompson <ryan@sasknow.com>
Cc:        "jslivko@4evermail.com" <jslivko@4evermail.com>, freebsd-questions@FreeBSD.ORG
Subject:   Re: Loads on a Web/Shell Server 
Message-ID:  <200110191326.f9JDQmf34019@cwsys.cwsent.com>
In-Reply-To: Your message of "Thu, 18 Oct 2001 20:51:55 MDT." <Pine.BSF.4.21.0110182037080.211-100000@ren.sasknow.com> 

next in thread | previous in thread | raw e-mail | index | archive | help
In message <Pine.BSF.4.21.0110182037080.211-100000@ren.sasknow.com>, 
Ryan Thomp
son writes:
> jslivko@4evermail.com wrote to stable@FreeBSD.ORG:
> 
> > Hello, 
> > 
> > I have a shell/web hosting company (4EverMail Hosting Services) and I
> > have a little bit of a problem with the loads on my machine. I am
> > hosting a few IRC servers, eggdrop bots and a few apache websites
> > (mainly my own), and the loads are already at 0.15 and so on. 
> 
> Load averages confuse a lot of people. 0.15 is quite low. On the server I
> am logged in to:

It is true that load averages are confusing at first.  Once it is 
understood how the system calculates the load average, it can be a 
useful tuning tool.  Take for example the example:

>  8:37PM  up 205 days, 17:52, 6 users, load averages: 2.44, 2.51, 2.45

If the load average was chronic and vmstat showed that the CPU was at 
100%, this would be an indication that a faster CPU might be required.

If on the other hand very little CPU was being used or most of it was 
system state, another area to look at would be I/O.  Is there a disk 
that is too busy?

Or, the load average might be indicative of a memory problem.  Is the 
scan rate or page out rate too high?  Are we seeing a lot of page 
reclaims?

Generally the load average tells us that there might be a shortage or 
misalignment of resources.  It's a simple way to get a feel for how 
well the system is doing without getting into a detailed analysis of 
the metrics that will ultimately point us in the right direction.   
It's akin to looking up at the sky and saying "it might just rain 
today", then looking at the detailed weather charts to confirm or deny 
one's suspicion (that being the detailed analysis after looking at the 
load average).

For example, on an Alpha system running Tru64-UNIX with a load average 
of > 1.  If the system is performing well, you can bet the system is 
CPU bound.  If on an Alpha system with a load average > 1 is performing 
sluggishly, you can bet the problem is I/O.

FreeBSD systems can take load averages of 6 or 7 before they appear 
sluggish.

On Sun Sparc systems a load average of 3 to 5 seems to be the 
borderline.

Having said all that, a load average of 12, for example, caused by CPU 
bound processes that have been niced will appear to run quite nicely.

Having an intimate familiarity of the hardware, software, and 
applications running on your system and how it normally runs goes a 
long way to understanding load average on a particular system.  What 
may be a high load average on one system may not necessarily be a high 
load average on another system running a different mix of applications.


Regards,                         Phone:  (250)387-8437
Cy Schubert                        Fax:  (250)387-5766
Team Leader, Sun/Alpha Team   Internet:  Cy.Schubert@osg.gov.bc.ca
Open Systems Group, ITSD
Ministry of Management Services
Province of BC



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200110191326.f9JDQmf34019>