2002 03 07

                    LINUX NEWS
        Resources & Links From CramSession.com
             Thursday, March 7, 2002


1) Sean’s Notes

2) Linux News

It's Hammer Time
Linux Will Prevail
I'm Not the Only One
Open Source - Now!

3) Linux Resources

CVS Book
Put That Alpha to Good Use
More Uses for Alphas
Bootable CD Firewall
Summary:  Don't Delete libc

4) App o’ the Week

~~~~~~~~~~~~~~~~~~~~~~ ADVERTISEMENT ~~~~~~~~~~~~~~~~~~~~~~~

Serebra Learning Corporation knows that it’s true, you get paid more if you have the skills. Learn at your own pace with our dynamic training programs for the skills needed to succeed in today’s IT market. The Best Way to Learn Anything, Anywhere, Anytime. Check out this month’s specials!


For information on how to advertise in this newsletter
please contact mailto:adsales@CramSession.com or visit

1) Sean's Notes

In the past, I've written about the various ways to deal with
your system when it's slow.  The "top" command gives you a
real-time look at the processes running on your system, ranked
in order of CPU usage.  The load average, which appears at
the top of the "w" or "uptime" commands gives you a look at
the size of the run queue - how many processes are waiting for
the CPU.  The "free" command will show you where Unix has
allocated your memory, if the applications use it all, you're
going to have to hit the disk for swap space.

What I haven't touched on is how to tell if you've overloaded
your machine?  Disk too slow?  Not enough CPU or memory?

The "vmstat" command is a great tool to show you your disk,
swap, and CPU activity all at once.  The main option used
with vmstat is how many seconds to wait between measurements;
I usually choose 5 seconds.

$ vmstat 5
   procs                      memory    swap          io     system         cpu
 r  b  w   swpd   free   buff  cache  si  so    bi    bo   in    cs  us  sy  id
 2  0  0   7156   4620  53276 124432   0   0    22    12   96   168   2   7  90
 2  0  0   7156   4620  53300 124432   0   0     0    22 1014  2316   2   3  95
 1  0  0   7156   4620  53308 124560   0   0    26     2  987  2198   1   3  96

One thing to remember is that the first line represents the
average values since your system has booted, and subsequent
lines are instantaneous readings.  If you want to see how
everything is doing at this instant, then you'll skip over
the first line.

The information from vmstat has to be divided into columns
representing the various measurements (top row).  Each column
heading is right-justified such that procs refers to r, b,
and w (second row), and memory refers to swpd through cache.

"procs" gives you an indication of how many processes are
running, but not in the same sense that the load average does.
The "r" column tells you how many processes are waiting right
now to be run (note that unlike top, vmstat excludes itself
from this count).  "b" is the processes that are blocking on
IO (ie waiting for disk or some other device).  We're going to
ignore "w" for now.

A high number of "r" processes tells us we've got a lot of
processes waiting for the CPU at this instant.  Over time,
this will translate to a high load average.  Lots of blocked
processes means your processes are twiddling their thumbs,
waiting for external resources.

If you run vmstat for a bit, then stop it and run "free",
you'll see the four memory values repeated, which might make
it easier to figure out how it all adds up.  Linux will try
to allocate all your memory for *something*, so it's normal
to see a small amount of memory reported as free.  The two
columns to the right, buff and cache, are where most of your
"free" memory normally is (buffering data, or caching data
for later use).  If your applications need it, Linux takes
memory away from here and gives it to them.  Thus, the "free"
column is derived from your physical memory minus what's been
allocated to applications, buffer, and cache.  The memory
space left for your applications to grow is then the sum of
the free, buff, and cache columns.

Warning flags here are high values of "swpd", meaning that
memory is being swapped to disk.  Since disk is significantly
slower than memory, excessive swap usage can cause a huge
performance hit.

Speaking of swap, the "si" and "so" give you more details on
the rate (in kbytes/sec) at which memory is being swapped in
and out from the disk.  High values of "si" means that your
memory is swapped to disk, and you're forced to bring it into
RAM.  "so" means you're busy dumping RAM to disk in order to
free it for other uses.  Practically, high swap usage is a sign
you might need more RAM, so it doesn't matter if it's going in
or out to disk.

Next we have "bi" and "bo", which refer to disk blocks in and
out respectively.  If you have multiple drives, this is derived
from the activity of all your drives.  Each block is a kilobyte
(on a hard drive), so high numbers mean you're working the disks
hard.  The level at which you become concerned depends on the
hardware, and the layout of the disks.  Numbers above your
normal average would prompt you to delve deeper into your disk
statistics with other utilities such as "iostat".

The system section tells you about your hardware.  "in" is the
number of interrupts per second your computer is generating,
and "cs" is the context switches per second.  A context switch
happens when the scheduler switches processes on the CPU, or the
process transfers control to the kernel (ie a system call).
Interrupts and context switches are basically overhead.  Lots of
interrupts might be from devices such as busy network cards.
More advanced cards, such as some of the 3COMs, are more efficient
on interrupts, using DMA to transfer larger amounts of data.

CPU, as the name implies, refers to the processor itself.  The
last three columns should add up to 100, give or take rounding
errors.  "us" means the percentage of the CPU that is tied up
with user tasks, "sy" the percentage devoted to system (i.e.
kernel) tasks, and "id" is the percentage idle.  A system that
sits at zero idle and high "us" is begging for more or faster CPUs.

Your computer is a complex device, and its smooth operation
involves watching many key figures for signs of a bottleneck.
vmstat is an efficient tool for telling you what general area
is causing you problems, be they memory, IO, CPU, or other.
Like most performance related utilities, numbers vary widely
from system to system.  Periodically running vmstat for a few
samples can give you a baseline from which to compare the next
time things slow down.

Long live the Penguin,


2) Linux News

It's Hammer Time
AMD's next generation of processors, using the so called
"x86-64" technology, will enjoy support and optimizations
from the stock Linux kernel. SuSE are the ones making the
patches, and odds are that they'll find their way into the
2.4 kernel.


Linux Will Prevail
It's always good to hear opinions on what is needed to make
Linux succeed on the desktop. The author of this article
brings up a couple of new points, though, that are worth a read.


I'm Not the Only One
Looks like I'm not the only author to ditch his trusty email
client in favour of Evolution. Citing ease of use and features,
this guy went from Sylpheed to Evolution, and seems to be happy.


Open Source - Now!
This is an interesting advocacy project started by Red Hat.
They're targeting schools, and why not? That's where many
habits and preferences are formed, not to mention that it
would save school systems money that could be better spent
on teachers.


3) Linux Resources

CVS Book
While there are many documents out there on how to use CVS
to keep track of changes to files, this one is unique in
that it focuses on how to administer a CVS server, and to
deal with uncommon uses of CVS.


Put That Alpha to Good Use
x86 isn't the only architecture that Linux runs on, but it's
the most popular. Old Alpha boxes can be had for a song on
eBay, and can provide great experience on running Linux on
non-standard hardware. This article explains how to get
Debian running on Alpha hardware.


More Uses for Alphas
If Linux isn't your thing (then why are you reading this?),
but you still want to do something with that Alpha, maybe
VMS is more your style. Unix and VMS have quite the history
together (mostly as competitors), and I've found using VMS
makes me appreciate Linux even more!


Bootable CD Firewall
This is a brilliant idea: making a bootable CD that runs as
a firewall. No worries about anything being changed on a
read-only medium! Though the example is for FreeBSD, all the
concepts are the same for Linux--only the files are different.


Summary:  Don't Delete libc
This is an interesting account of a user deleting an
important symlink from /lib. Learn from his mistake, trust
me. If you do want to try it out for real, make sure it's
on a machine you don't mind losing.


4) App o' the week
Webmin is a system to allow you to manage your system over
the web. Users can be assigned privileges, so one user can
edit DNS, while not being able to add users, and so on.
Between the standard modules and the third-party modules,
there is very little you can't do.


(C) 2002 BrainBuzz.com, Inc. All Rights Reserved.

         This message is from CramSession.com.

You are currently subscribed to the
   Hottest Linux News and Resources
   as: sean@ertw.com

To un-subscribe from this newsletter by e-mail:
   send a blank email message to:


To Subscribe to this newsletter by e-mail:
   send a blank email message to: