Beowulf and Big Brother

[Robert G. Brown]

On 12 Nov 2002, Jim Meyer wrote:

> Hello!
> 
> On Mon, 2002-11-11 at 11:55, Ollisl wrote:
> > [...] I was just wondering if it is reasonable or smart to monitor
> > these master nodes with Big Brother? Or is there even ready-made shell-
> > scripts for that?
> 
> Another thing to look at is Ganglia (http://ganglia.sourceforge.net/)
> which not only monitors services but tracks both aggregate and local
> statistics. Mind you, I've never installed it; the last place I worked
> never outgrew Big Brother (and never had time to experiment with it =\)

You can also play with xmlsysd and wulfstat.  I finally got around to
setting up a page of sorts for them:

  http://www.phy.duke.edu/brahma/xmlsysd.html

from which they can be retrieved.  You should be able to install the
binary rpm's, chkconfig xmlsysd on (or you may want to hand edit
/etc/xinetd.d/xmlsysd to adjust e.g. the number of connections or
allowed client connections), restart xinetd, and verify the daemon's
functionality via telnet hostname 7887.  When you connect, enter "init"
(no quotes) and then "send" to see the full set of non-PID values
supported.  "quit" to exit.

To use wulfstat, create a ~/.wulfhosts in your home directory like:

<?xml version="1.0"?>
<wulfstat>
  <hostrange>
    <hostfmt>b%02d</hostfmt>
    <imin>0</imin>
    <imax>15</imax>
  </hostrange>
</wulfstat>

(to monitor b00 to b15) and then run wulfstat in a 80x35 xterm.

There are a couple of minor bugs in wulfstat -- a couple of displays on
the menu don't really exist, it can hang if you page up or down to
quickly -- but it is harmless to kill it and restart it at any time and
I'm working on the bugs fairly aggressively.

I think you'll like what it displays.

   rgb

Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu