On Monday 09 December 2002 16:21, Seth Hollen wrote:
> I use nagios a little to monitor some servers at one location and such. But
> I have a tough one here.
>
> A friend called me up today looking for a solution.
>
> Problem:
> A friend of his has a business with about 9 computers running various
> programs in windows and DOS (I know I know) For example one only does
> shipping labels, another does inventory, etc. these are all controlled from
> 1 monitor/keyboard/mouse, with a KVM switch.
>
> The problem is that being windows the programs occasionally lock up. Not
> the whole computer mind you (that happens too) but the program will crash
> or freeze. Since no one is looking at the screen it is hours or days till
> the problem is noticed.
>
> I am not familiar enough with nagios to write plug-ins. I was wondering if
> it's possible for nagios to somehow monitor a remote process? Perhaps
> through SNMP? While windows may be the problem here, I was hoping Linux was
> the answer.
>
> Any ideas on a solution to this would be appreciated :)
It is possible to do this. The plugin documentation thoroughly describes the
two methods of creating Netsaint/Nagios check scripts (positional
command-line arguments vs option based command-line arguments).
If you can write a command-line tool that queries the state of the remote box,
you can monitor it with nagios.
The easiest way might be to setup cygwin sshd on all of the windows boxen,
then inside your Nagios check-script merely ssh in and run scripts on the
servers themselves. If you're running code locally on the boxes, it's
relatively simple to run whatever command-line tools you may need. I
recommend a copy of the Win2k resource kit, as well as utilities from
Sysinternals.com. Take careful note of cygwin native tools however: for
example, you can get a full Win32 process list with "ps -eW".
I'm never really keen on SNMP, largely for security reasons. If you can get
the Microsoft SNMP agent to enumerate running processes, you may be able to
monitor a process to see if it is running using purely SNMP. I wouldn't take
this route, personally.
This is what we do at NKS when forced to build/manage Win2k boxen. Our
auto-build environment is entirely based off of a MS Client TCP/IP DOS grub
netboot floppy that maps to a Samba share and uses unattended files to
auto-build the boxen. I've also hacked up a rather neat OEM "boot" directory
that has stages that run during multiple reboots of the system, to
automatically hotfix and install baseline files we use to manage customer
boxes. All NKS auto-install packages are built with the NSIS installer
toolkit (http://nsis.sf.net) I can build a new machine from scratch, without
ghost and associated Windows driver problems, in about 20 minutes.
If they're going to make me support these blasted boxes, I'm not going to
suffer the GUI hell and RIS/Active Directory nightmare that is Microsoft's
"standard" way of adminning them.
Every Win2k box gets an sshd - it's a matter of principle ;)
We use a modified version of Nagios for our internal monitoring purposes. It
works wonderfully.
-- - Ian C. Blenke <icblenke@nks.net>(This message bound by the following: http://www.nks.net/email_disclaimer.html)
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 18:56:05 EDT