Single Windows 2003 box not reporting memory

[rjholtz]

I have Cacti 0.8.6i running on a linux box, monitoring ~150 hosts and ~2000 sources. All are good except for one Windows 2003 Server w/SP1 that has never reported it's physical memory to Cacti (we knew, but never took the time to figure out why). Last week we rebooted cause it was acting funky and all of a sudden it reported it's Memory, and let us know it was going WAY into Virtual Memory. So we upgraded its RAM and now it won't report memory anymore. All other Server that are being monitored are fine. What can I look at on this Windows box that will show me what is wrong?
Thanks everyone!
Ray

Additional Info:
We also run Nagios to alert us of problems. Nagios also cannot see the memory on this server, not can it see the uptime of the server. Just wanted to let that be known in case it helps.

[BSOD2600]

It could be the DS max value is too small for the amount of memory you installed, so its discarding the data. Turn the logging level to medium so you can see the values that are being returned from the server. If its returning memory data but the rrdfile doesn't, then the DS in the rrd file needs to be increased (along with the data template).

[rjholtz]

With Medium logging enabled, this line came back:

Code: Select all

04/20/2007 03:40:27 PM - CMDPHP: Poller[0] Host[37] DS[3200] CMD: perl /usr/local/cacti/scripts/nt_memuse.pl apps01, output: total:0.00 used:0.00

I'm not sure what you mean by the DS value... but it looks like the server isn't returning data. Any idea what to look for now?
Thanks for your help,
Ray

[BSOD2600]

Look in that perl script for what its doing to get the memory value and then troubleshoot from there.

If its using WMI, then look into possibly re-registering the WMI hooks for windows. run wmiadap /f.

[rjholtz]

Well, it looks like it is using a Nagios checker, below is the nt_memuse.pl file that Cacti calls. which would be why Nagios can't see the mem usage either. I guess a former employee set it up like this, I don't think that Cacti would use Nagios material and not make it known. If anyone here might be able to show me what might be wrong, I'd love it, but I think I will have to check on the Nagios Forums.

Thanks for helping me find this BSOD!

Code: Select all

#!/usr/bin/perl

$response = `/usr/local/nagios/libexec/check_nt -H $ARGV[0] -p 1248 -v MEMUSE`;
chomp $response;
($used) = ($response =~ /used: (\d+\.\d+) Mb/);

($total) = ($response =~ /total:(\d+\.\d+) Mb/);

print "total:$total used:$used\n";