Wrong values returned by HP OID's for pagefile used/free

[dr0h]

Hi Everyone,

I've been running into some weird problems after starting to graph some server performances with Cacti. I have a bunch of HP ProLiant servers (dl380's, dl385's, dl580's and dl585's) all installed with either Windows 2000 or Windows 2003. I've tested on PSP (ProLiant Support Pack) 7.10 through 7.40b.

I'm using the HP template to graph out the servers information, including pagefile used/free data. The OID's I'm using are:

Paging Memory Size
.1.3.6.1.4.1.232.11.2.13.3.0
Paging Memory Free
.1.3.6.1.4.1.232.11.2.13.4.0

The problem is that these OID's don't return the proper values. Most of my servers have two pagefiles (4096 Mb each on seperate drives). When I query the OID's, I normally only received 3.5 Gb as the total page file size, when the system detects it properly at 8 Gb.

Oddly enough, even the amount free sometimes exceeds the total amount (as shown in the attachment).

Has anyone run into this problem before?


Thanks,
-Joshua

[BSOD2600]

Being that you're using a HP ProLiant snmp addon, I suggest you contact HP about this problem.

Could use a WMI script to get this data (with a bigger performance hit).

[dr0h]

I'm trying to stay away from WMI

I figured I'd try here first since I found the actual OID's and templates in the forums


Thanks..
-Joshua

[Swordfish]

I think you'll find that, as with most things with an OID, that you are pulling two values for one paging file.
Do an snmpwalk from a smaller OID base (take one or two digits off the right), then you will probably see that there are seperate values for each file.
You will then need to add them together, or I'd suggest stacking them, so you still see the two files.
If the used portion of each paging file is important, rather than the total only, then I've seen others negate one of the values, so that you get a centre line at zero and +4gb and -4gb.
The used values, do the same, and you get the graph growing both sides of the mid line.

Fishy