Monitor Disk I/O

[mkeey]

Would like to start monitoring server disk i/o. Cannot find a wizard that will add that. Please advise if there's any documentation on how to add that "service" to a host in nagios xi.

[swolf]

If you're trying to do this on a Linux server, there are several plugins on the Nagios Exchange that should match what you need. In particular, this one seems to be fairly highly-rated. Here is the documentation for managing new plugins in your Nagios XI environment.

If you're working with Windows servers, you should be able to find the relevant Windows performance counters (which will change if you're using, say, a local disk vs. an SMB share) and monitor those via NCPA.

Let us know if you have any other concerns.

[mkeey]

We will want to eventually do this on both Linux & Windows and possibly AIX at some point.

However, I'm primarily concerned with Windows at this time. I did do a goggle search and found this...
check_xi_service_nsclient!!COUNTER !-l "\LogicalDisk(C:)\% Disk Read Time"!!!!!

But, all that returns is a "0". Not sure if that's working the way I want. Would like some specific instructions on how to code the monitor for Disk I/O.

[mkeey]

My apologies, let me get a little more specific. We have VM's in both Windows and Linux. They are using SAN storage. So, we really need to monitor Disk I/O Wait and then later on IO Queing (aka depth). But, I need to concentrate on Wait for now so that we can detect when there are problems with data retrieval from the SAN. So, I really need instructions on how to setup a service to monitor that Disk I/O Wait.

[npolovenko]

@mkeey

However, I'm primarily concerned with Windows at this time. I did do a google search and found this...
check_xi_service_nsclient!!COUNTER !-l "\LogicalDisk(C:)\% Disk Read Time"!!!!!

But, all that returns is a "0".

You may need to open the Performance Monitor application on the windows server to see which performance counter displays the information you need. Go to Monitoring Tools -> Performance Monitor, right click on the white part of the screen and select "Add counters". From there you can add different counters to see which one will return the information you're looking for.

[mkeey]

Have been side-tracked with other priorities. This case is still important but I've been delayed in working on this. Please leave case open as I hope to readdress this week - MKeey

[npolovenko]

@mkeey, No problem, will leave this opened.

[mkeey]

Just to keep you posted and this thread open, I finally got a chance to look at performance monitor and got a list of LogicalDisk counters. We're now investigating which one will work best for what we're looking for.

May need some assistance into incorporating that into a service monitor so please leave this open a while longer. I appreciate your patience.


Martin

[npolovenko]

@mkeey, No problem, I'll leave this open. Thanks for the update.

[jvaira]

I was about to post a separate topic for an IO question but it sounds like @mkeey may run into this as well so I'll post it here. We are using NRPE with the checkcounter command to pull IO data from windows performance monitor. Although this works perfectly in most situations we are running into an issue when monitoring disk seconds per write and seconds per read. The issue is that windows performance monitor measures this in seconds and not milliseconds which means that a typical reading may be 0.005 seconds. When looking at this data in the Nagios performance graphs we just see a flat zero. Is this because the performance graph in Nagios registers any reading less than zero seconds as a zero?