Nagios Support Forum

attached is the xml.

thanks!
Adam

I'm wondering if this is just a counter issue. For some reason that I never understood, all of the PNP template arrays have to start at 1 instead of 0, so I think when you use the foreach loop, it's starting by assigning things to $def[0], which is just an empty placeholder. I would experiment with either doing

$count++;

on line 5, OR starting the loop at an index of 1 instead of 0. There is a graph logger in XI for graphs that failed to draw. (If it's not there, touch it, and set perms as 777).

/usr/local/nagios/var/graphapi.log

Access a particular graph that's now showing correctly, and run: You can also try tailing the apache log to see if there are any PHP errors coming from the template.

I tried both count++ and the for string you gave me, and like always the graphs are being generated, but not showing in the frontend.

I created the file with 777 and started looking at the graphs, however, nothing was ever written to that file

the only errors in the error_log were as follows:

PHP Notice: Undefined variable: sa in /usr/local/nagiosxi/html/config/configobject.php on line 933, referer: https://nagiosxi.domain.example.com/nag ... xample.com
[Thu Mar 29 13:42:09 2012] [error] [client 134.48.25.59] PHP Notice: unserialize() [<a href='function.unserialize'>function.unserialize</a>]: Argument is not a string in /usr/local/nagiosxi/html/includes/components/xicore/status-utils.inc.php on line 302, referer: https://vs-nagiosxi/nagiosxi/includes/c ... e=overview

Like i alluded to before, i am able to get to the missing graphs, however i have use the following link and change the source= value:
https://nagiosxi.domain.example.com/nag ... 1333126779

Ok, so before I go and set up a passive test with this template, what happens if you make a copy of this template and name it:

check_by_ssh-cpu.php

Based on the XML file for this service, it's looking for that file as a template. Nagios XI associates graph templates with the command definition that the service uses. So for example, a service that uses the defined command: check_local_disk will look for the template check_local_disk.php for a template.

mguthrie wrote:Ok, so before I go and set up a passive test with this template, what happens if you make a copy of this template and name it:

check_by_ssh-cpu.php

that is what the template is currently named:

Ok, so after way more digging than I was expecting, I discovered a bug/limitation in our current code. It appears as though our code doesn't currently support dynamic counts of datasources in the PNP templating. So what appears to be keeping the additional graphs from being displayed is noted below:

This works, because XI can identify an integer for the datasource value.
$def[1] = "DEF:var1=$rrdfile:$DS[1]:AVERAGE " ;

This currently doesn't work because of how XI currently parses for datasource. I did some snooping on how we could add this support in XI, but at the moment the way it's set up it's going to be a royal pain to add support for it, since PNP templates aren't really templates, they're just code that people can setup however they want. You could modify the template to work differently, the key string that XI is looking for is:

$def[1] and $def[2] and so on.

mguthrie wrote:Ok, so after way more digging than I was expecting, I discovered a bug/limitation in our current code. It appears as though our code doesn't currently support dynamic counts of datasources in the PNP templating. So what appears to be keeping the additional graphs from being displayed is noted below:

I appreciate all your work on this!! do you know if it will be possible to put this in as a bug so that it gets fixed in a later release? (also, what would happen if i defined too many datasources initially EX: use the same template for a 2 and a 24 cpu server and have 24 ds's defined?)

do you know if it will be possible to put this in as a bug so that it gets fixed in a later release?

Yeah, go ahead and file this on our tracker: tracker.nagios.com so that way it doesn't fall off our radar

I think what would happen at the moment is that for the 2CPU machine you'd get 2 good graphs and a bunch of blank ones. Although it'd be a little bit of a pain, you could create a template for a 2CPU, 4CPU, 8CPU and so on. Not very elegant, but it would work in the meantime.

I have a feeling the fix for this would ultimately be a small change to the code once we figured out what the solution is, but in the meantime I might have to let this idea marinate for a little while until we can come up with something that's not a drastic overhaul of the code ; )

Bug filed here:

http://tracker.nagios.com/view.php?id=249

Thanks again for all your help!

Has there been any updates to this bug? I don't see any in the tracker.

Even if someone could point me to the file in which the DS logic is run, I could probably figure this out myself.

I have a template I've created with about 12 different dynamically created DS's. I've confirmed that if I use this template as the "default.php" template, all graphs are displayed correctly. But if I run it as intended (i.e. leave default.php alone and run this template specifically for the check it's meant for) I only get 2...

I'm sure it's a relatively easy fix... I know the suggestion was to specify $def[1] etc... manually, but because these are generated dynamically, the number of counts will vary from host to host.

Cheers,

Ben