NAN Graphs for localhost

[InferiorWang]

I've been banging my head against this for a couple of hours. The poller seems to be working, no errors in the debug enabled logs that I can see outright. I've been working my way through Gandalf's debug docs.

In the rrd info section where you check out the max and min values, I have the following output:

Code: Select all

ds [proc].min = 0.0000000000e+00
ds [proc].max = 0.0000000000e-01
...
ds [proc].value = 1.1492382000e+03

Does that seem right or is something off there? If it's fine, I can't sort out what's going on here.

[gandalf]

The MAX one looks strange. Please try to

Code: Select all

rrdtool tune --maximum ...

to a value of U (equals: no limit)
Reinhard

[InferiorWang]

So from the command line, and in the rra directory, I would type:

Code: Select all

rrdtool tune localhost_proc_7.rrd --maximum proc:U

Am I getting that right? Is there anything else I would want to check/change? Thanks.

[gandalf]

Yes. Please check rrdtool info output afterwards to see the change
Reinhard

[InferiorWang]

It comes up with the max being ds[proc].max = NaN

The funny thing is, checking the templates, it looks like everything is in order as far as the max/min settings. I'm stumped.

[gandalf]

What are the settings of the data template for this item?
Reinhard

[InferiorWang]

Unix - Processes | Unix - Get System Processes | Active

Internal Data Source Name = PROC
Min = 0
Max = 1000
Data Source Type = GAUGE
HeartBeat = 600
Output Field = Proc - Number of Processes

Does this have anything to do with the fact that I'm trying to query against the local host?

[gandalf]

MAX is too low to hold data for values such as 1.1492382000e+03. And "Unix*" Templates are supported only for localhost. See the Script used to get an idea what source is used to fetch that data
Reinhard

[InferiorWang]

Ok, max value being too small makes sense.

I checked the resources directory and the data queries section of the console and I only have the follow entries:

Code: Select all

Karlnet - Wireless Bridge Statistics   	 Get SNMP Data (Indexed)  	 
Netware - Get Available Volumes 	Get SNMP Data (Indexed) 	
Netware - Get Processor Information 	Get SNMP Data (Indexed) 	
SNMP - Get Mounted Partitions 	Get Script Server Data (Indexed) 	
SNMP - Get Processor Information 	Get Script Server Data (Indexed) 	
SNMP - Interface Statistics 	Get SNMP Data (Indexed) 	
ucd/net - Get Monitored Partitions 	Get SNMP Data (Indexed) 	
Unix - Get Mounted Partitions 	Get Script Data (Indexed)

Could that be why the unix processes graph still reads out NaN?

[gandalf]

Unix Processes Data Template should call the following script

Code: Select all

#!/usr/bin/perl

open(PROCESS, "ps ax | grep -c : |");
$output = <PROCESS>;
close(PROCESS);
chomp($output);
print $output;

Please execute this from cli running as cactiuser to see if it works
Reinhard

[InferiorWang]

The output displays the numbers 80 or 81 each time I run it. I gather that's the number of processes running.

[gandalf]

Seems to be ok. Did you run it as cactiuser or as root? Did you alreday check second link of my signature for NaN Debugging?
Reinhard

[InferiorWang]

Yes on both counts. For the hell of it, I added a cisco catalyst switch as a device. Cacti can talk to it over snmp, all my interfaces and vlans come up. I'm getting real numbers with it. That leads me to believe there is something goofy with my local host.

I have gone through your NaN tutorial once. I'll give it another whirl and report back.

Thanks for the help.

[gandalf]

Someone lately reported strange issues on localhost in a specific cacti version. I do not remember, wheter it was 0.8.6h,i or j. What are you running?
Reinhard

[InferiorWang]

I'm running rev J.