[mrtg] Broken gauge seemingly without cause

[Edwin A. Epstein III]

Hello, 

I've been using MRTG to graph the number of stations connected to various Ubiquiti access points. This was working quite well with no issues till apparently a new firmware broke MRTG functionality in some way that I can't identify. The working configuration: 

"Target[192.168.1.1_stcount]: CnTWaLK1. 3.6.1.4.1.14988.1.1.1 .2.1.3&CnTWaLK1. 3.6.1.4.1.14988.1.1.1 .2.1.3:public at 192.168.1.1 +0 
WithPeak[192.168.1.1_stcount]: ymw 
SetEnv[192.168.1.1_stcount]: MRTG_INT_IP="192.168.1.1" MRTG_INT_DESCR="stcount" 
Options[192.168.1.1_stcount]: gauge,growright,nopercent,noi 
Title[192.168.1.1_stcount]: The-Broken-Gauge 
MaxBytes[192.168.1.1_stcount]: 256 
XSize[192.168.1.1_stcount]: 600 
YTics[192.168.1.1_stcount]: 1 
YLegend[192.168.1.1_stcount]: Station Count 
Legend2[192.168.1.1_stcount]: stcount 
LegendO[192.168.1.1_stcount]: &nbsp stcount 
ShortLegend[192.168.1.1_stcount]: STc 
PageTop[192.168.1.1_stcount]: <h1>192.168.1.1 (stcount)</h1> 
<div id="sysdetails"><table> 
<tr><td>Nagios Name:</td><td>The-Broken-Gauge</td></tr> 
<tr><td>IP Address:</td><td>192.168.1.1</td></tr> 
<tr><td>Make & Model:</td><td>(NSM5) Ubiquiti Networks NanoStation M5</td></tr> 
<tr><td>MRTG Database ID:</td><td>548966</td></tr> 
<tr><td>MRTG Server:</td><td>Node 45 (10.10.3.3 )</td></tr> 
<tr><td>Maintainer:</td><td> noc at rhinobee.com </td></tr> 
<tr><td>Gauge Type:</td><td>stcount</td></tr> 
<tr><td>Gauge Name:</td><td>Station Count Graph</td></tr> 
</table></div>" 

I'm aware (I think), that MRTG and snmpwalk use different libraries to manage the traffic. Hence, WaLK on a target line allows me to graph values otherwise unreadable/unreachable. This is the case with much of Ubiquiti equipment that is using a MIB from MikroTik (The OID above). I have two different versions of this gauge that support the two different MIBs that Ubiquiti uses, so I'm quite confident about my OIDs being correct. On the gauges that have issues a value of 0 is always received and graphed from the count walk. However, performing a snmpwalk still shows: 

"SNMPv2-SMI::enterprises. 41112.1.4.5.1.5.1 = INTEGER: -58" 

A count walk should return a value of 1, but never does. I can't see that anything has changed in the configuration file, or even the output from snmpwalk. I'm at a loss on how to proceed diagnostically here and would like some assistance in debugging MRTG and/or logging better information to find a fix here. Everything seems like it should be working as normal.... 


Sincerely, 

Edwin A Epstein, III 
Rhinobee Internet Services 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.oetiker.ch/pipermail/mrtg/attachments/20160713/1a286dc6/attachment.html>