Archived community.zenoss.org | full text search
Skip navigation
1 2 3 Previous Next 6533 Views 30 Replies Latest reply: Sep 10, 2011 11:39 PM by nilie RSS
stevez Rank: Green Belt 131 posts since
Aug 12, 2010
Currently Being Moderated

Sep 8, 2011 2:04 PM

Guage RRD Behavior

Looking at ifAdminStatus (1.3.6.1.2.1.2.2.1.7). I'm pulling it as a guage. It should only return a 1 or a 2 (up/down) of type integer. Whenever I do an SNMP walk that is all I see:

 

[zenoss@bigzenosstest CustomComponents]$ snmpwalk -vXXX -cXXX X.X.X.X 1.3.6.1.2.1.2.2.1.7

IF-MIB::ifAdminStatus.1 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10101 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10102 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10103 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10104 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10105 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10106 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10107 = INTEGER: down(2)

IF-MIB::ifAdminStatus.10108 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10109 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10110 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10111 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10112 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10113 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10114 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10115 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10116 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10117 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10118 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10119 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10120 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10121 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10122 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10123 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10124 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10125 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10126 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10127 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10128 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10201 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10202 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10501 = INTEGER: up(1)

IF-MIB::ifAdminStatus.10502 = INTEGER: up(1)

 

 

Here is my RRD Archive header - as you can see it is of type gauge:

 

<!-- Round Robin Database Dump --><rrd> <version> 0003 </version>

        <step> 300 </step> <!-- Seconds -->

        <lastupdate> 1315504776 </lastupdate> <!-- 2011-09-08 13:59:36 EDT -->

 

        <ds>

                <name> ds0 </name>

                <type> GAUGE </type>

                <minimal_heartbeat> 900 </minimal_heartbeat>

                <min> NaN </min>

                <max> NaN </max>

 

                <!-- PDP Status -->

                <last_ds> 2.0 </last_ds>

                <value> 5.5352637800e+02 </value>

                <unknown_sec> 0 </unknown_sec>

        </ds>

<!-- Round Robin Archives -->   <rra>

                <cf> AVERAGE </cf>

                <pdp_per_row> 1 </pdp_per_row> <!-- 300 seconds -->

 

                <params>

                <xff> 5.0000000000e-01 </xff>

                </params>

                <cdp_prep>

                        <ds>

                        <primary_value> 1.6305668200e+00 </primary_value>

                        <secondary_value> 1.0000000000e+00 </secondary_value>

                        <value> NaN </value>

                        <unknown_datapoints> 0 </unknown_datapoints>

                        </ds>

                </cdp_prep>

                <database>

 

When I shut/noshut the interface I would expect only 1s and 2s in my RRA. However when I shut the interface down for one poll I get 1.61 and then it gets up to 2 on the next. Likewise, when I do a no shutdown (should go 2 to 1) I get a 1.37 between the 2 and 1 polls. It's my understanding that this should not be happening with Guage. Has anyone see this behavior before? This is on a Cisco 3560 for reference.

 

 

                   

<!-- 2011-09-08 12:35:00 EDT / 1315499700 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 12:40:00 EDT / 1315500000 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 12:45:00 EDT / 1315500300 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 12:50:00 EDT / 1315500600 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 12:55:00 EDT / 1315500900 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:00:00 EDT / 1315501200 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:05:00 EDT / 1315501500 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:10:00 EDT / 1315501800 --> <row><v> 1.6104536167e+00 </v></row>
                    <!-- 2011-09-08 13:15:00 EDT / 1315502100 --> <row><v> 2.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:20:00 EDT / 1315502400 --> <row><v> 2.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:25:00 EDT / 1315502700 --> <row><v> 2.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:30:00 EDT / 1315503000 --> <row><v> 2.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:35:00 EDT / 1315503300 --> <row><v> 1.3743194867e+00 </v></row>
                    <!-- 2011-09-08 13:40:00 EDT / 1315503600 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:45:00 EDT / 1315503900 --> <row><v> 1.0000000000e+00 </v></row>
                    <!-- 2011-09-08 13:50:00 EDT / 1315504200 --> <row><v> 1.6305668200e+00 </v></row>
  • jmp242 ZenossMaster 4,060 posts since
    Mar 7, 2007
    Currently Being Moderated
    1. Sep 8, 2011 2:06 PM (in response to stevez)
    Re: Guage RRD Behavior

    I think this is related to the RRD averaging. There are a lot of forum posts on this that explain it and other configs you can do also to change this in some cases.

     

    --

    James Pulver

    Information Technology Area Supervisor

    LEPP Computer Group

    Cornell University

  • jmp242 ZenossMaster 4,060 posts since
    Mar 7, 2007
    Currently Being Moderated
    3. Sep 8, 2011 2:41 PM (in response to stevez)
    Re: Guage RRD Behavior

    You're storing averages though, per RRD docs:

    AVERAGE

     

    the average of the data points is stored.

     

    You probably want to use LAST, though I'm not entirely sure.

     

    --

    James Pulver

    Information Technology Area Supervisor

    LEPP Computer Group

    Cornell University

  • jmp242 ZenossMaster 4,060 posts since
    Mar 7, 2007
    Currently Being Moderated
    4. Sep 8, 2011 2:45 PM (in response to jmp242)
    Re: Guage RRD Behavior

    Point 2: RRD entries are always a rate. You're very unlikely to get whole numbers.

     

    --

    James Pulver

    Information Technology Area Supervisor

    LEPP Computer Group

    Cornell University

  • Shane Scott ZenossMaster 1,373 posts since
    Jul 6, 2009
    Currently Being Moderated
    5. Sep 8, 2011 2:47 PM (in response to jmp242)
    Re: Guage RRD Behavior

    Steve:

     

    James is correct. If you use LAST, you'll get the last value, unconsolidated. You wont be able to graph whole numbers since everything in RRD is a rate.

  • Shane Scott ZenossMaster 1,373 posts since
    Jul 6, 2009
    Currently Being Moderated
    7. Sep 8, 2011 2:52 PM (in response to stevez)
    Re: Guage RRD Behavior

    Steve:

     

    Thats right, but its still going to graph a rate for slopes derived between collected numbers. If your current falls on a collection point it will be whole, if not, it'll be a rate.

  • Shane Scott ZenossMaster 1,373 posts since
    Jul 6, 2009
    Currently Being Moderated
    8. Sep 8, 2011 2:52 PM (in response to Shane Scott)
    Re: Guage RRD Behavior

    Steve:

     

    Why are you graphing the up/down status anyway?

  • Shane Scott ZenossMaster 1,373 posts since
    Jul 6, 2009
    Currently Being Moderated
    11. Sep 8, 2011 3:07 PM (in response to stevez)
    Re: Guage RRD Behavior

    Steve:

     

    Oh gotcha. My mistake. Acctual values are be stored, but I don't think non rate values can be presented via rrdtool less the step is the same as the period, essentially zero slope. Really its still giving you a rate but they'll be whole numbers you're looking for.

  • Shane Scott ZenossMaster 1,373 posts since
    Jul 6, 2009
    Currently Being Moderated
    12. Sep 8, 2011 3:13 PM (in response to Shane Scott)
    Re: Guage RRD Behavior

    Steve:

     

    Right, I understand. You'll need a step divisible by 10 that is equal to the period. *I think* There is a way to achieve slopes of zero.  Any chance of using snmp traps for int up/down events? Much easier and quicker response time.

1 2 3 Previous Next

More Like This

  • Retrieving data ...

Legend

  • Correct Answers - 4 points
  • Helpful Answers - 2 points