MRTG errors

 Hi,

I get a lot of errors resembling this:

> 2010-12-14 07:29:32 -- SNMP Error:
> Error decoding response PDU:
>   Expected sequence tag 48, got 78
>     %{%i%s%*{%i%i%i%{% <at> 
>       ^
> SNMPv2c_Session (remote host: "10.236.71.1" [10.236.71.1].161)
>                    community: "public"  
>                   request ID: 2119154287
>                  PDU bufsize: 8000 bytes
>                      timeout: 10s
>                      retries: 10
>                      backoff: 1)
>  at /usr/bin/../lib/mrtg2/SNMP_util.pm line 492
> 2010-12-14 07:29:32 -- Use of uninitialized value in string ne at
> /usr/bin/../lib/mrtg2/SNMP_Session.pm line 870.

I'm monitoring about 750 hosts, most are Linux boxes running either
ucd-snmpd-4.2.3 or net-snmpd-5.4.1+ - these errors occur for both kinds.

Any ideas what these are and/or how to get rid of them?

  Regards, Danilo

Re: MRTG errors

Which version of MRTG are you using?  Try upgrading to 2.16.4, and make sure you're not using an older BER.pm
or SNMP_session.pm from a previous version.

Steve

Steve Shipway
University of Auckland ITS
UNIX Systems Design Lead
s.shipway <at> auckland.ac.nz
Ph: +64 9 373 7599 ext 86487

________________________________________
From: mrtg-bounces+s.shipway=auckland.ac.nz <at> lists.oetiker.ch
[mrtg-bounces+s.shipway=auckland.ac.nz <at> lists.oetiker.ch] on behalf of Danilo Godec [danilo.godec <at> agenda.si]

I get a lot of errors resembling this:

> 2010-12-14 07:29:32 -- SNMP Error:
> Error decoding response PDU:
>   Expected sequence tag 48, got 78
>     %{%i%s%*{%i%i%i%{% <at> 
>       ^

Re: MRTG errors

 On 12/14/2010 09:21 AM, Steve Shipway wrote:
> Which version of MRTG are you using?  Try upgrading to 2.16.4, and make sure you're not using an older BER.pm
or SNMP_session.pm from a previous version.

That's the version I'm using. I compiled it myself on OpenSuSE 10.3
x86_64, based on some other OpenSuSE SRPM.

There were two 'BER.pm' files on the system - one belonging to 'mrtg'
(/usr/lib/mrtg2/BER.pm) and the other belonging to 'smokeping'
(/usr/lib/perl5/vendor_perl/5.8.8/BER.pm). I removed 'smokeping' as
we're not using it, restarted 'mrtg' but it is still the same.

As far as I can tell the remaining 'BER.pm' and 'SNMP_session.pm' files
are correct and belong to 2.6.4.

   Regards, Danilo

> Steve
>
> Steve Shipway
> University of Auckland ITS
> UNIX Systems Design Lead
> s.shipway <at> auckland.ac.nz
> Ph: +64 9 373 7599 ext 86487
>
>
> ________________________________________
> From: mrtg-bounces+s.shipway=auckland.ac.nz <at> lists.oetiker.ch
[mrtg-bounces+s.shipway=auckland.ac.nz <at> lists.oetiker.ch] on behalf of Danilo Godec [danilo.godec <at> agenda.si]
>

Re: MRTG errors

Hi Danilo,

Today Danilo Godec wrote:

>  On 12/14/2010 09:21 AM, Steve Shipway wrote:
> > Which version of MRTG are you using?  Try upgrading to 2.16.4, and make sure you're not using an older
BER.pm or SNMP_session.pm from a previous version.
>
> That's the version I'm using. I compiled it myself on OpenSuSE 10.3
> x86_64, based on some other OpenSuSE SRPM.
>
> There were two 'BER.pm' files on the system - one belonging to 'mrtg'
> (/usr/lib/mrtg2/BER.pm) and the other belonging to 'smokeping'
> (/usr/lib/perl5/vendor_perl/5.8.8/BER.pm). I removed 'smokeping' as
> we're not using it, restarted 'mrtg' but it is still the same.
>
> As far as I can tell the remaining 'BER.pm' and 'SNMP_session.pm' files
> are correct and belong to 2.6.4.

I have asked simon leinen (the author of SNMP_session) about this
and he sais that the erros does not look anything like he has ever
seen ... he suggest to use packet capture (wireshark) to identify
the offending packet.

cheers
tobi

>    Regards, Danilo
>
>

Converting mebibytes

Hi,

we use a Panda GateDefender Integra for which I want to monitor the  
traffic of each port (using mrtg 2.16.2-3 (Debian Lenny)).

I don´t get a response for "interfaces.ifTable.ifEntry.ifInOctets" (or the  
corresponding OID), so I have to use "1.3.6.1.4.1.23473.1.1.2.2.3.1.1.4".  
Unfortunately, these values are presented in mebibytes and as these grow  
slowly I don´t get a graph in MRTG.

I read about multiplying the received values but that apparently can only  
use another OID and I don´t think there will be any OID that has the  
needed number and that also doesn´t change over time.

How can I solve this problem? Will there be the possibility to use numbers  
instead of OIDs for multiplying in a future version of MRTG?

Lars

_______________________________________________
mrtg mailing list
mrtg <at> lists.oetiker.ch
https://lists.oetiker.ch/cgi-bin/listinfo/mrtg

Re: MRTG errors

On 2010-12-14 10:32, Danilo Godec wrote:
>  On 12/14/2010 09:21 AM, Steve Shipway wrote:
>> Which version of MRTG are you using?  Try upgrading to 2.16.4, and make sure you're not using an older
BER.pm or SNMP_session.pm from a previous version.

> As far as I can tell the remaining 'BER.pm' and 'SNMP_session.pm' files
> are correct and belong to 2.6.4.

Just to be a bore: It _is_ a typo, that you say 'belong to 2.6.4, isn't it?

Jan Ferré

--

-- 
Jan Ferré - UNI-C Netdrift - DTU Bygning 304, DK-2800 Lyngby
tlf. +45 3587 8935, cell. +45 2160 8012, fax. +45 3587 8990
EAN-no.: 5798000555198

Expanded scale

_______________________________________________
mrtg mailing list
mrtg <at> lists.oetiker.ch
https://lists.oetiker.ch/cgi-bin/listinfo/mrtg

Re: MRTG errors

On 14.12.2010 10:46, Tobias Oetiker wrote:
> Hi Danilo,
>
> Today Danilo Godec wrote:
>
>>  On 12/14/2010 09:21 AM, Steve Shipway wrote:
>>> Which version of MRTG are you using?  Try upgrading to 2.16.4, and make sure you're not using an older
BER.pm or SNMP_session.pm from a previous version.
>> That's the version I'm using. I compiled it myself on OpenSuSE 10.3
>> x86_64, based on some other OpenSuSE SRPM.
>>
>> There were two 'BER.pm' files on the system - one belonging to 'mrtg'
>> (/usr/lib/mrtg2/BER.pm) and the other belonging to 'smokeping'
>> (/usr/lib/perl5/vendor_perl/5.8.8/BER.pm). I removed 'smokeping' as
>> we're not using it, restarted 'mrtg' but it is still the same.
>>
>> As far as I can tell the remaining 'BER.pm' and 'SNMP_session.pm' files
>> are correct and belong to 2.6.4.
> I have asked simon leinen (the author of SNMP_session) about this
> and he sais that the erros does not look anything like he has ever
> seen ... he suggest to use packet capture (wireshark) to identify
> the offending packet.
>
> cheers
> tobi

OK, here it is:

> 2010-12-14 18:54:48 -- SNMP Error:
> Error decoding response PDU:
>   Expected sequence tag 48, got 78
>     %{%i%s%*{%i%i%i%{% <at> 
>       ^
> SNMPv2c_Session (remote host: "10.224.104.1" [10.224.104.1].161)
>                    community: "lancom"
>                   request ID: 1270453324
>                  PDU bufsize: 8000 bytes
>                      timeout: 10s
>                      retries: 10
>                      backoff: 1)
>  at /usr/bin/../lib/mrtg2/SNMP_util.pm line 492

I cought it with 'tcpdump -s0' and opened the resulting file in Wireshark.

I found the 'offending' request ID and compared it to the next one (that
didn't show an error) - but I can't see a difference...

Attached is a .pcap file of these two request ID's...

  Danilo

_______________________________________________
mrtg mailing list
mrtg <at> lists.oetiker.ch
https://lists.oetiker.ch/cgi-bin/listinfo/mrtg

Re: Converting mebibytes

>I read about multiplying the received values but that apparently can only
>use another OID and I don´t think there will be any OID that has the
>needed number and that also doesn´t change over time.
>
>How can I solve this problem? Will there be the possibility to use numbers
>instead of OIDs for multiplying in a future version of MRTG?

You can already use constants in an MRTG Target definition.

e.g:

Target[xxx]: oid1&oid2:community <at> device * 1024

if the OIDs return values in k.  Note the required spaces around the operations and constants.  You can even
use brackets and +-*/ and so on.

Note though that SNMP generally returns integers so this may not help in your specific issue

It is generally Best Practice to use calculations on the Target line to convert the returned data to SI units
(IE, to bytes rather than leaving it as K) so that the stored values are in SI units; then, use the Factor[]
directive to make any display-time changes (eg, divide by 4096 if you want the value displayed in Blocks)

Steve

Steve Shipway
University of Auckland ITS
UNIX Systems Design Lead
s.shipway <at> auckland.ac.nz
Ph: +64 9 373 7599 ext 86487

Re: Expanded scale

_______________________________________________
mrtg mailing list
mrtg <at> lists.oetiker.ch
https://lists.oetiker.ch/cgi-bin/listinfo/mrtg