[mrtg] ifNumbers Not Working for Low Speed Interfaces When MRTG Parallelized

[Gordon, Bruce]

Wow you're good!

forcing it to SNMPv1 worked - thanks

________________________________
From: Karyn Stump [mailto:karyn at calarts.edu]
Sent: Wednesday, June 09, 2010 4:22 PM
To: Gordon, Bruce
Subject: Re: [mrtg] ifNumbers Not Working for Low Speed Interfaces When MRTG Parallelized

Is it able to pull the uptime of the affected device in either case ?

Have you tried "Target[$]: :<community>@<device>:::::1~" just to see what happens ?

On Wed, Jun 9, 2010 at 1:02 PM, Gordon, Bruce <Bruce.Gordon at nrc-cnrc.gc.ca<mailto:Bruce.Gordon at nrc-cnrc.gc.ca>> wrote:
Same graphs all the time - those not using SNMPv2 counters

running MRTG from the command line against the individual config works
running MRTG from the script does not
running MRTG from the script using cron does not

________________________________
From: Karyn Stump [mailto:karyn at calarts.edu<mailto:karyn at calarts.edu>]
Sent: Wednesday, June 09, 2010 3:53 PM
To: Gordon, Bruce
Subject: Re: [mrtg] ifNumbers Not Working for Low Speed Interfaces When MRTG Parallelized

Same graphs all the time ? Different graphs at different times ? Can you tell what is not being interpolated correctly from the errors you are getting ? Check your environment. Usually if you can run a script from the command line and then not from cron (for example) it is just a difference in the environment.

On Wed, Jun 9, 2010 at 7:18 AM, Gordon, Bruce <Bruce.Gordon at nrc-cnrc.gc.ca<mailto:Bruce.Gordon at nrc-cnrc.gc.ca>> wrote:
Hello,

We've been using MRTG on Linux for years.  Since one of our sysadmins updated our MRTG to 2.16.2 we have noticed that some of our graphs are breaking.

We manually roll our own configs which we then run from a script that reads in a file that contains subdirectory names and the subdirectories contains the config file.  The script runs mrtg against that cfg file in the background.  A "wait" command at the end of the shell script.  In this manner we are able to parallelize the execution of the configs (one per device) so one unresponsive device (some can have hundreds of interfaces) does not wrap the monitoring interval.

I know it's old-school and RRD would solve all our problems but ... It works.  Or it did.

Now the majority of the devices polled have SNMP ver2 counters so we tend to put them at the head of the file with a "Target[$]: :<community>@<device>:::::2~" string then reset the Target string to "Target[$]: :<community>@<device>~" before the sections dealing with i/fs with non v2 counters.

What we now get is:

2010-06-08 15:13:27: WARNING: Can not determine ifNumber for <community>@<device>   ref: 'Name'     key: 'Et1/1'
2010-06-08 15:13:27: WARNING: Can not determine ifNumber for <community>@<device>   ref: 'Ip'       key: '192.168.1.1
2010-06-08 15:13:28: ERROR: Target[et1_1][_IN_] ' $target->[5]{$mode} ' did not eval into defined data
2010-06-08 15:13:28: ERROR: Target[et1_1][_OUT_] ' $target->[5]{$mode} ' did not eval into defined data
2010-06-08 15:13:28: ERROR: Target[tu1][_IN_] ' $target->[6]{$mode} ' did not eval into defined data
2010-06-08 15:13:28: ERROR: Target[tu1][_OUT_] ' $target->[6]{$mode} ' did not eval into defined data

When running the script.

The weird thing is when we run mrtg directly against the target (i.e. "mrtg mrtg.cfg") we get no error - i.e. it determines the ifNumber

Any ideas? (Other than spanking the sysadmin)

Regards,

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



--
Karyn Stump
Manager, Network and Server Operations
California Institute of the Arts




--
Karyn Stump
Manager, Network and Server Operations
California Institute of the Arts

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/mrtg/attachments/20100609/b8ec1431/attachment.htm