Outbound (ifoutoctets) data is wrong for some of the routers

Post general support questions here that do not specifically fall into the Linux or Windows categories.

Moderators: Developers, Moderators

Post Reply
User avatar
fmendez75
Posts: 7
Joined: Tue Dec 05, 2006 1:25 pm
Location: NYC, NY

Outbound (ifoutoctets) data is wrong for some of the routers

Post by fmendez75 »

Hi,
Please post in the right forum if this is the wrong one.

Problem:
I've noticed that for some interfaces and for some routers, my graphs are not showing the right outbound current data.
I have really tried to check through the templates and nothing has changed to the best of my knowledge. I check the interfaces.xml file and it looks ok. The data template and the graph template which I've used to graph the interfaces are the same as when I installed Cacti. This is why I think this is a weird case.

Have anybody encountered the same problem? or perhaps can show me the way to fix this.

I've tried to see if the CDEFs were wrong, but it looks fine.

When I do a snmpwalk on the command line, I get the results in bytes and they are the correct ones.

Please let me know if you need any other information that might help come up with a solution.

Thank you all,
Cacti Version - 0.8.6h
Poller Type - CMD.php
Server Info - Linux 2.6.16-2-686-smp
Web Server - Apache/2.0.55 (Debian) PHP/4.4.2-1.1
PHP - 4.4.2-1.1
PHP Extensions - zip, yp, xmlrpc, xml, wddx, tokenizer, sysvshm, sysvsem, sysvmsg, standard, sockets, shmop, session, posix, overload, mime_magic, mbstring, iconv, gettext, ftp, filepro, exif, dbx, dba, ctype, calendar, bz2, bcmath, zlib, pcre, openssl, apache2handler, mysql, snmp
MySQL - 5.0.24a-Debian_3-log
RRDTool - 1.2.11
SNMP - 5.2.3
Plugins
  • Thresholds (thold - v0.3.0)
    Device Monitoring (monitor - v0.7)
    Network Discovery (discovery - v0.6)
    Report Creator (reports - v0.1b)
    Network Tools (tools - v0.2)
    FlowView (flowview - v0.3)
    Cycle Graphs (Cycle Graphs - v0.3)
    Host Info (hostinfo - v0.1)
    Update Checker (update - v0.3)
User avatar
gandalf
Developer
Posts: 22383
Joined: Thu Dec 02, 2004 2:46 am
Location: Muenster, Germany
Contact:

Post by gandalf »

SNMP will never report the correct outbound data in bytes for the current interval but as a COUNTER it will report all bytes transferred since reboot. So if comparing snmpwalk ... interfaces output to current numbers in cacti, they will almost always not match.
If I misunderstood, please provide more infos
Reinhard
User avatar
fmendez75
Posts: 7
Joined: Tue Dec 05, 2006 1:25 pm
Location: NYC, NY

Post by fmendez75 »

Correct!,
That I understand, but in my case the outbound is way off.
Let see if I can make this clearer.
The data below was taken within the 5 minutes polling period.
Please compare the output for the outbound interface.

The graph should be in terms of bits per second.

Also the following is the output gathered through the command line.

snmpwalk -v 1 -c xxxxxxx host.domain.net ifoutoctets
IF-MIB::ifOutOctets.2 = Counter32: 2555023492

(I copied and pasted the output, and also just copied the interface in question.)
Attachments
wrong_outbound_data.JPG
wrong_outbound_data.JPG (120.51 KiB) Viewed 1942 times
User avatar
gandalf
Developer
Posts: 22383
Joined: Thu Dec 02, 2004 2:46 am
Location: Muenster, Germany
Contact:

Post by gandalf »

Ok, better. There are two possible issues. The first is related to clipping data when exceeding the given "Maximum" for tha data source. Visit the data template for the traffic graph and check both datasource's maximum. Perhaps you need to increase it.
Please pay attention: You will have to rrdtool tune --maximum alle datasources of all existing rrd files belonging to that data template as well!
The second issue is related to 32 bit counter wrap. This may occur on heavily loaded links if polling every 5 min. In this case, you will have to switch over to 64 bit counters. This will require using SNMP V2c (cacti knows this as snmp version 2). In this case, you will have to create a new set of graphs/rrd files (select the 64 bit ones from the dropdown when creatinbg traffic graphs).
Reinhard
User avatar
fmendez75
Posts: 7
Joined: Tue Dec 05, 2006 1:25 pm
Location: NYC, NY

Post by fmendez75 »

I thought about the datasource's max and I did modified to make sure it was correct.
I tried to do the rrdtool tune command and the --maximum was not taken.
however this command worked.

rrdtool tune host_in_id.rrd max

I just don't know if that command did what it was supposed to do.

Step 2 fixed the problem.
I went into my Device that was giving me trouble. Select Version 2 for my SMNP Version.
Then I created a new graph using the 64 Bit Counter, and it worked.

Now, the question is will I have to recreate all of my graphs again for that device?
Because I noticed that the old graph is still showing the wrong number, even though that the device SNMP version has changed. I guess the secret lies here in the 64 Bit Counter graph template.

Thank you very much for your help and and other ideas I could do for my current graphs?
Attachments
right output.JPG
right output.JPG (25.09 KiB) Viewed 1923 times
User avatar
gandalf
Developer
Posts: 22383
Joined: Thu Dec 02, 2004 2:46 am
Location: Muenster, Germany
Contact:

Post by gandalf »

You current graphs will be useless, unless you're sure when the counters did not wrap. In general, always when there is a sharp edge, the counter will have wrapped. If this is ok for you, you may simply rename the old rrd file to the new one (increasing the maximum doens't change the data representation of existing data points in the rrd file), so that it will get filled with larger values. This assumes, that the maximum is set correctly. The syntax for rrdtool tune is rrdtool tune <filename> --maximum <ds-name>:<max> where max is set to 0 to suppress any maximum. Check using rrdtool info <rrd file>.
If you consider renaming, make sure to have backups available!
Reinhard
Post Reply

Who is online

Users browsing this forum: No registered users and 7 guests