Cacti monitoring of bonded interfaces

Post support questions that directly relate to Linux/Unix operating systems.

Moderators: Developers, Moderators

Post Reply
nsanders
Posts: 21
Joined: Thu Oct 19, 2006 2:01 pm

Cacti monitoring of bonded interfaces

Post by nsanders »

Is there a problem with cacti or snmp in general when monitoring bonded interfaces under linux? On multiple machines we have eth0 and eth1 bonded into bond0 using 802.3ad bonding. This should result in a balanced load (I know it's not 50/50) but the odd thing is that the bond0 interface does not equal eth0 + eth1.. On some systems you don't even see similar traffic on the graphs.

Such as this:

Image


First off, bond0 (the top graph) does not equal eth0 + eth1. It's only 60Mb, the same as eth1...

Around 7:30am you see a 60mb inbound spike on eth1, why is this not on bond0?

At 4:15am you see a 60mb outbound spike on bond0.. Um, I don't see 60Mb of traffic on either eth0 or eth1 at this time. There is 20Mb on both, but 20 + 20 = 40, not 60...

Also, why are my graphs on eth0 half missing? You see multiple large spikes out outbound traffic that just appear/vanish without a completed graph. I get this a lot.

Is this a cacti problem or snmp?
nsanders
Posts: 21
Joined: Thu Oct 19, 2006 2:01 pm

Post by nsanders »

Oh, as a follow up here is the information of bandwidth usage provided by 'ifconfig'

bond0:
RX bytes:2013712826273 (1.8 TiB) TX bytes:30989991334 (28.8 GiB)

eth0:
RX bytes:1818309221041 (1.6 TiB) TX bytes:25900895497 (24.1 GiB)

eth1:
RX bytes:195403605232 (181.9 GiB) TX bytes:5089095837 (4.7 GiB)

As you can see everything makes sense. Eth0:TX + Eth1:TX = bond0:TX.

So there is for sure a problem with the sums on Cacti (or snmp).
nsanders
Posts: 21
Joined: Thu Oct 19, 2006 2:01 pm

Post by nsanders »

Nobody has any ideas about this?
nsanders
Posts: 21
Joined: Thu Oct 19, 2006 2:01 pm

Post by nsanders »

I seem to have stumped everyone.
User avatar
pestilence
Cacti User
Posts: 207
Joined: Fri Jul 25, 2003 10:37 am
Location: Athens/Greece
Contact:

Post by pestilence »

This has nothing to do with cacti summing, i suspect this might be a SNMP issue, could you check your SNMP output and post it here?
/* * Oops. The kernel tried to access some bad page. We'll have to * terminate things with extreme prejudice. */ die_if_kernel("Oops", regs, error_code);
nsanders
Posts: 21
Joined: Thu Oct 19, 2006 2:01 pm

Post by nsanders »

What information exactly do you want? You just want me to snmpwalk the host and paste the relevant info or is there some where in Cacti I can grab that easily.
nsanders
Posts: 21
Joined: Thu Oct 19, 2006 2:01 pm

Post by nsanders »

Anybody?
maxw
Posts: 8
Joined: Fri May 28, 2010 4:59 am

Re: Cacti monitoring of bonded interfaces

Post by maxw »

I'd like to bump this thread in the hope of fixing this issue. I see it on Windows and Linux. Here is my thread in the general section:
http://forums.cacti.net/viewtopic.php?f=21&t=40831
Post Reply

Who is online

Users browsing this forum: No registered users and 0 guests