SNMP issues with GbE2c

[Thomas Joebstl]

Hello,

got a C7000 with 4 GbE2c's handed down to have a look at the individual blades traffic and generate some stats.
The GbE2c's are running firmware 5.1.3 and I'm using cacti to get some nice graphs, but graphing the GbE2c's traffic doesnt really work that well.

Apparently the 64 bit traffic counters roll over at 2^32 bit as well so the graphs top out at around 100Mbit/s instead of going nearly to the port maximum when running iperf at 940Mbit/s for even half an hour...

snmpwalk shows this quite clearly:

# date ; snmpget -v2c -c xxxx 10.160.1.213 .1.3.6.1.2.1.31.1.1.1.10.277
Tue May 17 21:27:26 CEST 2011
IF-MIB::ifHCOutOctets.277 = Counter64: 4288293561

# date ; snmpget -v2c -c xxxx 10.160.1.213 .1.3.6.1.2.1.31.1.1.1.10.277
Tue May 17 21:27:27 CEST 2011
IF-MIB::ifHCOutOctets.277 = Counter64: 61872149

Now I'm wondering if
- I'm the only one seeing this weird behaviour?
- the Blade-Center admins have to enable something obscure on the switches to get the 64 bit counters to actually work as 64 bit counters?
- if I (and cacti) am using the wrong oid to query the 64 bit traffic counters?
- or this is really a firmware bug?

Regards,
Thomas

[Gerhard Roets · Accepted solution]

Hi Thomas

I can help you with one answer according to the MIB.
roetsroe@linux-mhe8:~> snmptranslate -Td -m all 1.3.6.1.2.1.31.1.1.1.10.277
IF-MIB::ifHCOutOctets.277
ifHCOutOctets OBJECT-TYPE
-- FROM IF-MIB
SYNTAX Counter64
MAX-ACCESS read-only
STATUS current
DESCRIPTION "The total number of octets transmitted out of the
interface, including framing characters. This object is a
64-bit version of ifOutOctets.

Discontinuities in the value of this counter can occur at
re-initialization of the management system, and at other
times as indicated by the value of
ifCounterDiscontinuityTime."
::= { iso(1) org(3) dod(6) internet(1) mgmt(2) mib-2(1) ifMIB(31) ifMIBObjects(1) ifXTable(1) ifXEntry(1) ifHCOutOctets(10) 277 }

It is a 64 bit counter.
It should wrap roughly at 3.68934881474e+19.

I would suggest looking at support with it.

HTH
Gerhard

[Thomas Joebstl]

Thanks, just did that.
I've verified by having a look at the port stats the switch reports via cli and it shows the same behaviour so I guess this is really a firmware bug.

[AviKlempner]

Well,

Long time passed from original message, but I face the same issue with firmeware 5.1.2.

Was this ever fixed? having problems with cacti monitoring the interface mbps (spped)

Thanks  & Regards,

Avi