RE: Re: Initial release of the Bacula SNMP Subagent - v1.2

On Fri, 2005-10-21 at 16:54, Darren Gamble wrote:

> > Mostly unrelated question: do you happen to know who to
> > complain to about Linux x86_64 interface traffic stats
> > from net-snmp being incorrect as reported by cacti?
> > The target is Centos3.x (should be the same as RHEL3) with
> > cacti running on a 32-bit Centos4 box.  After a reboot the
> > values are correct for a while, then go crazy, showing
> > 100 Meg traffic (the interface limit) most of the time.
> 
> Cacti (usually) just uses net-snmp's agent to get the data, so the issue is
> probably "garbage in, garbage out".  I had seen problems a few months ago
> (and posted here) where a 64-bit net-snmp agent will return a
> greater-than-Counter32 value for a Counter32 value, which causes funny
> results on the net-snmp 32-bit client side, especially when a snmpwalk is
> being done. Interestingly enough, if the client was also 64-bit, then it
> would "properly" report the bad value as it was transmitted.  I don't see
> this problem anymore with recent versions of net-snmp.

What's supposed to happen when the value doesn't fit in a Counter32?
On a box where:
IF-MIB::ifDescr.2 = STRING: eth0
IF-MIB::ifDescr.3 = STRING: eth1
snmpwalk says:
IF-MIB::ifOutOctets.2 = Counter32: 149182455
IF-MIB::ifOutOctets.3 = Counter32: 0
but ifconfig shows:
eth0   TX bytes:149179729 (142.2 Mb)
eth1   TX bytes:33033390619 (31503.0 Mb)

This is an older version but with RH based systems it's hard to tell
what fixes have been backported to old version numbers.

-- 
  Les Mikesell
   [email protected]




-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl