cause in previous version of net-snmp, there are some uggly bugs like:
on 64bits arch, the IF-MIB::ifInOctets.2 return a 64bits integer but must return
a 32bits integer and if you request this OID on a 32bits arch we have this
snmpwalk -c public -v1 Host IF-MIB::ifInOctets.2
IF-MIB::ifInOctets.2 = Counter32: 0
and on 64bits we have
IF-MIB::ifInOctets.2 = Counter32: 118735921106
another problem is the CPU Counter
not works at all before 5.3-dev4 version.
can we hope an upgrade of gentoo portage and add this?
i write a bug report to net-snmp and the staff tell me this bug is close since
5.3-dev4
Reproducible: Always
Steps to Reproduce:
*on 32 bits host to 64 bits host*
1) uname -msp
2) $snmpwalk -c public -v1 Host IF-MIB::ifInOctets.2
*on 64 bits host to 64 bits host*
3) uname -msp
4) snmpwalk -c public -v1 Host IF-MIB::ifInOctets.2
*for the CPU Usage*
5) snmpget host UCD-SNMP-MIB::ssCpuUser.0
6) on host "$ yes >/dev/null"
7) snmpget host UCD-SNMP-MIB::ssCpuUser.0
and the value never change id snmpd is 5.2.1.X or 5.3.20050624, but it s ok with
5.3-dev4
Actual Results:
1) Linux i686 Intel(R) Pentium(R) 4 CPU 3.00GHz
2) IF-MIB::ifInOctets.2 = Counter32: 0
3) Linux x86_64 AMD Opteron (tm) Processor 844
4) IF-MIB::ifInOctets.2 = Counter32: 118735921106
5) UCD-SNMP-MIB::ssCpuUser.0 = INTEGER: 6
6) -
6) a value not 100%/(number of CPU)
Expected Results:
2)IF-MIB::ifInOctets.2 = Counter32: 118735921106
6) UCD-SNMP-MIB::ssCpuUser.0 = INTEGER: 25 (cause 4 CPU)
it s not a gentoo bug but a net-snmp bug, the net-snmp staff tell me to upgrade
to the net-snmp 5.3-dev4 version, and i which to have this in a few in portage.
for now i must install on my server outside portage or work with old version.
