CentOS

I have a couple of (patched & updated) CentOS 5 boxes here and ever since I setup ALB bonding on the two interfaces, my log file is getting slammed with the following SNMP errors. Anybody have a clue as to why this is happening?

Jun 4 09:47:15 webbox2 snmpd[13664]: error on subcontainer '' insert (-1)
Jun 4 09:47:45 webbox2 snmpd[13664]: error on subcontainer 'ia_addr' insert (-1)

Here's my ifconfig and I don't see any problems here.... The only thing that I can think of is that bond0 uses the MAC addy of eth1, but this is the way it must work correct?

[root@webbox2 ~]# ifconfig
bond0 Link encap:Ethernet HWaddr 00:0B:CD:1C:E6:D5
inet addr:10.110.120.101 Bcast:10.110.120.255 Mask:255.255.255.0
inet6 addr: fe80::20b:cdff:fe1c:e6d5/64 Scope:Link
UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
RX packets:8343346 errors:0 dropped:0 overruns:0 frame:0
TX packets:10268849 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:1178946879 (1.0 GiB) TX bytes:893576188 (852.1 MiB)

eth0 Link encap:Ethernet HWaddr 00:0B:CD:1C:E6:D7
inet6 addr: fe80::20b:cdff:fe1c:e6d5/64 Scope:Link
UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
RX packets:4488633 errors:0 dropped:0 overruns:0 frame:0
TX packets:5405394 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:610419594 (582.1 MiB) TX bytes:434368673 (414.2 MiB)
Interrupt:177

eth1 Link encap:Ethernet HWaddr 00:0B:CD:1C:E6:D5
inet6 addr: fe80::20b:cdff:fe1c:e6d7/64 Scope:Link
UP BROADCAST RUNNING SLAVE MULTICAST MTU:1500 Metric:1
RX packets:3854713 errors:0 dropped:0 overruns:0 frame:0
TX packets:4863455 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:568527285 (542.1 MiB) TX bytes:459207515 (437.9 MiB)
Interrupt:185

Anyone seeing this at all? .... bump.

This is a known issue with RHEL5 and its derivatives (according to some googling) its expected to be fixed in an update.

So in short you can have snmp, or bonding but not both.

Hi,

I have this error with bonded nics on a Dell 2850 running latest Centos5 with virtualisation support enabled but not used yet

I have deleted the bonding from /etc/sysconfig/networking-scripts and also /etc/modprobe.conf

I did not look in the logs before enabling bonding

so I am guessing the problem is either not bonding related - or my sever has some memory of bonding I have not removed

any suggestions appreciated

Stuart

HI,

Got around the problem - now running with bonding and SNMP on 64bit

System appears to function as good as when on Centos4

copied /etc/init.d/snmpd from an existing system

OPTIONS="-LS c d -LF ! /dev/null -p /var/run/snmpd -a"

Stuart

Guys, I had a similar case (centos 5, net-snmp-5.3.1-14.0.1.el5, 2 bonding devices).

I got the same log messages (error inserting ...)

The reason appeared to be in the ipv6 support of the snmp. I disabled loading the ipv6 module in modprobe.conf, fixed my hosts file so that no ipv6 addresses are assigned to localhost, added a "IPV6INIT=no" line to the /etc/sysconfig/network-scripts/ifcfg-* and now everything works fine, no more annoying errors in the log.

Now the only error message I have is "snmpd: cannot open /proc/net/snmp6 ..."

Which can be fixed AFAIK by editing the snmpd configuration.

Anyway, bonding and snmp are possible on rhel5, it's the IPv6 support that causes the problem...

two eth? in a bonding device? or two bonding devices? thanks.