RedHat 5.4 bonding issue
 

I'm trying to do some network bonding in RHEL 5.4, and I'm not seeing the results I expect. I have set up a number of servers using mode 1 with no issue, but I need to configure a new server differently.

bond1 is created from eth1 & eth2 bonded together using mode 802.3ad with layer2 lacp

bond0 is created from bond1 & eth3 using mode active-backup with bond1 as the primary interface

When I bring the network services up, all traffic is passing over eth3 instead of bond1, which is the primary.

Below are the modprobe.conf * ifcfg-* files involved. Any help is appreciated.


My modprobe.conf:
alias eth0 e1000
alias eth1 e1000
alias eth2 e1000
alias eth3 e1000
alias scsi_hostadapter mptbase
alias scsi_hostadapter1 mptsas
alias scsi_hostadapter2 qla2xxx
alias scsi_hostadapter3 usb-storageDEVICE=bond0
alias bond0 bonding
alias bond1 bonding

ifcfg-bond0:
ONBOOT=yes
BOOTPROTO=none
IPADDR=10.10.10.10
NETMASK=255.255.248.0
GATEWAY=10.10.10.1
TYPE=Ethernet
USERCTL=no
IPV6INIT=no
PEERDNS=yes
BONDING_OPTS="mode=active-backup miimon=100 downdelay=200 updelay=200 primary=bond1"

ifcfg-bond1:
DEVICE=bond1
ONBOOT=yes
BOOTPROTO=none
MASTER=bond0
SLAVE=yes
USERCTL=no
BONDING_OPTS="mode=802.3ad xmit_hash_policy=layer2 miimon=100 downdelay=200 updelay=200"
PRIMARY=yes

ifcfg-eth1:
# Intel Corporation 82546EB Gigabit Ethernet Controller (Copper)
DEVICE=eth1
HWADDR=00:14:4F:40:68:6D
ONBOOT=yes
BOOTPROTO=none
MASTER=bond1
SLAVE=yes
USERCTL=no

ifcfg-eth2
# Intel Corporation 82546EB Gigabit Ethernet Controller (Copper)
DEVICE=eth2
HWADDR=00:14:4F:40:68:6E
ONBOOT=yes
BOOTPROTO=none
MASTER=bond1
SLAVE=yes
USERCTL=no

ifcfg-eth3:
# Intel Corporation 82546EB Gigabit Ethernet Controller (Copper)
DEVICE=eth3
HWADDR=00:14:4F:40:68:6F
ONBOOT=yes
BOOTPROTO=none
MASTER=bond0
SLAVE=yes
USERCTL=no
PRIMARY=no

I think you will have to renamed the bond1 to something like zbond1.

I think I have seen this before and I battled with a bonding issue a while back.

I can't remember where I read a work-around for this, but try changing the name on the bond.

See if that works?

That fixed it. I renamed bond1 to aggr1 and had to hard code the MAC address for eth1 in ifcfg-aggr1. Now that shows up as the active interface in /proc/net/bonding/bond0:

cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v3.4.0 (October 7, 2008)

Bonding Mode: fault-tolerance (active-backup)
Primary Slave: aggr1
Currently Active Slave: aggr1
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 200
Down Delay (ms): 200

Slave Interface: aggr1
MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:14:4f:40:68:6d

Slave Interface: eth3
MII Status: up
Link Failure Count: 0
Permanent HW addr: 00:14:4f:40:68:6f

I am glad that fixed it, I read the same problem and the re-named worked.

I had a similar problem, however I had a brand new Dell R610 with 4 on board ports & 2 extra ports on a added in Intel Network card.

What was strange the eth0 kept pointing to the Intel card that was added on, I yanked the Intel card and the on-board Broadcom 1G ports worked perfectly.

Bonding nics appears to be simple in appearance, but it is very challenging working with different hardware or configurations on stand-alone servers.