Anyone ever have issues with the bonding driver? I want to see if this is a
mis-configuration or a known issue before I dive into the driver source code.
I'm trying to bond two interfaces, eth0 and eth1, using active-backup mode (1).
My problem is that if the first port added to the bonding driver (eth0, in my
case,) is down (i.e. not plugged into the switch,) the driver will NOT use eth1
as the active link when it is added (eth1 is plugged in.)
To my way of thinking, this is a problem. It should use the first active
interface it finds in all cases. To have a good link and not use it is a bug,
I think.
Here are the particulars of my setup:
Kernel : 2.6.16 (this is for an embedded application, and no, I cannot
change kernels.)
Mod Load : /bin/mount -t sysfs sysfs /sys
/sbin/insmod /ASE/bonding.ko max_bonds=4 miimon=1000 use_carrier=1
echo balance-rr >/sys/class/net/bond0/bonding/mode
echo balance-rr >/sys/class/net/bond1/bonding/mode
echo active-backup >/sys/class/net/bond2/bonding/mode
echo active-backup >/sys/class/net/bond3/bonding/mode
ifcfg_bond2 : DEVICE=bond2
IPADDR=192.168.242.16
NETMASK=255.255.255.0
BROADCAST=192.168.242.255
NETWORK=192.168.242.0
ifcfg_eth0 : NAME="Rear Ethernet 1"
DEVICE=eth0
SLAVE=yes
MASTER=bond2
ifcfg_eth1 : NAME="Rear Ethernet 2"
DEVICE=eth1
SLAVE=yes
MASTER=bond2
So. my questions are:
1) Is the above configuration correct?
2) Is the behavior I'm seeing normal?
3) Is this a known bug?
-- Robert King
Sevis Systems
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to [EMAIL PROTECTED]
Please read the FAQ at http://kernelnewbies.org/FAQ
