I have a related issue... we don't really want to use the primary option. When there is a failure it should just stick to that path. However we do want it to be predicatable on startup as to which interface it is going to use. Is there any way of specifying which interface should be the current active slave.

Regards

________________________________________
From: rhelv5-list-bounces@(protected)
Sent: 01 April 2008 17:12
To: Red Hat Enterprise Linux 5 (Tikanga) discussion mailing-list
Subject: [rhelv5-list] bonding problems

Hi
I have a simple setup in my office on which I doing some bonding testing and have run into a small problem

I have an HP server with a onboard broacom network card and an pci intel network card.

This uses bnx2(broadcom) and e1000 (intel) drivers.

The bonding settings are:
mode=active-backup
miimon=5
primary=eth0

eth0 is plugged into a Cisco 2950 called A
eth2 is plugged into a separate Cisco 2950 called B
Both switches are connected to one another on port 24.
There is a laptop plugged into switch A or B, results are the same.

I then do the following test:
ping the server continuosly

Shutdown eth0 on the switch
The bond fails perfectly to eth2.
Wait for one minute
Enable eth0 again.

Watching /proc/net/bonding/bond0 this shows that it immediately  makes eth0 active again.

However immediately upon doing that the ping starts to fail for about 30 seconds and then after 30 seconds starts working again.

When I shutdown eth0 I see a arp reply packet on eth2 followed by multiple arp who-has packets
When I bring eth0 up again I see a arp reply packet on eth0

I have no idea why this failure is happening... is this a switch config problem or a bondig problem...

Regards


_______________________________________________
rhelv5-list mailing list
rhelv5-list@(protected)
https://www.redhat.com/mailman/listinfo/rhelv5-list