[SOLVED] Network Bonding

[MyKroFt]

I have a supermicro with 2 gig nics - I have setup bonding and have configured to the 2 ports on my smart netgear switch - and have set bonding to yes and mode to balance -rr - this seems to be working fine and is using both nics together finally pulling my entire avail bandwidth when necessary.

 

my quest is looking at syslog I am getting the following and am not sure what to even look for:

 

Aug 25 13:04:43 Tower kernel: br0: received packet on bond0 with own address as source address

 

any ideas?

 

Thanks

Myk

 

[bonienl]

I have a supermicro with 2 gig nics - I have setup bonding and have configured to the 2 ports on my smart netgear switch - and have set bonding to yes and mode to balance -rr - this seems to be working fine and is using both nics together finally pulling my entire avail bandwidth when necessary.

 

my quest is looking at syslog I am getting the following and am not sure what to even look for:

 

Aug 25 13:04:43 Tower kernel: br0: received packet on bond0 with own address as source address

 

any ideas?

 

Thanks

Myk

 

Your switch needs to support port channeling to use this mode, otherwise it will treat both interfaces as single connections which may result in a loop.

 

This is indicated by the syslog message, unRAID sees its own message coming back.

 

If you want increased throughput without special functionality on the switch, you can use either mode balance-tlb or balance-alb (one way or two way balancing).

 

[MyKroFt]

Its a netgear GS748T, and I have the 2 ports in what they call a LAG group - all other 46 ports are indv - I am 99.99% positive I have this set correctly as when I didnt and plugged in the 2nd network cable - the whole network came to a stop and I could see the looping happening.  Unplug the 2nd cable and the network could come back to normal.

 

Both cables plugged in right now and network performance is as expected except for the br0 message.  I needed to set the br0 so my 2 VMs would get independent ips from my dhcp server.

 

I am watching the switch monitoring interface and there are no errors anywhere happening on the network....

 

So I am still not sure where to look for this error.

 

Myk

 

[MyKroFt]

Here are the two router config screen and how they are setup

 

Myk

[bonienl]

Its a netgear GS748T, and I have the 2 ports in what they call a LAG group - all other 46 ports are indv - I am 99.99% positive I have this set correctly as when I didnt and plugged in the 2nd network cable - the whole network came to a stop and I could see the looping happening.  Unplug the 2nd cable and the network could come back to normal.

 

Both cables plugged in right now and network performance is as expected except for the br0 message.  I needed to set the br0 so my 2 VMs would get independent ips from my dhcp server.

 

I am watching the switch monitoring interface and there are no errors anywhere happening on the network....

 

So I am still not sure where to look for this error.

 

Myk

 

If your switch supports link aggregation with LACP then I would advice to use mode 802.3ad (4) on your unRAID system. This has the advantage that an active protocol monitors the links in the bundle and any change would be automatically detected at both ends, keeping them in sync.

 

[MyKroFt]

Changing to 802.3ad stopped the br0 messages

 

Thanks!

 

Myk