I have just upgraded to unraid6 with unraid as the host. My previous configuration was a debian server with unraid running as a xen guest, the server has 2 NICs so the debian dom0 bonded them into a bond0 in LACP mode. The switch is configured appropriately & server used DHCP, this all worked fine ... i.e. the physical setup supports this configuration without issues.

 

Switching to unraid I thought I would replicate that setup but it seems extremely unstable.

 

If I turn on bonding in the web ui and use DHCP, system can't obtain a lease.

If I switch to a static IP with a valid gateway & dns, bond0 is up but unuseable (e.g. can't reach anything via the network)

If I turn off bonding but turn on bridging, whole network seems unstable with frequent dropouts (especially when a VM is added and running, everything went v v v slow at that point)

If I turn on bonding and bridging then that seems to bring down the switch it is connected to so I suppose some sort of storm is going on somewhere

 

I have searched the forum but am unable to find a clear statement of what the correct configuration is here.

 

Are the few values in the network.cfg the only values to tweak or are there some hidden params in there?

 

Obviously I can drop back to a single NIC but am curious as to what a working bond0 configuration actually is, can anyone shed light on this? (NB: I have searched the forums but the various threads are quite opaque and/or refer to old unraid versions).

 

If I do drop back to a single NIC, do I need to physical pull the 2nd cable?

OK so now I notice the help mode in the GUI which says

 

Caution: if bonding is also not enabled, do not connect two or more ethernet ports to the same switch unless you have STP enabled and the switch supports STP (most consumer switches do not).

 

Doing so will cause an "ARP broadcast storm" and can bring down your entire network (unplugging all sever ethernet ports except one typically will restore your network).

 

I suppose that explains the extreme flakiness 

I have enabled STP on the switch and now br0 comes up ok.

 

Therefore my issue seems to boil down to 802.3ad doesn't seem to work (with hardware that is known to work) for me. My setup is

 

Enable bonding: Yes

Bonding mode:802.3ad (4)

Setup bridge:Yes

Bridge name:br0

Bridge enable STP:No

Bridge forward delay:0

Obtain IP address automatically:Yes

IP address:192.168.1.10

Network mask:255.255.255.0

Default gateway:192.168.1.1

Obtain DNS server address automatically:Yes

DNS server 1:192.168.1.1

DNS server 2:

DNS server 3:

 

if I disable the link aggregation group (LAG) in the switch then I, unsurprisingly, get a message as follows in syslog

 

bond0: Warning: No 802.3ad response from the link partner for any adapters in the bond

 

the links comes up ok though

 

if I assign the ports to the LAG in the switch then it fails to get a DHCP lease and gives up after 60s

 

if I switch to active-backup mode on the bond then it comes up ok with no obvious issues

 

Ideas welcome on what is wrong

 

for reference, the mobo is an ASRock Z87 Extreme6 and the NICs are

 

00:19.0 Ethernet controller: Intel Corporation Ethernet Connection I217-V (rev 05)

03:00.0 Ethernet controller: Intel Corporation I211 Gigabit Network Connection (rev 03)

 

the latter is eth0 and the former is eth1

I have the same motherboard and have never been able to get 802,3ad working.  I have a Netgear managed switch and when I set up dynamic bonding on unraid and enable LAG for the server ports on the switch, the network virtually stops.  I can barely get to the unraid webgui and didn't bother trying to get at the log.  I just gave up.

 

I use alb which does not need switch support and gives higher throughput and redundancy.

ok thanks, sounds like it's not worth bothering trying to debug what unraid is doing then

On 1/4/2016 at 5:15 PM, dlandon said:

I have the same motherboard and have never been able to get 802,3ad working.  I have a Netgear managed switch and when I set up dynamic bonding on unraid and enable LAG for the server ports on the switch, the network virtually stops.  I can barely get to the unraid webgui and didn't bother trying to get at the log.  I just gave up.

 

I use alb which does not need switch support and gives higher throughput and redundancy.

Same thing happened to me.  After doing the setup, the page barely loads...like some stuff.  I have no idea why this happens.