Working Link Aggregation Config RT-AC66U

[mrengles]

it's good to know there's alternate methods

and it certainly can't hurt to try

but from his post above

root@Gateway:~# cat /sys/class/net/bond0/bonding/slaves
vlan3 vlan4

so both vlan's are already slaves to bond0
but still show different aggregator ID regardless of this

Regardless of how I add the slaves still have the same issues. I'm still searching the web for a solution...

 

[RMerlin]

Not that it helps much now, but ifenslave is pretty much depreciated. Because DD-WRT provides a newer userland, I would guess you can use the ip link command instead of sysfs (a little easier to remember/type IMO) to set slaves.

ip link set vlan1 master bond0
ip link set vlan2 master bond0

Asus' copy of iproute2/kernel is too old to support this syntax, unfortunately.

Do you think it's just iproute2? Which version would be required?

Asus upgraded iproute2 for the ARM devices - might be because it was required to work with the 2.6.36 kernel those devices use.

 

[saintdev]

Do you think it's just iproute2? Which version would be required?

Asus upgraded iproute2 for the ARM devices - might be because it was required to work with the 2.6.36 kernel those devices use.

I suspect it requires kernel support as well, but I don't know for sure. Since the MIPS kernel has sysfs bonding support it may already have the required code, though.

 

[KAD]

Regardless of how I add the slaves still have the same issues. I'm still searching the web for a solution...

how's the switch set up
what does the switch show for status

active/passive
link speed, etc ...
does switch show LACP running

just currious, what the other side of this link shows

 

[RMerlin]

I suspect it requires kernel support as well, but I don't know for sure. Since the MIPS kernel has sysfs bonding support it may already have the required code, though.

Check if maybe Entware might have a newer iproute2, you could test with it.

Or if you have a working build environment, have your build use the newer version of iproute2, and copy the resulting binary to your router to try out.

 

[mrengles]

how's the switch set up
what does the switch show for status

active/passive
link speed, etc ...
does switch show LACP running

just currious, what the other side of this link shows

I haven't tested my netgear switch yet (it's at my office but has worked with every other LAG bond I through at it.), I'm trying to setup with a MacPro with dual GigE intel Ethernet. From some earlier posts another user setup his Mac Pro without issue.

 

[KAD]

I haven't tested my netgear switch yet (it's at my office but has worked with every other LAG bond I through at it.), I'm trying to setup with a MacPro with dual GigE intel Ethernet. From some earlier posts another user setup his Mac Pro without issue.

So what does the macPro say about the link
does it give you any info on how/if it's running and what settings are used

 

[mrengles]

So what does the macPro say about the link
does it give you any info on how/if it's running and what settings are used

Only thing i can find... Both en0 and en1 show "No Partner" and when selected says... The port on the switch that the device is connected to doesn't seem to have 802.3ad Link Aggregation enabled.

I'm still going through the logs to find more detail. If I connect the Mac Pro to my older airport, 802.3ad activates right away. Same thing with an un-managed net gear GS108, But I can't change any settings or look at logs with that cheap 8-port switch to give you any real answers.

Thanks again for all your helps guys!

-mrengles

 

[KAD]

Only thing i can find... Both en0 and en1 show "No Partner" and when selected says... The port on the switch that the device is connected to doesn't seem to have 802.3ad Link Aggregation enabled.

I'm still going through the logs to find more detail. If I connect the Mac Pro to my older airport, 802.3ad activates right away. Same thing with an un-managed net gear GS108, But I can't change any settings or look at logs with that cheap 8-port switch to give you any real answers.

Thanks again for all your helps guys!

-mrengles

the airport I don't know about,
but the unmanaged netgear GS108 definetely would not work correctly with 802.3ad, so if the macpro says linkagg works with it, something is not right

for linkagg to work, all ports need to be configured correctly to match and added to a bonding interface like bond0
so I don't see any way to do that with unmanaged switch
I've also never seen any unmanaged switch that supports 802.3ad

 

[mrengles]

the airport I don't know about,
but the unmanaged netgear GS108 definetely would not work correctly with 802.3ad, so if the macpro says linkagg works with it, something is not right

for linkagg to work, all ports need to be configured correctly to match and added to a bonding interface like bond0
so I don't see any way to do that with unmanaged switch
I've also never seen any unmanaged switch that supports 802.3ad

The GS108 definitely supports 802.3ad, same as the Airport, its all setup automatically without need to configure the firmware though.

http://www.netgear.com/images/GS105v3_GS108v3_DS_27Apr1118-4903.pdf

 

[KAD]

The GS108 definitely supports 802.3ad, same as the Airport, its all setup automatically without need to configure the firmware though.

http://www.netgear.com/images/gs108tv2_ds_10dec0918-5336.pdf

ok, very surprised, documentation says it does in fact support LACP

 

[mrengles]

ok, very surprised, documentation says it does in fact support LACP

I was shocked myself. It's cheap but you have no control over the trunk. I'm pretty sure it only supports one bond, but can't say for sure only tested with one setup.

I tried connecting the gs108 but it did not see the bond on the rtac66u as expected

 

[saintdev]

ok, very surprised, documentation says it does in fact support LACP

Note, it's under the section Administrative Switch Management. That is one of the managed features of the switch. It's not an unmanaged switch.

Alternatively, you can use SNMP-based software to manage your Smart Switch. NETGEAR ProSafe Smart Switches are ideal for adding basic management to your unmanaged networks or extending your managed networks.

– IEEE 802.3ad Link Aggregation (manual or LACP)

This is not saying "manual or automatic". This is more accurately "static or dynamic". See the Wikipedia Link aggregation page: 2.2.1 Advantages over static configuration for what this means.

You will have to configure the switch so that the ports are set to LACP.

 

[mrengles]

@RMerlin

Are you using newer Broadcom drivers then ddwrt at the moment?

@ AnyoneElseWhoCares

Well I have been reading online there's a bug in the Linux Kernel 3.10.x that allows ARP monitoring in 802.3ad mode, which should not be the case. The current builds of dd-wrt are based on 3.10.25, and I'm pretty sure its missing the necessary (patch) to correct the problem.

The good news we can disable arp monitoring with the linux driver by setting arp_interval = 0, but I don't know because of the bug if it will take the setting?

arp_interval

Specifies the ARP link monitoring frequency in milliseconds. If ARP monitoring is used in an etherchannel compatible mode (modes 0 and 2), the switch should be configured in a mode that evenly distributes packets across all links. If the switch is configured to distribute the packets in an XOR fashion, all replies from the ARP targets will be received on the same link which could cause the other team members to fail. ARP monitoring should not be used in conjunction with miimon. A value of 0 disables ARP monitoring.The default value is 0.

The snippet that will hopefully fix my Aggregate ID, and Link Speed issues?

echo 0 > /sys/class/net/bond0/bonding/arp_interval
echo 0 > /sys/class/net/bond0/bonding/use_carrier
echo layer2+3 > /sys/class/net/bond0/bonding/xmit_hash_policy

I'm at work and can't test this setting yet, but will do so the second I get home.

If anyone is interested in the bug itself please read below, search the page for "802.3ad" otherwise your in for a long read.

https://www.kernel.org/pub/linux/kernel/v3.x/ChangeLog-3.10.23

Keep your fingers crossed, so that I can get out of your hair soon.

-Robert (mrengles)

Afterthought:

Looking through the ddwrt source code. I found these lines of code; which represent the fix, Hope its not in the build i'm using for some reason, because then I still have a chance of an easy fix, with a know solution.

        }
   	        if (bond->params.mode == BOND_MODE_ALB ||
   	            bond->params.mode == BOND_MODE_TLB ||
   	            bond->params.mode == BOND_MODE_8023AD) {
   	                pr_info("%s: ARP monitoring cannot be used with ALB/TLB/802.3ad. Only MII monitoring is supported on %s.\n",
   	                        bond->dev->name, bond->dev->name);
   	                ret = -EINVAL;
   	                goto out;

 

[RMerlin]

@RMerlin

Are you using newer Broadcom drivers then ddwrt at the moment?

No idea what DD-WRT uses. I think BSlayer has an SDK licence, so it's probably quite different code from what Asus uses, since Asus does changes to it, while BSlayer is probably just recompiling the Broadcom code targetted at his kernel. I don't have an SDK licence, so I just use the binary blobs from Asus.

Strictly at an SDK level, Asus is on 6.37.14.62 (r437318) for ARM devices, and 6.30.102.9 (r366174) for the SDK6 MIPS devices (I got newer GPL code for those devices but didn't have time to check if they had upgraded to a newer SDK).

 

[mrengles]

We'll my tested setting above... Doesn't change anything.

I've PM-ed the two main developers of ddwrt for this router and asked them to look at my posts, keeping my fingers crossed.

Thanks everyone for the help!

Should I try to contact the linux bond driver developers, or not even bother?

Robert (mrengles)

 

[RMerlin]

Should I try to contact the linux bond driver developers, or not even bother?

Considering how old the kernels used by these routers are, I doubt you'll get any support from these, to be honest.

 

[saintdev]

You will have to configure the switch so that the ports are set to LACP.

See the Link Aggregation Groups section of the GS108Tv2 User Manual (starts on page 89 of the PDF).

 

[mrengles]

See the Link Aggregation Groups section of the GS108Tv2 User Manual (starts on page 89 of the PDF).

Unfortunately it's not Netgear switch set up. It has to do with RT-AC66U 802.3ad settings (or bug). I've tested several different combinations of active-passive and nothing changes.

 

[saintdev]

Unfortunately it's not Netgear switch set up. It has to do with RT-AC66U 802.3ad settings (or bug). I've tested several different combinations of active-passive and nothing changes.

Apologies. I guess I was a little confused, because earlier you said..

The GS108 definitely supports 802.3ad, same as the Airport, its all setup automatically without need to configure the firmware though.

then

I was shocked myself. It's cheap but you have no control over the trunk.

So I assumed you had not configured the switch properly.