Balance-alb channel bonding crashing RT-AC66U with 3.0.0.4.354 firmware

[shade99gt]

I have a QNAP TS-419P II with both NIC's bonded using Balance-alb and plugged into the RT-AC66U. This has worked great with all other ASUS RT-AC66U firmwares, but ever since I upgraded to 3.0.0.4.354 (both stock and Merlin's), the router becomes unresponsive in about 30 seconds to a minute and then reboots constantly reboots.

As soon as I unplug one of the bonded NICs, it's fine. If I switch the bonded mode to Balance-tlb instead of Balance-alb, it's also fine. However i'd like to use Balance-alb because I get big performance increases for doing backups to my QNAP since Balance-alb increases the rx and tx speeds whereas Balance-tlb only increases tx speeds.

Any ideas? Is there something switch related in this new firmware that could be causing this that I can try tweaking?

Thanks

 

[canuckle]

Although it doesn't directly answer your issue, I went with a dedicated managed switch for LAG on my NAS, assuming you are looking for internal transfer/streaming bandwidth. I looked at doing it via the router but needed more than 2 other LAN ports anyway

I went with the Netgear GS108T, it's been flawless, truly set and forget. Keeps firmware upgrades from messing with it too.

YMMV

 

[shade99gt]

Although it doesn't directly answer your issue, I went with a dedicated managed switch for LAG on my NAS, assuming you are looking for internal transfer/streaming bandwidth. I looked at doing it via the router but needed more than 2 other LAN ports anyway

I went with the Netgear GS108T, it's been flawless, truly set and forget. Keeps firmware upgrades from messing with it too.

YMMV

From what i've read, using switches that support LAG's with the QNAP port trunking modes that required LAG's didn't really yield much (if any) performance boosts over Balance-alb which doesn't require any special switch.

Just seems odd that this firmware breaks Balance-alb when it was working fine before.

I'm also hoping not to have to spend any more $$$ to get something working that worked fine before.

 

[RMerlin]

Just a shot in the dark, but have you tried disabling Spanning-Tree Protocol support on the router?

nvram set lan_stp=0
nvram commit
reboot

I never tried any NIC bonding before, so I'm not sure if it might run into any issue with STP enabled or not.

You can revert that change by setting it back to "1".

 

[shade99gt]

Just a shot in the dark, but have you tried disabling Spanning-Tree Protocol support on the router?

nvram set lan_spt=0
nvram commit
reboot

I never tried any NIC bonding before, so I'm not sure if it might run into any issue with STP enabled or not.

You can revert that change by setting it back to "1".

I actually just disabled Spanning-Tree Protocol about an hour ago and it seems to be running ok now. Looks like that did it.

Thanks for the reply!

 

[RMerlin]

Awesome

Good thing you found out already, cause I just noticed I had mistyped the nvram variable name (I just edited my post to correct it).

 

[shade99gt]

Awesome

Good thing you found out already, cause I just noticed I had mistyped the nvram variable name (I just edited my post to correct it).

Just wanted to give a small update in case anyone else runs into this problem:

After further testing, it appears that disabling Spanning Tree Protocol only delayed the router crashing when Balance-alb port trunking is used on the QNAP NAS. The crashing also occurred even if the QNAP NAS using Balance-alb was plugged into another switch on the network (not directly into the RT-AC66U ports).

To stop the RT-AC66U from crashing and rebooting, I actually had to disable both STP and HW Acceleration on the RT-AC66U. However, disabling HW Acceleration also resulted in a noticable drop in performance when I did read/write tests to and from the QNAP NAS.

I actually went the route that Canuckle did and bought a Netgear GS108T switch that supports LACP 802.3ad port trunking and used that mode with the QNAP NAS.

To summarize:

The RT-AC66U 3.0.0.4.354 firmware has issues when Balance-alb port trunking is used on the network if Hardware Acceleration and/or STP is enabled under "Switch Control" on the RT-AC66U. Other "general" port trunking modes did not cause crashing, but also are lower performing.

 

[RMerlin]

FW 3.0.0.4.354 has at least one known issue with its switch driver on the RT-AC66U (random latency spike up to 10 ms). I would try the latest non-beta version (270) to see if it's more stable in your particular case, tho using a dedicated switch would certainly work as well.

 

[shade99gt]

FW 3.0.0.4.354 has at least one known issue with its switch driver on the RT-AC66U (random latency spike up to 10 ms). I would try the latest non-beta version (270) to see if it's more stable in your particular case, tho using a dedicated switch would certainly work as well.

I did not have any crash issues with all previous firmware verions (including the latest non-beta 270). The problem was introduced with 3.0.0.4.354.

Just wanted to put it out there for anyone else experiencing crashing issues when using Balance-alb port trunking with a QNAP NAS or other devices and this firmware.