Michael McNamara

technology, networking, virtualization and IP telephony

Lenovo ThinkPad T460 Yoga with Intel AC 8260 Wireless Issues

by

I recently came across an issue where the Lenovo ThinkPad T460 Yoga with Intel AC 8260 wireless adapter was having all sorts of issues connecting to and passing traffic across a Cisco 5508 Wireless LAN Controller with 1262N and 3702E Access Points running 8.0.133.0 software, the most recent release at the time of the issue. The first thing we tried was upgrading the driver for the Intel Dual Band Wireless-AC 8260 to 19.1.0.4 (7/16/2016) which was the latest available at the time. Unfortunately that didn’t help any, we also tried applying an 8.0.135.5 software version to the Cisco WLC, again that didn’t help any.

The laptop would often connect to the SSID but the laptop would be unable to get a webpage to render with all IP traffic essentially stalling. ICMP ping times would jump from 1 ms to 3,900 ms with multiple dropped packets scattered all about the constant ping. Without any load you could occasionally get 1 ms response times for a couple of minutes at a time but the instant you opened a web page the traffic would stall and the ICMP pings would start timing out.

The Intel engineer that was assisting me provided the hint, letting me know that Cisco IT had actually stumbled across this very same issue the week earlier internally with their own employees. Cisco had intentionally disabled A-MPDU on their WLCs, the workaround was to enable A-MPDU for 802.11n on their WLCs. I went ahead and checked our WLCs and sure enough we also had A-MPDU disabled – not exactly sure who or why it was disabled.

802.11n Status: 
    A-MPDU Tx: 
        Priority 0............................... Disabled 
        Priority 1............................... Disabled 
        Priority 2............................... Disabled 
        Priority 3............................... Disabled 
        Priority 4............................... Disabled 
        Priority 5............................... Disabled 
        Priority 6............................... Disabled 
        Priority 7............................... Disabled 
        Aggregation scheduler.................... Enabled 
        Frame Burst.............................. Automatic 
            Realtime Timeout..................... 10 
    A-MSDU Tx: 
        Priority 0............................... Enabled 
        Priority 1............................... Enabled 
        Priority 2............................... Enabled 
        Priority 3............................... Enabled 
        Priority 4............................... Enabled 
        Priority 5............................... Enabled 
        Priority 6............................... Disabled 
        Priority 7............................... Disabled 
    Rifs Rx ..................................... Enabled 
    Guard Interval .............................. Any

I used the following CLI commands to enable A-MPDU; (note that I had to temporarily disable the 802.11a network to make the change – you’ll want to schedule this off-hours)

config 802.11a disable 
y 
config 802.11a 11nsupport a-mpdu tx priority 0 enable 
config 802.11a 11nsupport a-mpdu tx priority 1 enable 
config 802.11a 11nsupport a-mpdu tx priority 2 enable 
config 802.11a 11nsupport a-mpdu tx priority 3 enable 
config 802.11a 11nsupport a-mpdu tx priority 4 enable 
config 802.11a 11nsupport a-mpdu tx priority 5 enable 
config 802.11a enable

Why doesn’t the Intel AC 8260 wireless adapter negotiate using A-MSDU?

I hope to be able to bring you that answer from either Cisco or Intel.

I hope you enjoyed the article Tim.

Cheers!

Update: December 7, 2016

Intel has released a new driver for the AC 8260 that is designed to address the issue.
https://downloadcenter.intel.com/download/26465/Intel-PROSet-Wireless-Software-and-Drivers-for-Windows-10
https://downloadcenter.intel.com/download/26469/Intel-PROSet-Wireless-Software-and-Drivers-for-IT-Admins

I’m currently testing the driver but haven’t had enough time to comment yet.

Issues with wireless roaming and Ralink chipsets?

by

We recently needed to replace the USB based wireless adapter we use in our CoWs (cart on wheels). The legacy D-Link DWL-AG132 (802.11a/g) was no longer being manufactured and/or available from suppliers so we selected the D-Link DWL-160 (802.11n/a/b/g) and that’s were the fun started. We utilize HP Thin Clients on our carts and dwa-160have been pairing them with USB adapters for the past year and a half (saves on having to purchase the chassis expansion kit and the PCI card).

Our initial tests showed no problems in connecting to our Motorola RFS7000 with AP300 (802.11a/b/g) radios utilizing 802.1x (WPA/TKIP). A few weeks later though we learned we had a problem with the combination and we eventually discovered that the problems revolved around roaming. We noticed that the DWL-160 didn’t want to roam until it had lost complete signal to the AP it was associated to and then only after about 10 – 15 seconds would it roam to another APl. We had a really large project and the time frame was extremely tight so we decided to run out and pick-up a different adapter. So we purchased a Cisco Linksys wusb600n(WUSB600N) Wireless-N USB Network Adapter with Dual-Band. We decided to simplify our testing environment by removing the Windows XP Embedded (Thin Client) and test on a simple Windows XP SP2 laptop. We also removed the authentication (802.1x) and encryption (WPA/TKIP) and just test using an open network. We quickly noticed that the problem was not only evident at the 2.4Ghz frequency (802.11b/g) but also evident at the 5 Ghz frequency (802.11a). The big surprise came we noticed that the Linksys behaved exactly like the D-Link in that it would not roam which would lead to poor connectivity. We started to peel back the onion and almost immediately found that both products were based on the Ralink Technologies chipset (RT-2870).

We’ve performed multiple wireless packet traces using AirPCAP and WireShark and they don’t show any issues with the wireless access points, however, they do show a lack of probing and some odd behavior by the wireless adapters (STA).

We’re in contact with both Motorola and Ralink Technologies and we have tried Ralink’s reference drivers (1.4.1) along with a beta driver they have provided but we haven’t seen any real improvement. We did find that if you enable “Fast roaming”, which can be found the in the Advanced Properties of the RA utility the STA behaves much better but it’s still a world apart from the roaming performance that we’re accustom to.

Has anyone else seen any issues with either the D-Link DWL-160 or the Linksys WUSB600N in a corporate network with respect to roaming? I can’t imagine that we’re the first folks to stumble upon this issue given that Ralink Technologies chipsets are in all these products.

Cheers!

Power over Ethernet Plus (PoE+)

by

I just recently learned that the majority of 802.11n products in design will likely out pace the current 13-15 watts of power provided by the 802.3af specification. It seems the IEEE is already working on 803.at, a new specification labeled “PoE+” by some.

What does this mean for the thousands of PoE (802.3af) ports already deployed throughout organizations?

Here’s a good article, A Look at POE Plus, in Network Computing by Peter Morrissey.

There are also some interesting articles over at Network World regarding 802.11n.

I’m not sure about everyone else out there but I won’t be rushing to deploy 802.11n or 802.3at gear anytime soon. We’ve actually standardized on using PoE capable network switches throughout the network going forward. The price cost between a PoE switch and a non-PoE switch is almost negligible when you consider the time and effort required to replace that switch in the future if PoE is required for some new application.

If you’re seriously thinking about deploying 802.11n you’ll need to consider how you’re going to power those devices.

Cheers!

Recent Posts

Categories

SUBSCRIBE BY EMAIL