We have a project where the customer has 3 Cisco AP1250 Autonomous APs in a parking lot. They have upgraded to the latest Cisco firmware and have disabled the 2.4GHz interface, so the APs only run 5GHz. They do not support WDS and the application we are supporting needs layer 2 device discovery, so we have to run Station Psuedobridge or Psuedobridge Clone on the RB411U boards which are client bridges.

The APs are using 8dBi gain 120 deg sector 3x3 mimo antennas from Mobile Mark. We are using 2x2 mimo 4dBi trunk mount antennas on the vehicles.

Testing with only one AP and one station, at 100 feet on 5825 with 40/below, we first were able to get 70Mbps TCP throughput as tested with IPerf using 8 threads. No other traffic in the area. AES was enabled, along with WMM, chains 1 and 2 on the station and chains 1,2 and 3 on AP, APMPDU of 0.

We sniffed the streams and saw really no collisions. CCQ was about 80/80, which was low considering, but we just figured that maybe we had too much gain (40 dB SNR). Turned tx gain down on both the AP and station with no effect.

So then we turned up the second AP on a non-overlapping channel (5745 40 above). Used a different SSID, and AES key. As soon as we turned this up, the first AP dropped to 2Mbps throughput and the second was getting about 25Mbps. Tried changing the scan list so that we were not on default and just picked one channel. No difference.

Then we turned up the third AP on 5180 40 above. All three links dropped to 2Mbps throughput.

Sniffing the channels show no collisions, but the CCQ drops to 11/11 on all three links.

Then the really strange thing...we turn off the second and third APs and stations, but the first pair stays at 2Mbps throughput. It never recovers back to 70Mbps. In fact we power off and power on everything and can never get back to the speed we first had.

Yes, when we put a Routerboard with ROS running as AP and use NV2 or NStreme all is good. But the customer has already purchased and installed all of the Ciscos, so we are stuck and have to support them.

The Cisco AP1250 is a total piece of crap, but we have to support it. I could post all sorts of diagnostics, but I was hoping that before this thread gets loaded up, someone might have experience with the AP1250 and have a trick.

We were reading about something called "deadlocking" that was related to AMPDU priorities. It kind of described what we were seeing.

Any ideas. This is a real mystery.

Bump...Nobody has any input on this one???

I've been in those situations. I think it would be in your best interest as well as your customer's to sell the Cisco equipment to someone on Ebay who doesn't know any better and replace it with Mikrotik.

I've got some x86 running ROS absolutely running circles around Cisco routers, CMTS and L3 switches, all the while doing things that aren't even possible with Cisco. As far as I'm concerned Cisco is surviving on their old reputation and AT&T. I have had more issues with Cisco switches and WAPs than I have with anything other than Netgear smart equipment.

The world has changed and Cisco is stuck somewhere in the late 90s.

edit: on the RF side, how far apart are the APs and sectors? Do you have any kind of shielding?