Hi there,

we have a Accespoint with three 120 degrees sector antennas. The AP is running 4.13 ... three of four clients (all connected to the same antenna) run 4.13 also, only one client runs old 3.20

The AP is using nstreme (exact size) and 5Ghz turbo Mode. All sectors are using different frequencies (5800, 5290 and 5210)

Now i have the issue, that the 4.13 clients are disconnected within some days. All show "lost connection, not polled for too long" .. the only client that stays up is the one with 3.20 running .. all clients have nearly the same link quality (ccq is 100%) all clients have LOS .. HW retries is set to 7 (default)

I have no idea what's causing that.

Any hints or ideas?

Thanks and regards,
Joerg

Hi there,

we have a Accespoint with three 120 degrees sector antennas. The AP is running 4.13 ... three of four clients (all connected to the same antenna) run 4.13 also, only one client runs old 3.20

The AP is using nstreme (exact size) and 5Ghz turbo Mode. All sectors are using different frequencies (5800, 5290 and 5210)

Now i have the issue, that the 4.13 clients are disconnected within some days. All show "lost connection, not polled for too long" .. the only client that stays up is the one with 3.20 running .. all clients have nearly the same link quality (ccq is 100%) all clients have LOS .. HW retries is set to 7 (default)

I have no idea what's causing that.

Any hints or ideas?

Thanks and regards,
Joerg

I see this behavior on APs with interference. On one AP I see this even with 50MHz
Frequency spacing. The problems increased with 11n Cards. But I'm not sure the
cards are the problem. Maybe the newer ROS-Versions introduced it.
You'll find some posts here which leads to the assumption ROS3.30 was more
stable regarding wireless than 4.x. Running without Nstreme on this links
increased stability for me.

Had clients dropping out and reassociating used setting by TomjNorthIdaho and then used NV2, which solved the issue, but if this does not work, and interference is suspect time to hire out a spectrum analyzer.

Advanced:
ACK: dynamic
Periodic Calibration: enabled
Calibration Interval: 00:10:00
Hw Retries: 15
Hw Protection Mode: cts-to-self
Preamble Mode: [x] long
Disconnect Timeout: 00:00:12
On Fail Retry Time: 900

Nstream:
Enable Nstream - not checked
Enable Polling - not checked
Disable CSMA – checked

client radio set to use: RTS/CTS
client rts/cts threshold: 32

I was getting the "lost connection, not polled for too long" disconnect issue when using Nstream on point to multipoint set up using an Omnitik as AP

Routing through all the posts I can find on the subject and based on users experience I have come to this conclusion with regards best Practice.

AP Specific Settings:

Wireless settings:
Band = A only
Protocol = nstream only
Default authenticate = no (use access list)

Nstream:
Enabled = Yes
Polling =yes
CSMA = disabled
Framer policy = Best Fit
Framer Limit 3200

Client specific Settings

Wireless settings:

Default authenticate = No (set AP details in Connect list)

I found this combination solved the polling disconnect and links are stable

Hope this helps

My set up is Omnitik to SXT ros ver 5.24

Thanks karina,

as I used your tip, mi link is stable and it did not ever had an uptime like this!
I'm using 3 SXT connected to a Groove5 all on RouterOS 6.7