Bug 643211 - Networkmanager with iwlagn stops for about 3 seconds exactly every 2 minutes
Summary: Networkmanager with iwlagn stops for about 3 seconds exactly every 2 minutes
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Fedora
Classification: Fedora
Component: NetworkManager
Version: 13
Hardware: i386
OS: Linux
low
medium
Target Milestone: ---
Assignee: Dan Williams
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2010-10-14 22:04 UTC by boggiano
Modified: 2011-09-06 13:31 UTC (History)
3 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2010-10-18 17:35:59 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
Output of iwevent command (1.38 KB, text/plain)
2010-10-14 22:05 UTC, boggiano
no flags Details

Description boggiano 2010-10-14 22:04:20 UTC
Description of problem:
Every 120 seconds my connection stops for about 2 seconds.
It disconnects from AP then reconnects immediately.

Kernel is 2.6.34.7-56.fc13.i686.PAE.
NetworkManager-0.8.1-6.git20100831.fc13

How reproducible:
Always

Steps to Reproduce:
1. Start networkmanager 
2. Looks into logs
  
Actual results:


Expected results:


Additional info:
The connection is "insecure" (not WPA or WEP).
120 is the "max time", sometimes even less. (as you can see in the attach file iwevent.txt).
If I stop NW and I use iwconfig by hand, this problem stops but I need a "ping -c www.google.com" in background or the  connection drops.

If I lunch the command:
while : ; do iwconfig wlan0|grep 'Freq' ;sleep 1; done
this is the output during the "disconnects":

Mode:Managed  Frequency:2.412 GHz  Access Point:xx:xx:xx:xx:xx:xx   
Mode:Managed  Frequency:2.412 GHz  Access Point:xx:xx:xx:xx:xx:xx   
Mode:Managed  Frequency:2.412 GHz  Access Point: Not-Associated   
Mode:Managed  Frequency:2.412 GHz  Access Point: Not-Associated   
Mode:Managed  Frequency:2.412 GHz  Access Point: Not-Associated   
Mode:Managed  Frequency:2.412 GHz  Access Point:xx:xx:xx:xx:xx:xx   
Mode:Managed  Frequency:2.412 GHz  Access Point:xx:xx:xx:xx:xx:xx

Comment 1 boggiano 2010-10-14 22:05:23 UTC
Created attachment 453591 [details]
Output of iwevent command

Comment 2 Dan Williams 2010-10-18 17:35:59 UTC
This is due to periodic scanning.  If you don't want periodic scanning (which allows intra-SSID roaming and faster reassociation on connection loss) then you can disable it by "locking" your connection to a specific BSSID in the connection editor.

Right-click on the network applet and choose Edit Connections... or run 'nm-connection-editor', click the Wireless tab, double-click on your wifi connection, and in the Wireless tab in that edit window, enter your AP's MAC address.  Then click Apply, and re-select your wifi network from the applet's menu.

NM will then tell the supplicant and driver to lock the connection to that specific AP, and will no longer request periodic scans.

Comment 3 boggiano 2010-10-18 20:39:16 UTC
Yes!!
It works very well!
Thanks a lot!!!

Comment 4 Steven Ellis 2011-01-10 22:35:27 UTC
I've just tried the above fix on a HP 6730b laptop running Fedora 14 and the machine is still re-connecting to the same BSSID Mac address every 20 minutes.

Comment 5 Dan Williams 2011-01-10 22:51:31 UTC
Just because *NM* isn't asking for scans doesn't mean something else like the supplicant itself isn't asking for them.  Believe it or not, there are valid reasons to do a scan while connected.  Like if the driver decides to disconnect for some reason, or if the AP timed out because somebody turned on a microwave, etc.  Take a look at 'dmesg' output from that time, and it's likely that the AP failed to respond to probe requests, which causes a disconnect, which causes the supplicant to scan for an AP to reconnect to.

The fix above just means that *NM* isn't asking for periodic scans, it doesn't suppress *all* scans by the wireless stack entirely, because there's often good reason why they are done.

Comment 6 Steven Ellis 2011-01-10 22:57:49 UTC
Problem is when the scan kicks in I loose my wireless connection for 2-3 seconds which is a real pain if I'm using ssh into a server and can no longer type.

Comment 7 Maciej Sitarz 2011-09-06 13:31:13 UTC
I'm having the same problem. My connection is interrupted every few minutesduring iwevent showing "Scan request completed". When using SSH, synergy or any other 'live connection' the work is troublesome.

I don't care if NM scans, but I think the connection shouldn't be interrupted during that. The workaround with setting exact BSSID MAC address in NM configuration works, but is the connection interrupted by design?
Maybe it's iwlagn module/firmware bug? or NetworkManager's fault?

Detailed info:
- hardware: 03:00.0 Network controller: Intel Corporation Centrino Advanced-N 6200 (rev 35)
- OS Fedora 14 (with latest updates)
- kernel: 2.6.35.14-95.fc14.x86_64 #1 SMP Tue Aug 16 21:01:58 UTC 2011 x86_64 x86_64 x86_64 GNU/Linux
- iwlagn module srcversion: 683C5EFF5AC57D530DD7FC4
- iwlagn firmware version 9.221.4.1 build 25532
- NetworkManager version: 0.8.4-2.git20110622.fc1


Note You need to log in before you can comment on or make changes to this bug.