RT61 CVS 2008102312 + WPA2 Tx Errors

Live forum: http://rt2x00.serialmonkey.com/viewtopic.php?t=5056

timb80

26-10-2008 09:02:18

Hi,

I have configured my card successfully to use WPA2-PSK (without wpa_supplicant) using iwpriv, short story is that on CentOS/RHEL I needed to add the iwpriv commands to /etc/rc.local to setup the card for WPA2 after boot. But these would only work in the following order, note the iwconfig eth1 key off - this seemed crucial to diable WEP prior to enabling WPA2 for some reason, maybe a conflict from the distro which only uses WEP... anyhow-
[code2jhx9383]
# Set up wireless WPA2 on boot (rt61)
/sbin/iwlist eth1 scan

/sbin/iwconfig eth1 essid youressidhere
/sbin/iwconfig eth1 key off

/sbin/iwpriv eth1 set AuthMode=WPA2PSK
/sbin/iwpriv eth1 set WPAPSK=yourpassphrase
/sbin/iwpriv eth1 set EncrypType=AES
[/code2jhx9383]

However, now that I have successfuly set it up - I have benchmarked a few transfers and noticed that I am getting approx 1 to 1.2mbps in both directions which is a little low for 802.11g/54mbps (max on network approx 2.5mbps), and a lot of Tx Errors. This appears to be about 1 in every 10 Tx packets, Rx packets are much lower... i'd love to improve the throughput and remove the errors - any clues?? I note from the CVS Issues sticky that Tx collision rates are improved but the thread no longer exists?

[code2jhx9383]
eth1 Link encap:Ethernet HWaddr 00:11:50:DD:51:2F
inet addr:192.168.1.2 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::211:50ff:fedd:512f/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:446698 errors:47699 dropped:0 overruns:0 frame:0
TX packets:45018 errors:37 dropped:37 overruns:0 carrier:0
collisions:3240 txqueuelen:1000
RX bytes:132415288 (126.2 MiB) TX bytes:47502557 (45.3 MiB)
Interrupt:11
[/code2jhx9383]

Edit Debug output to follow...

timb80

28-10-2008 13:51:12

Hmm odd - the Tx errors only increase during LAN traffic (about 1 in 10 packets), and not traffic directed to my WAN Router.

This interface is placed on a server with an Gigabit NIC on a different subnet...does anyone know of any command line tools that will allow me to identify possible problems?

EDIT This boils down to the rate of throughput, the WAN traffic is only 2mbps so handled with ease, however LAN traffic is up to 54mbps and the card struggles when fixed at this speed probably due to strength of signal, see post below.

timb80

12-03-2009 13:17:05

For anyone with a similar problem, I found that the excessive collisions and errors were caused by inadvertently setting the rate to 54M in the ifcfg file - which then fixed it at that speed. This seemed to cause problems with the weaker signal, setting to 'auto' fixed or at least lowered the problems to more sensible levels (collisions ~5%, errors <0.1%).

it still operates mostly around 36M during peak transfers, and returns to 54M for small / light traffic...changing channels (from 1 to 8) seems to have helped this significantly, so i suspect some form of local interference

iwconfig eth1 rate auto

[codeuixd0de1]eth1 Link encap:Ethernet HWaddr 00:11:50:DD:51:2F
inet addr:192.168.1.2 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::211:50ff:fedd:512f/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:3313199 errors:193 dropped:0 overruns:0 frame:0
TX packets:1012864 errors:787 dropped:787 overruns:0 carrier:0
collisions:222572 txqueuelen:1000
RX bytes:645782719 (615.8 MiB) TX bytes:862407050 (822.4 MiB)
Interrupt:11
[/codeuixd0de1]