[ath][CURRENT] Triangle-shaped network performance

[Lytochkin Boris]

On Sun, Mar 25, 2012 at 10:44 PM, Adrian Chadd <adrian.chadd at gmail.com> wrote:
> Hi,
>
> The LOR is known. I'd love to fix it but I have other things to do first.
>
> Try running athstats 1 and watch the rate of errors change over time.
Here it is.
   input   output altrate   short    long xretry crcerr crypt  phyerr rssi  rate
   44332    38318      20       0   14735   1813   1536     0       0   60 MCS11
      39       41       0       0       2      0      0     0       0   69 MCS11
      32       64       0       0      21      3      0     0       0   53 MCS11
      38       66       0       0       4      0      0     0       0   64 MCS11
      37       40       0       0      14      2      0     0       0   60  MCS6
      28       32       0       0       1      0      0     0       0   56 MCS11
      34       46       0       0       1      0      1     0       0   51  MCS0
      44       37       0       0      18      3      0     0       0   64 MCS11
      32       61       0       0       4      0      0     0       0   62 MCS11
      41       60       0       0      20      3      1     0       0   66 MCS11
      45       54       0       0       7      0      0     0       0   59  MCS6
      43       64       0       0      20      2      0     0       0   65 MCS11
Somewere in the middle triangle-shaped ping reset (timed out and then
rolled over)
>
> Disable HT40 as on ongested channels it's going to behave very oddly.
That does not help always.
5 minutes ago I had the same  triangles on
~>ifconfig wlan0
wlan0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
        ether b0:48:7a:da:0e:59
        inet6 fe80::b248:7aff:feda:e59%wlan0 prefixlen 64 tentative scopeid 0x10
        inet 192.168.2.1 netmask 0xffffff00 broadcast 192.168.2.255
        nd6 options=29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>
        media: IEEE 802.11 Wireless Ethernet autoselect mode 11ng <hostap>
        status: running
        ssid starlink channel 13 (2472 MHz 11g ht/20) bssid b0:48:7a:da:0e:59
        regdomain ETSI country RU indoor ecm authmode WPA privacy MIXED
        deftxkey 3 AES-CCM 2:128-bit AES-CCM 3:128-bit txpower 50 scanvalid 60
        protmode CTS ampdulimit 64k ampdudensity 8 shortgi wme burst
        dtimperiod 1 -dfs bintval 1000

(the moment I took athstats snapshot shown in this mail).


Some more observations: it can be cured temporarily fetching some data
though wireless: first it will start at 10kb/s than in a minute or
something triggers and both speed goes up and pings are sane. If
transfer is stopped triangles come in 5-10 seconds after transfer
abort.

Here it is some kind of diagram of such behavior (reproducible), 1
ping per second:
Reply from 192.168.1.5: bytes=32 RTT=297ms TTL=64 <-- transfer
started, speed 10kb/s
Reply from 192.168.1.5: bytes=32 RTT=318ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=340ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=360ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=388ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=410ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=434ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=457ms TTL=64
Timeout.
Reply from 192.168.1.5: bytes=32 RTT=35ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=58ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=78ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=101ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=123ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=147ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=173ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=195ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=217ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=237ms TTL=64
Timeout.
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64 <-- speed goes to normal
Reply from 192.168.1.5: bytes=32 RTT=3ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=2ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=4ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=3ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=2ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=3ms TTL=64 <-- transfer aborted
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=2ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT<1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=1ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=299ms TTL=64 <-- triangles again!
Reply from 192.168.1.5: bytes=32 RTT=322ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=345ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=368ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=390ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=413ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=436ms TTL=64
Timeout.
Reply from 192.168.1.5: bytes=32 RTT=34ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=57ms TTL=64
Reply from 192.168.1.5: bytes=32 RTT=80ms TTL=64


-- 
Boris Lytochkin