frankd wrote: 
> Log files of the script (version 0.73) with two events capture, when the
> script could not reestablish the network (second one around 12:30, first
> one around 11:55).
Thanks for the log files, an interesting read!

Looking at wlanerr.log, an 11:55:26 outage (Line 3) was restored (l 13)
at 5 ping fails by a quick reset at 11:55:41, only to be followed 
by a quick reset that took 15 seconds in all. However, a second outage
occurred just 39 seconds later at 11:56:20 (line 15). The recovery
record (l 25) shows a 11:57:09 recovery at 3 ping fails. The entry shows
a hard reset entry at 11:56:37, then a quick reset at 11:57:06, and the
wireless recovering by 11:57:09, indicating the network connection was
successfully reestablished, and stayed that way until the next outage.

The next came at 12:05:06, but the ping statistics were not written to
the log file, as perhaps the radio was being rebooted. It came up at
12:07:00.

Again, the failure at 12:30:24 neither recovered nor wrote the ping
statistics to the log file (as it was also rebooted?), coming up at
12:31:50. That might have a little too soon to reboot, or there is
something else wrong. Those missing statistics are in your stats.txt
file.

The current thinking is that the quick reset introduced in >= 0.7.0 is
neglecting to keep the jive player "happy," and perhaps the radio showed
no connection when it actually was successfully pinging. A brand new
requirement of keeping jive happy is being addressed in 0.7.4 (currently
available on the development branch).

A few more observations: the Gastzimmer SB signal level of -67 to -69
dbm is pretty low. Logitech at the time commented (in the code) that 20
db of SNR would be ok, but perhaps not in that fierce environment to
which your radio is exposed. The 11:55:26 entry (line 12) shows a 14
competing access points, 5 on 2412 mHz alone, with signal levels close
to radio's AP. That list, sorted by frequency (channel) with unessential
info removed is below with the columns being frequency, internal signal
level measure, and SSID.

2412 188 Franky24 
2412 184 ABCDE 
2412 177 BREITBAND-430D 
2412 173 cbx-69595 
2412 169 tle-53241 
2432 171 HUAWEI_H122_40BC 
2437 166 PVK
2437 173 Little Hawaii 
2442 171 dlink-7B08
2442 170 dlink-7B08 
2462 177 BREITBAND-A533 
2462 170 Android piotr
2472 184 FrankyGast
2472 183 Franky2

That's a lot, especially with a low signal level, and the neighbors
similar levels. Try to boost your the signal level. My worst signal
level is -52 dbm, the best is -35 dbm. My worst radio has a -51 dbm
signal level, and is closest to the neighbors on the south. (This data
is from the :8080 web page.)

Try the 0.7.4 wlanpoke.sh in the GitHub development branch. One of its
goals is to keep the player UI reassured that the network is working.
This is the start of that requirement, so further improvements are
expected. The hope is that keeping jive "happy" will return the script
to "never fail" status, possibly with tweaking the detection
parameters.

A 12 hour test run of 0.7.4 (.4) last night at this now quieter
environment has yielded just a few (18) failures over 7 radios,
requiring 11 hard resets, and 18-11 7 successful quick resets. The
radios kept playing, and the network icon presumably did not turn red
during the quick resets, so this data is inconclusive. Too quiet hier.
Please test. Thank you.


------------------------------------------------------------------------
POMdev's Profile: http://forums.slimdevices.com/member.php?userid=70558
View this thread: http://forums.slimdevices.com/showthread.php?t=109953

_______________________________________________
Radio mailing list
[email protected]
http://lists.slimdevices.com/mailman/listinfo/radio

Reply via email to