[aprssig] aprx bug continues!
hessu at hes.iki.fi
Sun Mar 9 17:47:29 CDT 2014
On Sun, 9 Mar 2014, Steve Dimse wrote:
> Today there are a bunch more of big spikes in APRS-IS volume, after a
> bit of detective work each spike is being caused by a single station.
> mysql> select count(*) from raw where `call` = "YB6HJE-3" and time_rx > 20140309000000;
> | count(*) |
> | 174914 |
The spikes are quite massive, over 70000 packets per hour:
> Matti tells me this is from a feature in his program that sends out data
> that was missed while his program was sleeping, though today's spikes
> make me question that - they seem too close together to have generated
> that much data.
The problem was in the timer code handling - if the clock stops and then
jumps forward a bit due to a sleep, there will be a small spike.
If the clocks goes back to year 1996 (or whatever year your computer's
BIOS was made) due to a dead clock backup battery, and a little power
failure, aprx starts, and then NTP brings it forward to today, there will
apparently be a large spike.
I suppose this could also hit gateways running on systems without a
battery-backed real-time clock, like a Raspberry Pi or an OpenWRT-based
> I've tried multiple times to contact YB6HJE through the QRZ email. If
> anyone has a means to contact him please try. And don't use aprsx on any
> computer that has a sleep mode!
The name of the program is actually 'aprx'.
It'd probably be best to ugprade to the most recent version, aprx 2.08
(svn558/559), which already contains the fixes for the bug.
I just sent twitter & facebook messages to YB6HJE. At least the FB account
seems active, he's updated it just yesterday.
More information about the aprssig