[aprssig] Weird Behavior on the APRS-IS

[Stephen H. Smith]

On 2/17/2014 6:45 PM, Steve Dimse wrote:
>
> On Feb 17, 2014, at 6:27 PM, Stephen H. Smith <WA8LMF2 at aol.com> wrote:
>>
>> Normally, one assumes that whenever anything is fed into an APRS-IS server, it propagates to all the others within a second or two.   Here I was seeing a nearly 20-MINUTE delay.
>
> Were you only using the ISS position as the indicator of a 20 minute delay? Perhaps someone is feeding a bad position based either on clock offset or old keplerian elements.
>
> I've been chasing a problem with a 45 second delay but it seems to only involve the CWOP weather data. A 20 minute delay would not be impossible (say from a saturated digi that holds data for long periods), but it is much more likely that someone is just generating bad positions.
>
> I never trust the ISS position on the APRS IS because I've seen lots of bad data over the years. findU runs its own orbital code to generate ISS position for www.ariss.net, it also generates a +5 minute and +10 minute position so you can see where it is heading, which I find a lot more useful than the current position anyway!
>
> Steve K4HG
>
>

This has nothing to do with digipeaters


The ISS object I am using is generated by KJ4ERJ-15 and is injected directly 
into  fourth.aprs.net.    In turn, I have been taking the same object off two 
other servers to plot on two separate displays.

On one map (the 30-meter HF APRS display), where I am also plotting the NWS 
watches and warnings, I am taking the feed off  firenet.us  in order to get the 
full NWS feed.

On the satgate map, I am taking the feed off a "normal" APRS-IS server to avoid 
unnecessary loading on the firenet servers.    (I have two separate connections 
because the port 14580 filtering requirements are different for the two 
applications; else I would re-distribute a single connection to both apps.)

Normally the station is received within a second or two of the same time, in 
the same place, on both maps,as one would expect.  This is why the TWENTY 
MINUTE delay in the appearance of the identical object on second is so strange.

This is vaguely reminiscent of the KPC3 long-delayed-packets receive bug that 
has been discussed so many times, but this one is only related to Internet 
server-to-server connections.


[If I had a Windows app that could generate the ISS track locally from keps, 
and emulate an APRS-IS server on   localhost:portwhatever  I wouldn't bother 
with the Internet ISS feed at all.]