To add to the puzzle, the feeder also sends its data to another Raspberry Pi that is running the Docker ADSB Feeder Image. graphs1090 on that machine for the same 30 day period does not show the drop in Tracks Seen (Apologies for the different wide/narrow settings.)
For starters I would be finding a convenient time to terminal connect to the base computer and do a “sudo shutdown now” and then power down the Pi and repower shortly thereafter to give it a fresh bottoms up reboot to see if it resets the issue.
I had the system shutdown for about half an hour about 24 hours ago - the issue persists after that.
No I haven’t made any config changes to the feeder or graphs1090 in the time period shown.
The feeder is actually performing well - positions, aircraft seen and range are all unaffected - it is just the reported tracks seen that has changed on tar1090 on my main feeder.
To clarify what I am showing above. My main feeder if a Pi5B running readsb, tar1090 and graphs1090, plus piaware to feed Flightaware. On the same local network I have a Pi4B that is running Docker ADSB Feeder Image which feeds Planefinder, adsb Exchange and a number of other sites. ADSB Feeder on the Pi4 is getting its data from readsb on the Pi5, so the ADSB Tracks Seen chart on the Dark background chart should be the same as the first chart in my post. Each of the three charts should show the same data on both systems.
Its a bit confusing because the wide chart shows days and the narrow dark chart shows week numbers, but week 52 starts on December 23rd, so if the drop was real, it being showing for week 52 on the dark chart.
Do you live near a major airport and the runway assignment has changed due to maintenance or different flight route ? I live near the RWY 27 approach of EHAM and when they use that runway i get a lot less traffic due to the strong signals of the aircraft passing within 1 NM of my feeders. Once the runway assignment is back to 18R/18C/18L it goes back to normal. Due to using Dynamic range in the setup it tries to compensate for the strong signals and it turns the gain down when the RWY 27 approach is in use. Maybe that can clarify what you are seeing ?
No, the nearest major airport is about 40NM away. The traffic hasn’t changed. This is a comparison of my main feeder and my test system for the last two weeks.
The main system consistently sees more aircraft and gets more positions - this hasn’t changed (note the vertical scales on the charts are not the same for each system.) The test system has less Aircraft Seem/Tracked but more ADSB Tracks seen - this wasn’t previously the case.
I think it is a data capture/presentation problem rather than an actual drop in performance.
Have you tried disconnecting the Test system and letting the Main system standalone to see if the data reporting reverts back to normal levels on the Main?
Sorry, as usual, I’m not explaining my issue very well.
My main feeder and test systems are completely independent - different antenna, receiver and raspberry pi. The test system is also in a worse position, so it always receives less aircraft. What I was trying to show was the relativity between the Aircraft Seen/Tracked chart and the Tracks Seen (8 minute exponential moving average) chart. There should be a correlation between the two as seen with the test system.
I’ve fiddled with my graphs1090 settings to try to make what is worrying me clearer. Both of the charts below come from data generated by readsb on my main feeder. The first is from graphs1090 running on the same Raspi as the readsb instance. The second set come from graphs1090 on another Raspi (without a receiver) which is taking its data over the LAN from readsb on the main feeder.
As expected, the charts on the left are close to identical. @wiedehopf is there a reason why the 8 minute exp.moving avg. charts on the right aren’t also the same?
someone asked you to restart which is reasonable, you haven’t done that.
otherwise … check readsb versions / command line parameters.
you could read up on what graphs1090 looks at which is the stats.json produced by readsb.
and read the code, put in stuff to debug it.
but at least look at the json.
