Spiky upload bandwidth when using Piaware upload client.


#1

Using the performance graphs built into my dump1090-mutability release, I find strangely spiky output bandwidth from feeding through the Piaware package. I’d have thought I would see a more average kind of internet bandwidth being used rather than this odd burst-like usage. Any ideas would be very much appreciated.

This graph covers a week of Internet upload and download activity.

https://lh3.googleusercontent.com/-iEs94wFQXbM/V53inisa_eI/AAAAAAAAGfI/CrHUJMWgyKkzngSnAlUc2qkmXQfjigR2gCCo/s577/piaware%2Bbandwidth%2Bout.png

Obviously, I’d expect peaks and troughs to reflect the air traffic conditions through the days, but these peaks look disproportionately large to me.

I have Mlat enabled and I also upload to Planefinder, but they only take a small amount of positional data because they don’t do Mlat from setups like mine.


#2

I doubt that the spikes are from piaware; ADS-B data is on the order of 1-2kB/s and mlat is on the order of 5-10kB/s peak.
You might want to do some more digging to work out exactly where the traffic is coming from.

note that dump1090-mutability does nothing to monitor system network usage, you must mean something else.


#3

Also, check the piaware logs for information about how much traffic mlat is sending, it logs at 15 min intervals:



mlat-client(7811): Server:      0.1 kB/s from server    0.0kB/s TCP to server     0.4kB/s UDP to server



#4

Regular watching the aircraft/stats on your web browser can cause spikes like this.


#5

Thanks much for the replies guys. I suspect that the main possibility for spikes might have been identified by Boab. I need to look and see when these spikes are happening. If they come in the middle of the night, then it isn’t the running of the webserver to watch the traffic. I do play with the thing and sometimes leave it serving the dynamic map pages, so I expect that this is what the data is. I’m not really suspicious that my receiver has been hacked and is communicating with unauthorised bad guys - I have set up three separate receivers based on retired and formatted netbooks that I have acquired and these do this behaviour right out of the box so the chances of them being compromised by malware or other nasties straight off are not very high at all.

I will do some systematic tests recording the times of webpage serving and leave long timed intervals of not serving the pages to confirm the hypothesis.

Thanks again to Obj and Boab.


#6

Yes - after testing, the bog spikes are due as Boab says to watching the webserver in dump1090 for aircraft activity.

Thanks for the help.