Has anything changed recently to the algorithm that selects one’s feeder for the WITHIN THE LAST HOUR data?
While the number of aircraft has drastically reduced lately, I appreciate I don’t expect to see many flights listed on the WITHIN THE LAST HOUR table, but a quick look at my nearby sites here in the UK seems anomalous.
Out the 20-odd sites I looked at nearby, most have zero, a couple have one or two, but one has 15!
Even when I look at the top five sites in UK, I’d expect to see more WITHIN THE LAST HOUR table. Perhaps I have some sort of cache issue that is giving a rogue picture or maybe it’s something to do with the latency from my ISP: I haven’t a clue.
Anyway, this is just an observation, not a defect report: it just seems a little strange to me!
I don’t look at those very often but you’re right, my main receiver has twelve which is fewer than I’d expect.
Looking at Nearby Sites, the Last Seen column is a bit odd as well as they’re all showing around fifteen minutes. A few days ago they were all showing an hour.
This has been odd since the outage a few days ago which seemed to affect some feeders and not others. I lost four or five hours but my friend Tony who’s not far from here had no lost data at all.
His feeder type is a PiAware (Radarcape) 3.5.0 whereas mine is PiAware (Debian Package Add-on) 4.0 and has just a handful reported. Normally our figures for aircraft and positions are very close.
My display is also strange. tar1090 shows over 100 aircraft while FA shows 15. The FA entries also show from 5 to 300 seconds. The SkyAware map is missing with only test shown.
I would assume that these values depend if the data from your receiver is used or not.
In case the aircraft is reported by a different receiver, it is not shown in this table under your profile.
I always have a lower number of aircraft reported there compared to my current view.
I fully understand that I would only expect to see a small subset of the flights available at anytime in my view and I am more likely to see nearby aircraft, but in the past the number in my table has typically gone up or down depending on how busy the local airspace is.
All I am trying to say (and I’m in agreement with @keithma) since the network outage my observation is that the number of flights in that table of data is lower than expected, but for some sites it is higher than expected, i.e., the balance has changed.
On the main receiver that @keithma referred to, a few minutes before 08:00 when I would expect it to be quite busy (after all that site is the top in the UK) there were just two entries WITHIN THE LAST HOUR ! The number has since risen, but I would expect it to go up and down dynamically, it is only natural.
What I can’t get my head around is when two sites nearby which generally see the same traffic are now not showing similar proportions of the flights in that table. This has only happened since the network outage and is possibly some knock on effect. If it is not due to that event then I would suspect an algorithm change.
Have you restarted your feeder after the outage or at least restarted Piaware service?
I had a lower overall aircraft reporting the first few hours after the service came back. And it was back to normal after i restarted piaware.
I don’t know if this was related, but it did not hurt.
I’m pretty certain I restarted both my feeders, but it wouldn’t harm to do it again.
I’ll hold off until the end of the day since the situation may recover anyway.
I’ve noticed a few sites that are getting very low numbers in that table, for reasons unknown.
(if you have IPv6, you may have been less affected as IPv6 connectivity was less affected and we added IPv6 addresses to the piaware DNS name partway through the outage to try to mitigate the problems)
I also noticed during the outage that name webfarm-eu.flightaware.com changed IP from 70.42.6.251 to 70.42.6.250 - the former IP now resolves to lb-1.hou.flightaware.com according to my DNS.
Beginning to wonder if I have a stale DNS?
@obj
As I have IPv6 support from my ISP, what name should I use on my feeders to use in preference to the current IPv4 default? I would like to try it.
piaware.flightaware.com currently has both v4 and v6 addresses piaware-ipv6.flightaware.com is v6 only
There’s currently no v4-only name (though it might make sense to add one)
It’s always misleading to think there is a defect without concrete evidence, but just a quick sample of my own nearby sites within 3 miles of me, 4 have NO flights for the “within last hour”. Strange to say the least!
No doubt by the time this is read the situation will have changed but wouldn’t this be ironic if it turns about to be a browser rendering issue and the table just isn’t getting displayed rather than there being no flights within the last hour.
One thing definitely is that the “last seen” on the nearby sites is also less frequent than in the past (well for me at least). I’m wondering if some sites have gone onto some sort of “slow poll”, while others are still acting normally?
I have reason to believe this is a rendering issue:
Looking at a nearby site that has NO “within last hour” data i.e., the table just doesn’t appear in my version of Chrome, but I see contradictory evidence in the Dev Tools.
This suggests there are 19 flights under the “considered” key…
But without more knowledge about how this is meant to work, I have no idea why the table is not showing the flights. Hopefully more knowledgeable people can comment.
Something’s changed I’m convinced: maybe I need to try a different browser!
Just throwing a theory out there (well i’m actually quite certain that’s how it works):
You can click on a track log of each flight and each point will be attributed to exactly one feeder (but you can only see the exact station if it’s your station otherwise it’s hidden).
I’m pretty sure “Flights with positions from this feeder” only include those flights where the tracklog of the flight contains a point attributed to your station.
With less flights and lots of feeders this will favor people with big range.
Areas with lots of receivers will see a significantly reduced number for each receiver.
That statistic has never shown ALL flights you see.
You’ll tend to see more flights there if you have more “unique” coverage (coverage where there are few receivers that can see there), but other than that it’s fairly hard to say how many flights “should” be there.
This data presentation actually dates back to a much older way of processing data. It’s actually more complicated now, because we combine data from multiple feeders into a single record that corresponds to the tracklog entry. That is complex to display… the tracklog point still shows only the feeder that first provided the position, even if auxiliary data was integrated into that point from other receivers.
