Feeder Has Odd Dropouts of ADS-B Tracks and CPU Utilization | Official ADS-B Exchange Group | Page 1

rotund vault Jan 12, 2026, 2:45 PM

#

My feeder is a recent addition and the first one I have assembled, UID QLvS_Der7bJk. It is built from a Pi 4, Nooelec Flycatcher, Nooelec SAWBird+, and self built antennas with wired ethernet. I had been running the antennas initially in the basement for about the first week, moved the 1090 to the first floor for about a week, and then both into the attic during Jan. 11th. Overnight, my feeder experienced a complete halts of all CPU activity that required hard power of and on reset to get going again. I had tried a "save and restart" via the http://local_ip_addr/config/ link, but it was not enough to restore feed status. There were 2 short halts before I went to sleep and the feeder stopped overnight while I was asleep for 3 hours and again required a hard power reset.

Since that reset, I have made sure to close all local feeder web server pages and to not tax the Pi. I think the issue has to do with keeping multiple instances of the http://local_ip_addr/tar1090/ page open on multiple devices in my house. Overall CPU utilization less than 10%. Keeping the local page serving traffic to the Pi at minimums seems to allow the feeder to keep running.

What I am seeing now are odd dropouts of ADS-B 1090 Tracks and ADS-B CPU Utilization at about 20 minute intervals. I spoke too soon, while I have been writing this, the feeder completely halted again. With the power off reset, the feeder seems to have lost 9 hours of Performance Graph data.

I have a lot of screen capture and some log data captured. What would be most useful for me to post for diagnosing this trouble?

Thank you!

orchid wadi Jan 12, 2026, 3:34 PM

#

The first question we always ask in cases like this is; what power supply are you using for the Pi? If it's not an official Pi one, or at least one designed for the purpose (i.e. not a phone charger, etc), it could cause issues.

Assuming the ADSBx pi image, graphs data is only written to the SD card once a day, to minimise "wear", so it's normal to lose data since the last write in the event of an "ungraceful shutdown".

rotund vault Jan 12, 2026, 4:10 PM

#

It is a qualified 3A wall wort that was previously supplied to me with a rather power hungry LTE hotspot. I did measure the voltage supplied amperage draw by the Pi and receivers. From the current System Info...

#

Voltage Drops:

Status: 0x0
Undervolted:
Now: NO
Run: NO
Throttled:
Now: NO
Run: NO
Frequency Capped:
Now: NO
Run: NO
Softlimit:
Now: NO
Run: NO

CPU temp=54.5'C
Redline=80'C

#

The dropouts and CPU halting only started when I moved the 1090 & 978 antennas to their final intended location. The activity level should not have been a problem overnight as traffic seems to be much less at night.

orchid wadi Jan 12, 2026, 4:22 PM

#

Load should never make it completely lock up, although I'm not entirely sure what you mean by "complete halt". You said you "Save & restart services" ... if it was completely halted, how did you even get to that page? Anyway, it sounds like a hardware issue.

rotund vault Jan 12, 2026, 4:25 PM

#

#

Complete CPU halt.

#

Generally, I try to use 'sudo shutdown -h now', but the time I did not, I lost 6 hours of local log data. Understandable, but it shows my growing frustration.

#

Dseven makes a good point, it cannot be a complete CPU halt as the Pi responds to SSH. As a feeder, it is doing nothing at those times.

#

tar1090.service log has things like this in it that seem to coincide with the 20 minute cycle.
Jan 12 16:10:46 adsbexchange tar1090[647]: No aircraft.json found in /run/readsb during the last 30 seconds! Try restarting dump1090 or reinstalling tar1090 if you switched dump1090 to readsb!
Jan 12 16:10:50 adsbexchange tar1090[647]: Found aircraft.json in /run/readsb, continuing operation as per usual!
Jan 12 16:10:50 adsbexchange tar1090[17691]: /usr/local/share/tar1090/tar1090.sh: line 104: 978.json: No such file or directory
Jan 12 16:11:00 adsbexchange tar1090[647]: No aircraft.json found in /run/readsb during the last 30 seconds! Try restarting dump1090 or reinstalling tar1090 if you switched dump1090 to readsb!

#

The short (recoverable) dropouts seem to coincide with the big spikes of Disk I/O and the above message in the tar1090.service log.

orchid wadi Jan 12, 2026, 4:46 PM

#

Might be worth trying a new SD card too. Assuming you're not doing anything else on the Pi4, this doesn't seem normal, so still suspecting a hardware issue of some sort.

rotund vault Jan 12, 2026, 4:50 PM

#

Sigh! I had a tough time getting 978 to work. Yes, this Pi is dedicated to only ADS-B/UAT feeding. The hesitation is rebuilding the software stack from scratch. For some that is trivial, not so trivial for me. How do I keep from causing a new feeder identity?

orchid wadi Jan 12, 2026, 4:52 PM

#

Keep a copy of /boot/adsbx-uuid. It shouldn't be complicated to setup - the whole purpose of the image is to make it simple. I've never used a Flycatcher, but it looks like its pretty much two SDRs (with LNAs?) duct-taped together?

rotund vault Jan 12, 2026, 5:03 PM

#

Yes, the Flycatcher is only on the Pi-Hat format to make it stupid simple to mount. It still has and requires separate USB power cables to each of 1090 and 978. Back to SSH to find the UUID file. Thank you. I'll report back to this thread when I get that accomplished.

fierce radish Jan 15, 2026, 2:27 PM

#

I have had nothing but problems with even the "official" pi wallwarts, I recomend a 25W PS, I got mine from aliexpress... 5V 5A 5$. 😄

rotund vault Jan 22, 2026, 7:03 PM

#

orchid wadi Keep a copy of `/boot/adsbx-uuid`. It shouldn't be complicated to setup - the wh...

I seem to have it put back together under the original feeder ID but with a replaced CD card. I seem to have everything working, except UAT/978 data in /graphs1090/. UAT/978 data is reporting. Feeder ID = QLvS_Der7bJk

orchid wadi Jan 22, 2026, 7:05 PM

#

rotund vault I seem to have it put back together under the original feeder ID but with a repl...

There are two types of feeder ID - the full UUID (the one you saved), and the short feeder ID, which is a reduction of it. Is it actually causing a problem?

rotund vault Jan 22, 2026, 7:10 PM

#

orchid wadi There are two types of feeder ID - the full UUID (the one you saved), and the sh...

I reported the short ID to make it easier to find the statistics. I did use the UUID from the old SD card and put it on the new SD card. I had to Nano edit the long ID into the file as I did not have permissions set to FTP the old UUID file into the new card. I don't believe either is causing a problem. The statistics seem to be updating on the ADS-B Exchange website for both 1090 and 978 data and under the expected short feeder ID.

orchid wadi Jan 22, 2026, 7:12 PM

#

oh, so you don't have 978 at all in graphs1090, or graphs with no data?

rotund vault Jan 22, 2026, 7:13 PM

#

orchid wadi oh, so you don't have 978 at all in graphs1090, or graphs with no data?

The first part - None of the 978 graphs are shown. Not even place holders in graphs1090.

orchid wadi Jan 22, 2026, 7:14 PM

#

hmm. I don't have 978, but, so less experience with it, but... maybe if you just enabled it, graphs1090 might need a restart to pick it up, or something. Reboot might do it?

rotund vault Jan 22, 2026, 7:16 PM

#

I'll try another reboot. Seems like I have done many. To me it feels like an enable statement is missing in the graphs1090 config, but I have not found the file I think I am looking for.

orchid wadi Jan 22, 2026, 7:24 PM

#

if that doesn't do it, show the output of grep 978 /etc/collectd/collectd.conf

rotund vault Jan 22, 2026, 8:10 PM

#

Reboot (actually a shutdown, power off, power on and normal startup) has not given me 978 graphs.

#

The grep command only returned "#URL_978 http://localhost/skyaware978"

orchid wadi Jan 22, 2026, 8:11 PM

#

OK... does /run/adsbexchange-978/aircraft.json exist?

rotund vault Jan 22, 2026, 8:13 PM

#

yes

orchid wadi Jan 22, 2026, 8:14 PM

#

then I would try running sudo bash /usr/share/graphs1090/install.sh

rotund vault Jan 22, 2026, 8:18 PM

#

Odd result...

#

pi@adsbexchange:/ $ sudo bash /usr/share/graphs1090/install.sh
16406 (process ID) old priority 0, new priority 10
/ /
remote: Enumerating objects: 47, done.
remote: Counting objects: 100% (47/47), done.
remote: Compressing objects: 100% (22/22), done.
remote: Total 27 (delta 18), reused 9 (delta 5), pack-reused 0 (from 0)
Unpacking objects: 100% (27/27), 14.02 KiB | 398.00 KiB/s, done.
From https://github.com/wiedehopf/graphs1090

branch master -> FETCH_HEAD

639ac71...c6cf415 master -> origin/master (forced update)
HEAD is now at c6cf415 airspy: DF counts, add DF20 somehow this got missed
/

Install in progress, this shouldn't take longer than a minute or two ..........
/usr/share/graphs1090/install.sh: line 161: Plugin: No such file or directory
[ERROR] Error in line 161 when executing: face "'$iface'"' /etc/collectd/collectd.conf && continue
# only add interface starting with et en and wl
case $iface in
et*|en*|wl*)
sed -i -e '/ "interface" Interface "'$iface'" < Plugin > /{a
pi@adsbexchange:/ $

GitHub

GitHub - wiedehopf/graphs1090: Graphs for readsb / dump1090-fa / du...

Graphs for readsb / dump1090-fa / dump1090 (based on dump1090-tools by mutability) - wiedehopf/graphs1090

#

Just in case something useful happened, I rebooted. Still no 978 graphs.

orchid wadi Jan 22, 2026, 8:26 PM

#

it's like the script got corrupted somehow

#

what's the output from grep -C3 Plugin /usr/share/graphs1090/install.sh (please enclose it in triple backticks for readability)

rotund vault Jan 22, 2026, 8:38 PM

#

Not sure what you meant by "backticks". The output in the native SSH under Windows is no joy to read.

orchid wadi Jan 22, 2026, 8:39 PM

#

I mean ``` then the output then ``` again

rotund vault Jan 22, 2026, 8:40 PM

#

is this right?'''

#

'''pi@adsbexchange:~ $ grep -C3 Plugin /usr/share/graphs1090/install.sh
# only add interface starting with et en and wl
case $iface in
et*|en*|wl*)
sed -i -e '/<Plugin "interface">/{a
Interface "'$iface'"
}' /etc/collectd/collectd.conf
;;
pi@adsbexchange:~ $'''

orchid wadi Jan 22, 2026, 8:41 PM

#

those are forward ticks - backtick is top-left corner of the keyboard (usually)

rotund vault Jan 22, 2026, 8:41 PM

#

Yes, finally figured it out - ```

#

    # only add interface starting with et en and wl
    case $iface in
        et*|en*|wl*)
sed -i -e '/<Plugin "interface">/{a\
    Interface "'$iface'"
}' /etc/collectd/collectd.conf
        ;;
pi@adsbexchange:~ $```

orchid wadi Jan 22, 2026, 8:42 PM

#

that's the one 👍

#

That looks normal. The error you got is bizarre. Trying to think what would cause that....

#

but I reproduced it.... investigating....

#

meanwhile, try this instead: sudo bash -c "$(curl -L -o - https://github.com/wiedehopf/graphs1090/raw/master/install.sh)"

rotund vault Jan 22, 2026, 8:46 PM

#

  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0
100 11109  100 11109    0     0  38233      0 --:--:-- --:--:-- --:--:-- 38233
14302 (process ID) old priority 0, new priority 10
/home/pi /home/pi
remote: Total 0 (delta 0), reused 0 (delta 0), pack-reused 0 (from 0)
From https://github.com/wiedehopf/graphs1090
 * branch            master     -> FETCH_HEAD
HEAD is now at c6cf415 airspy: DF counts, add DF20 somehow this got missed
/home/pi
------------------
Install in progress, this shouldn't take longer than a minute or two ..........
---------
write reducing measures already enabled, no need to do anything for this script!
---------
--------------
--------------
All done! Graphs available at http://192.168.1.45/graphs1090
It may take up to 10 minutes until the first data is displayed
pi@adsbexchange:~ $```

orchid wadi Jan 22, 2026, 8:47 PM

#

Good... now grep 978 /etc/collectd/collectd.conf again?

rotund vault Jan 22, 2026, 8:47 PM

#

UAT Graphs Now Report! Outstanding! Thank you!

orchid wadi Jan 22, 2026, 8:47 PM

#

Sweet! 👍

rotund vault Jan 22, 2026, 8:48 PM

#

No data in them, yet. Cautiously optimistic.

#

Data has begun. Nice!

#

It will take a few days to see if I have achieved the goal of stability. The original trouble I eventually realized that the whole feeder software was halting and exiting. Very few clues as to why.

rotund vault Jan 22, 2026, 8:52 PM

#

orchid wadi Good... now `grep 978 /etc/collectd/collectd.conf` again?

Odd, same grep result

#

URL_978 "file:///usr/share/graphs1090/978-symlink"
pi@adsbexchange:~ $```

orchid wadi Jan 22, 2026, 8:53 PM

#

That's not the same 😉

#

before it was #URL_978 http://localhost/skyaware978

#

that's "commented out", so doesn't do anything - it's the default when there's no 978 data found

rotund vault Jan 22, 2026, 8:54 PM

#

You're right. I missed the minor change - details are everything in programming

orchid wadi Jan 22, 2026, 8:54 PM

#

indeed!

rotund vault Jan 22, 2026, 8:56 PM

#

Is your feeder dseven_wl ?

orchid wadi Jan 22, 2026, 8:56 PM

#

yes

rotund vault Jan 22, 2026, 8:56 PM

#

It does well

orchid wadi Jan 22, 2026, 8:57 PM

#

about as well as I can do from this location. I'm at sea level, in a low traffic area, with hills most of the way around me

rotund vault Jan 22, 2026, 9:00 PM

#

Outside antenna?

orchid wadi Jan 22, 2026, 9:02 PM

#

Yes, colinear on the roof of an outbuilding

rotund vault Jan 22, 2026, 9:04 PM

#

My 1090 and 978 are separate, simple ground planes and in the attic. The only way I can put any outside would be to hide them in a tree. Not optimum either way.

orchid wadi Jan 22, 2026, 9:05 PM

#

1090 spider is tiny - shouldn't be too noticable.... weatherproofing may be an issue, though

rotund vault Jan 22, 2026, 9:07 PM

#

For now, I am hoping it is stable. I have another project coming up where I plan to build another feeder into a large RV (caravan?) I will make sure to add a USB GPS to take advantage of the MLAT switching (off when moving.)

orchid wadi Jan 22, 2026, 9:08 PM

#

Cool. Fingers crossed on the stability 🤞

rotund vault Jan 22, 2026, 9:08 PM

#

Thanks again for your help!

#Feeder Has Odd Dropouts of ADS-B Tracks and CPU Utilization