New Anomalies Messages

[beckerm13]

Witht the new anomalies messages showing on the ADS-B home page, I am seeing a message that my RPi (B+ Model) CPU is running near 100%. Looking at the top command output, I see that Dump1090-Mut and the FA MLAT Client are taking up about 82% of CPU. Below is the top output. Should Dump1090-Mut and the FA MLAT Client be using this much CPU?

This my secondary RPi that has its antenna in the window. My primary RPi is an older B Model that runs regular Dump1090 and no MLAT, with the antenna in the attic. It sees about 300 more aircraft and about 100,000 more positions a day that the secondary B+ Model. The I have been thinking of upgrading the B Model to a spare B+ Model. However, after seeing how much the current B+ Model working with Dump1090-Mul and MLAT on fewer aircraft and positions, I am reluctant to upgrade it.

Thoughts?

Marty

  PID USER      PR  NI  VIRT  RES  SHR S  %CPU %MEM    TIME+  COMMAND
 2564 pi        20   0 22284  13m 1944 R  46.9  3.0 523:54.28 dump1090-mutabi
 2805 root      20   0 12748 9872 5556 R  36.0  2.2 290:53.88 fa-mlat-client
 2316 root      20   0 45452 2976 2156 S   8.3  0.7  65:24.33 pfclient
 2778 root      20   0 18900 8240 4820 R   4.8  1.8  31:15.73 piaware
 7038 pi        20   0  4676 2488 2152 R   1.3  0.6   0:00.12 top
    3 root      20   0     0    0    0 R   0.6  0.0   4:02.31 ksoftirqd/0
 2803 root      20   0  2656 2136 1876 S   0.6  0.5   6:43.78 faup1090
    7 root      20   0     0    0    0 S   0.3  0.0   4:10.59 rcu_preempt
 6983 root      20   0     0    0    0 S   0.3  0.0   0:01.94 kworker/0:0
 7012 pi        20   0  9436 3680 3068 S   0.3  0.8   0:00.11 sshd

[toverfield]

The B+ is no faster than the old B, so that won’t help with the overloaded CPU. I don’t know if software speedups are on the way, but I upgraded to the Pi 2 to solve this problem. The Pi 2 is very much faster. It completely fixed this problem.

[obj]

@beckerm13

Should Dump1090-Mut and the FA MLAT Client be using this much CPU?

Yes that sounds about right, especially given your location. In busy US airspace with mostly Mode S only traffic, you are not going to fit much other than dump1090/piaware/mlat on a B/B+.

The mlat client could use some improvement, it doesn’t really need to be taking that much CPU, but there is some heavy lifting involved (it’d need a rewrite into C) and there is a relatively simple workaround (use a Pi 2 or move other stuff off the Pi) so it is a low priority.

[rjankowski]

Weird, I am not having that high of utilization.

top - 15:45:25 up 4 days, 21:10,  1 user,  load average: 0.03, 0.04, 0.05
Tasks: 115 total,   2 running, 113 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.5 us,  0.2 sy,  0.0 ni, 99.3 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
KiB Mem:   4047464 total,  3616824 used,   430640 free,   194976 buffers
KiB Swap:  4194300 total,        0 used,  4194300 free.   553232 cached Mem

  PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
25823 dump1090  15  -5  118176  11032   1428 S   2.3  0.3  11:22.12 dump1090-mutabi
    7 root      20   0       0      0      0 R   0.3  0.0   8:13.19 rcu_sched
   10 root      20   0       0      0      0 S   0.3  0.0   5:56.44 rcuos/2
    1 root      20   0   33504   2808   1472 S   0.0  0.1   0:01.90 init
    2 root      20   0       0      0      0 S   0.0  0.0   0:00.25 kthreadd
    3 root      20   0       0      0      0 S   0.0  0.0   0:02.23 ksoftirqd/0
    4 root      20   0       0      0      0 S   0.0  0.0   0:00.00 kworker/0:0
    5 root       0 -20       0      0      0 S   0.0  0.0   0:00.00 kworker/0:0H

All kidding aside, moving off of my B and onto a real machine (even though it’s a VM) has really improved the performance of piaware. I know there are other boards out there and I was actually going to look to see if the Intel Edison types of boards can be adapted for piaware. If I do pick one up to replace one of my Arduinos, I’ll try to get it to work to see if I can find a performance difference. They’re not a whole lot better (if at all) but I wonder if the x86 architecture would be more performance over a ARM.

[beckerm13]

As a comparison, this is the top output from my B Model, that uses regular Dump1090 and no MLAT, but processes more aircraft and positions.

Marty

  PID USER      PR  NI  VIRT  RES  SHR S  %CPU %MEM    TIME+  COMMAND
 2085 pi        20   0 15784 6056  900 R  47.4  1.4 267:16.31 dump1090
 2254 root      20   0 45448 2020 1124 S  10.1  0.5  48:41.39 pfclient
 2074 pi        20   0 12984 7516 3156 S   6.2  1.7  15:06.95 python
 2420 root      20   0  2404 1012  712 S   1.0  0.2   4:40.68 faup1090
 3081 pi        20   0  4672 1352 1020 R   1.0  0.3   0:00.23 top
    7 root      20   0     0    0    0 S   0.3  0.0   1:27.65 rcu_preempt
 1649 root      20   0  1752  512  428 S   0.3  0.1   0:05.60 ifplugd
 3056 root      20   0     0    0    0 S   0.3  0.0   0:00.16 kworker/0:0
    1 root      20   0  2148  720  616 S   0.0  0.2   0:02.91 init
    2 root      20   0     0    0    0 S   0.0  0.0   0:00.00 kthreadd
    3 root      20   0     0    0    0 S   0.0  0.0   1:14.74 ksoftirqd/0
    5 root       0 -20     0    0    0 S   0.0  0.0   0:00.00 kworker/0:0H
    6 root      20   0     0    0    0 S   0.0  0.0   0:02.65 kworker/u2:0
    8 root      20   0     0    0    0 S   0.0  0.0   0:00.00 rcu_bh
    9 root      20   0     0    0    0 S   0.0  0.0   0:00.00 rcu_sche

[toverfield]

@rjankowski

I wonder if the x86 architecture would be more performance over a ARM.

No need at all. The Pi 2 is much faster than necessary for this.

top - 15:05:00 up 1 day, 13:54,  1 user,  load average: 0.34, 0.37, 0.40
Tasks:  95 total,   1 running,  94 sleeping,   0 stopped,   0 zombie
%Cpu(s):  8.1 us,  1.5 sy,  0.0 ni, 90.0 id,  0.0 wa,  0.0 hi,  0.3 si,  0.0 st
KiB Mem:    948252 total,   927516 used,    20736 free,    51456 buffers
KiB Swap:        0 total,        0 used,        0 free,   751136 cached

  PID USER      PR  NI  VIRT  RES  SHR S  %CPU %MEM    TIME+  COMMAND
 1995 dump1090  15  -5 23236  13m 1868 S  28.4  1.4 605:31.26 dump1090-mutabi
 2530 root      20   0 11224 9524 5232 S   5.3  1.0  88:31.56 fa-mlat-client
 2145 root      20   0 17456 7940 4536 S   1.6  0.8  20:41.24 piaware
 2165 root      20   0 73948 6232 3240 S   1.6  0.7  23:03.87 fr24feed
 2423 root      20   0  2656 1588 1420 S   0.7  0.2   8:07.50 faup1090

[beckerm13]

@obj

Yes that sounds about right, especially given your location. In busy US airspace with mostly Mode S only traffic, you are not going to fit much other than dump1090/piaware/mlat on a B/B+.

Yes, the Chicago is a busy area. Looking at the FA statistics for US only users, I am 6th in the US. There is one other user ahead of me here in Chicago and one ahead of me who is just north of here in Madison WI. The other two top 5 US users are on the east coast.

Marty

[obj]

@rjankowski

(even though it’s a VM)

Be aware that sometimes there are problems with VMs (vmware in particular seems to have trouble) where they can silently drop USB traffic. This isn’t a big deal for ADS-B, you just lose a few messages, but it’s disastrous for mlat.

[rjankowski]

@obj

@rjankowski

(even though it’s a VM)

Be aware that sometimes there are problems with VMs (vmware in particular seems to have trouble) where they can silently drop USB traffic. This isn’t a big deal for ADS-B, you just lose a few messages, but it’s disastrous for mlat.

Guess I’ll find out sooner or later.

[beckerm13]

I decided to turn off my MLAT due to the high load it puts on the RPi B+.

Marty

[dblake5356]

@beckerm13

Witht the new anomalies messages showing on the ADS-B home page, I am seeing a message that my RPi (B+ Model) CPU is running near 100%. Looking at the top command output, I see that Dump1090-Mut and the FA MLAT Client are taking up about 82% of CPU. …

Marty

Does anyone know if over-clocking the RPi might help reduce the CPU load?

[LitterBug]

@dblake5356

@beckerm13

Witht the new anomalies messages showing on the ADS-B home page, I am seeing a message that my RPi (B+ Model) CPU is running near 100%. Looking at the top command output, I see that Dump1090-Mut and the FA MLAT Client are taking up about 82% of CPU. …

Marty

Does anyone know if over-clocking the RPi might help reduce the CPU load?

I had overclocked my B+ but was still running over 100%. Switched to a Pi2 and am purring along fine now.

LitterBugs

[beckerm13]

With my MLAT now turned off, I was still getting the Anomalies message that the CPU was near 100%. When I checked the top output, now Dump0190-Mut is using around 75%.

Marty

  PID USER      PR  NI  VIRT  RES  SHR S  %CPU %MEM    TIME+  COMMAND
 2564 pi        20   0 23452  13m 1944 R  74.3  3.0   1378:33 dump1090-mutabi
 2316 root      20   0 45452 3220 2156 S  20.1  0.7 210:45.99 pfclient
 8302 root      20   0 18900 8292 4920 S   1.6  1.9  17:07.28 piaware
 8328 root      20   0  2656 2088 1840 S   1.3  0.5   9:40.36 faup1090
13371 pi        20   0  4676 2472 2136 R   1.0  0.6   0:00.85 top
    3 root      20   0     0    0    0 S   0.3  0.0   9:19.33 ksoftirqd/0
    7 root      20   0     0    0    0 S   0.3  0.0  10:04.64 rcu_preempt
   43 root      20   0     0    0    0 S   0.3  0.0   0:02.83 jbd2/mmcblk0p6-
 1747 root      20   0     0    0    0 S   0.3  0.0   3:26.28 RTW_CMD_THREAD
13348 root      20   0     0    0    0 S   0.3  0.0   0:00.36 kworker/0:0
    1 root      20   0  2152 1308 1204 S   0.0  0.3   0:07.84 init
    2 root      20   0     0    0    0 S   0.0  0.0   0:00.06 kthreadd

[dblake5356]

@beckerm13

With my MLAT now turned off, I was still getting the Anomalies message that the CPU was near 100%. When I checked the top output, now Dump0190-Mut is using around 75%.

Marty

If this just started to happen, to help trouble shoot the issue, you might try swapping your SD cards between your two RPis to see what happens. Maybe your SD card or maybe even your RPi has a problem and is getting ready to die. Then of course it might be a power supply issue.

[obj]

@beckerm13

With my MLAT now turned off, I was still getting the Anomalies message that the CPU was near 100%. When I checked the top output, now Dump0190-Mut is using around 75%.

Most of the time I’ve seen that it’s because the dongle gain is set too high and dump1090 is spending a lot of time trying to decode loud noise.
Normal CPU use for dump1090-mutability with --oversample --phase-enhance is somewhere around 40-50% of a B/B+

[beckerm13]

@obj

@beckerm13

With my MLAT now turned off, I was still getting the Anomalies message that the CPU was near 100%. When I checked the top output, now Dump0190-Mut is using around 75%.

Most of the time I’ve seen that it’s because the dongle gain is set too high and dump1090 is spending a lot of time trying to decode loud noise.
Normal CPU use for dump1090-mutability with --oversample --phase-enhance is somewhere around 40-50% of a B/B+

Right now the gain is at 44. I dropped it to 42 and there was no major change.

Marty

[GDoggRaspberry]

@beckerm13

@obj

@beckerm13

With my MLAT now turned off, I was still getting the Anomalies message that the CPU was near 100%. When I checked the top output, now Dump0190-Mut is using around 75%.

Most of the time I’ve seen that it’s because the dongle gain is set too high and dump1090 is spending a lot of time trying to decode loud noise.
Normal CPU use for dump1090-mutability with --oversample --phase-enhance is somewhere around 40-50% of a B/B+

Right now the gain is at 44. I dropped it to 42 and there was no major change.

Marty

Is there a reason you’re not using the AGC (gain -10) setting? Are you getting better performance by specifying the gain?

[obj]

@GDoggRaspberry

Is there a reason you’re not using the AGC (gain -10) setting? Are you getting better performance by specifying the gain?

For ADS-B, the tuner AGC is ineffective and basically works like a fixed gain setting of around ~52dB. It does not adjust to the incoming signal level.

[triggers]

Before I splash out on a Cray XC40 Have any of you guys chained a couple of Pis together to share the CPU load? I now have 2 B+ which are on the upper limit of CPU. I don’t think I can get a Pi 2 past the wife !!

[sjacket99]

A Pi cluster would be cool. Like this one…

The RPiCluster: youtu.be/i_r3z1jYHAc