OpenCPN always crashes when using radar_pi with ARPA ON for 2 minutes to over an hour

[ERousseletChorus]

Hi,

I've been using OpenCPN and the plugin Radar Pi to detect ARPA and generate xxTTM messages for months with a Halo20 radar. To this day, OpenCPN ended up crashing every single time after from 2 minutes to over an hour of normal functionning, without me interacting with it in any way after starting the radar.

This is the crash report folder I got from a crash generated this morning: 1 hour and 10 minutes of normal functionning with ARPA detected from 10m to 100m (I'm not at sea right now so I just generate ARPA in a room): crash_report.zip

This is the opencpn.ini file used: opencpn_ini.txt

And this is the logfile generated during this test (from 8h53 to 10h04): opencpn_log.log

I hope that I provided enough information. I have no idea about the cause of this issue. I've reinstalled OpenCPN recently and did not change much in its initial configuration appart from installing the plugins Radar_Pi and NMEA_Converter_Pi directly from the plugin manager in OpenCPN

Thanks for reading me, and have a nice day !

[douwefokkema]

I just had the radar (Halo 20+) running for 4 hours on Windows, no crash. I have ARPA on but no NMEA_Converter_Pi. I will continue for some hours more. Same versions of software as you have. Question: Does it also crash with ARPA off? Does it crash without NMEA_Converter_Pi?

[ERousseletChorus]

I've had the radar running for 2,5 hour this morning with ARPA OFF and NMEA_Converter_Pi enabled with no crash. I just started it with ARPA ON and NMEA_Converter_Pi disabled. Since I don't have a compass, I'm missing the heading parameter, so the radar is not able to detect ARPA in this configuration even if they are set ON.

I'll edit this comment when I have a result (2h30 with no crash or a crash after XXhXX).

EDIT

The radar has be running for 2,5 hours with ARPA ON and NMEA_Converter_Pi disabled with no crash.

Since I can't generate ARPA in this configuration (no heading data), it seems clear that the crashes are somehow related to the aquisition of ARPA and are not caused by NMEA_Converter_Pi.

The two main differences between our tests are the radar used (Halo20 // Halo20+) and the computer used. So, maybe there is a problem of compatibility with the Halo20, or maybe I am missing some .dll file rarely used by OpenCPN on my computer ?

These are all the lines with errors in yesterday's logfile: 08:53:25.801 ERROR radar_pi.cpp:1557 Config Loaded RadarCount=1 08:53:25.801 ERROR radar_pi.cpp:1663 Config Save RadarCount=1 08:53:55.296 MESSAGE RadarInfo.cpp:470 Error ProcessRadarSpoke range is zero 08:53:57.289 MESSAGE RadarInfo.cpp:470 Error ProcessRadarSpoke range is zero

The last two errors likely come from me editing the inner range from 0.1km to 0.01km before starting the radar.

I'll start a run with the configuration of yesterday's crash with a Wireshark capture of all xxTTM sentences to provide more data.

[Hakansv]

One thought: Do you happens to have the PPI window open receiving AIS targets? If so you may try to switch off the PPI window. There's a detected memory leak in O 5.8.4 for that case. (Now corrected in Github master)

[ERousseletChorus]

One thought: Do you happens to have the PPI window open receiving AIS targets? If so you may try to switch off the PPI window. There's a detected memory leak in O 5.8.4 for that case. (Now corrected in Github master)

Yes, I have the parameter AIS/ARPA on PPI ON. I start a test with yesterday's configuration and this parameter OFF immediatly !

[ERousseletChorus]

One thought: Do you happens to have the PPI window open receiving AIS targets? If so you may try to switch off the PPI window. There's a detected memory leak in O 5.8.4 for that case. (Now corrected in Github master)

Yes, I have the parameter AIS/ARPA on PPI ON. I start a test with yesterday's configuration and this parameter OFF immediatly !

The radar has been running for 2,5 hours without any crash right now. So I think this was the issue. Sorry if the AIS/ARPA on PPI issue was already reported somewhere and I missed it.

Huge thanks to you two for helping me solving this issue, and have a great day !

[Hakansv]

@ERousseletChorus You're welcome and sorry for the O bug creating this. You've to await a new OCPN release. (Technically it's possible to use a unstable alpha version from the ongoing developing work. But maybe not suitable for navigation?)