Cumulus MX has recently started crashing

[cmr77]

Hello,
I've been using Cumulus MX 3.16.1 build 3183 with my LaCrosse WS2300 for a few years now and it's been super reliable - not seen any issues at all.
In the past week it's started crashing every few hours.
Attached is a zip of the MXDiags folder. I cleaned the folder out - restarted the program and waited for it to crash so the folder isn't too busy.
Any help would be much appreciated.

Carl

[PaulMy]

Hi,
v3.16.1 is not that old (likely May 2022) but updating generally cannot hurt.
The MXdiags file shows many new records for low pressure

Code: Select all

2022-11-16 18:01:12.515 New all-time record: New time = 2022-11-16 04:32, new value = 978.400 "Low pressure" prev time = 2022-11-15 10:57, prev value = 978.900
2022-11-16 18:01:12.552 New monthly record: month = 11: New time = 2022-11-16 04:32, new value = 978.400 "Low pressure" prev time = 2022-11-15 10:57, prev value = 978.900

Possibly this is correct, and pressure was dropping fast? or a sign of faulty sensor?

and

Code: Select all

2022-11-16 20:01:01.113 New all-time record: New time = 2022-11-16 20:00, new value = 963.200 "Low pressure" prev time = 2022-11-16 19:55, prev value = 963.300
2022-11-16 20:01:01.124 New monthly record: month = 11: New time = 2022-11-16 20:00, new value = 963.200 "Low pressure" prev time = 2022-11-16 19:55, prev value = 963.300
2022-11-16 20:01:36.027 !!! Unhandled Exception !!!
System.IndexOutOfRangeException: Index was outside the bounds of the array.
   at CumulusMX.WS2300Station.Ws2300PressureTrendAndForecast(String& pressuretrend, String& forecast)
   at CumulusMX.WS2300Station.GetAndProcessData()
   at CumulusMX.WS2300Station.Start()
   at System.Threading.ThreadHelper.ThreadStart_Context(Object state)
   at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state, Boolean preserveSyncCtx)
   at System.Threading.ExecutionContext.Run(ExecutionContext executionContext, ContextCallback callback, Object state, Boolean preserveSyncCtx)
   at System.Threading.ExecutionContext.Run(ExecutionContext executionContext, ContextCallback callback, Object state)
   at System.Threading.ThreadHelper.ThreadStart()
2022-11-16 20:03:00.457 *** Data input appears to have stopped

and the *** Data input appears to have stopped message continues until CumulusMX stopped.
Again, possibly sensor or station failure?

Also you should fix this:

Code: Select all

2022-11-17 00:00:00.153 Duplicate entry at line 1090 of dayfile.txt: 11/11/22. Please correct this by editing the file
2022-11-17 00:00:00.153 Duplicate entry at line 1091 of dayfile.txt: 12/11/22. Please correct this by editing the file
2022-11-17 00:00:00.153 Duplicate entry at line 1092 of dayfile.txt: 13/11/22. Please correct this by editing the file
2022-11-17 00:00:00.153 Duplicate entry at line 1093 of dayfile.txt: 14/11/22. Please correct this by editing the file
2022-11-17 00:00:00.154 Duplicate entry at line 1095 of dayfile.txt: 15/11/22. Please correct this by editing the file

Enjoy,
Paul

[mcrossley]

I'll take a detailed look at this tomorrow, but meantime could you generate a log file with data logging enabled as well please?

[cmr77]

I've removed the duplicate date entries from dayfile.txt and enabled data logging. I'll post the MXdiag folder again when it next crashes - It's been fine so far today.
Carl

[cmr77]

It ran most of yesterday then crashed later in the evening. Attached are the logs (with logging enabled) so hopefully that can shed some light onto what is going on.
The log file 20221118-192000.txt shows the Unhandled Exception error 2022-11-18 19:43:03.

Carl

[mcrossley]

Thanks, I've fixed this for the next release.

[cmr77]

Mark,
Ok glad you found something - I'll upgrade when the next release is available. Thanks for your help.

Carl

[mcrossley]

It fixes the crash, but the root cause is bad data being received. I see there are quite a few serial port timeouts occurring, is that a normal "station thing" or do you have a small problem with the serial connection? Given you say this only started happening recently, I'd guess the latter.

[cmr77]

I will look into the serial data timeouts. Nothing has changed recently but the WS2300 has a particularly nasty serial interface that references to one of the serial control lines rather than 0V - there is no 0V connection - very odd. I'll get the scope out and take a look at the quality of the received data.

Carl