v3130 Fine Offset Lost Connection Changes

[euphbass]

Hi there, I've installed the latest build (3130). Now, when it loses connection with the Fine Offset base station, it does keep trying to reconnect, but can't manage it. I suspect this is a base station error, so not the main query here. However, in my previous version of CumulusMX (3120 I think it was) when it lost the connection, the data recording just stopped, and it caught up with the stored data once it was forced to reconnect (hard reset of base station, restart Cumulus). The missing time period data was offset, but self consistent. Now, when the connection is lost, the data (e.g. as viewed on the recent graphs) goes a bit mad and is full of odd values and spikes, some of which are caught by spike handling and some of which aren't. You get things like the attached temp graph (connection lost ~10:30am). Do you know if there is a way to avoid this? When resetting the base station to regain the connection, it no longer seems to be getting the data stored on it for the disconnected time (or if it is, it's all over the place, which it wasn't before). Not quite sure what's going on here...

To be fair, if this is all the fault of the base station, I don't expect any solution other than to contact the manufacturer! Thanks .

[mcrossley]

If it has all connection to the station then there should be no new values logged - or at worst case the last values "flat line".

If it is logging "odd" values then that would indicate that MX is receiving "something" from the station, but it is invalid.

I'd really need to the MXdiags file, ideally with debug and data logging enabled.

[euphbass]

This is the relevant log file - not sure if it includes everything you mentioned. The connection was lost around 10:30am or soon after. Thanks! I also wanted to say thanks for adding email alerts for errors - it now emails me when it loses connection so I can reset the base station and the software fairly soon after the connection is lost, so minimising the amount of odd data when it's down.

[mcrossley]

As I expected it does not show a lot.
The station sent some bad data around 10:50 which was ignored, then it appeared to go offline at 10:54

After that nothing is logged except it though it can connect to the station, read operations are timing out.

At 22:05 it stopped responding completely.

[Ulric2]

I have had a similar problem using a W3083, CumulusMX 3.15.0 as a service on Linux. After a while (5 hours or so) data logging just stopped until the base station was reset. The PC I run it on is also used as a HTPC so it tends to be turned on early evening and is set to shut down automatically at 11pm. Having noted that the period before a hang occurs is always the same and that it freezes the base station rather than Cumulus, I wondered if it could be that two timers running with different periods somehow coincide at a point in time and the simultaneous instructions crash the base station.

I set up a cron job (running as root in /etc/crontab) which simply restarts the service every 4 hours and the problem has completely gone away. Guess you could do the same with task scheduler in windows.

Hope that helps someone!