Weird shutdown

[HansR]

I normally shutdown CumulusMX daily at 9:00 hrs for a backup and about a minute later it restarts. That's been going on like that since I restarted June last year.
This morning I had some kind of failure (see last logile, no debugging).
I make the following remarks:

1. My experimentation with MQTT left cumulus trying to connect continuously with a non-existing MQTT server. It could have stopped that after failing three times (and not modifying the connection data) It floods the logfile.
2. There were repetitive stackdump failures at shutdown (see end of last logfile).
3. My Cumulus.ini file was completely reinitialised when starting up.

Had to restore the ini from my backup, system running again but it was kind of a shock. I deleted MQTT connection data of course. Did not yet activate debug logging, but in the light of recent changes in the shutdown code I thought it useful to mention. I'll keep my eyes open.

[mcrossley]

MQTT - yep early days, reconnection needs some love.

Errors on close: Seem benign, the data reading thread continued after the port was closed. We often see things like that. The shutdown code is not as complete as it could be.

Cumulus.ini - odd as you can see in the log it says it completed writing to Cumulus.ini, then wrote today.ini. I have a question to myself about if those files need to be written at all on shutdown.

[HansR]

Yes, the shutdown itself and the MQTT are not high priority. A message to the user to switch it off when continued failing might be useful.

Cumulus.ini - odd as you can see in the log it says it completed writing to Cumulus.ini, then wrote today.ini. I have a question to myself about if those files need to be written at all on shutdown.

I think files are only written if at least one parameter has changed value.