Nice problem. I’m glad the diagnostics and recovery worked. Here’s what it looks like to me.
The IoTaWatt took a software watchdog exception while trying to read the latest record from the current log to post that data to influxDB. It hung while trying to read the latest record on Monday, 20-May.
Your log had wrapped, which is normal. When the current log gets to a certain size, it wraps around and starts writing over old entries. I don’t know when you installed this system but the history log begins at May 24, 2018 and that would have been the start of the current log as well. When it wrapped on May 17 of this year, it started overwriting the 2018 entries. That was just a little over three days ago. Hard to dismiss that coincidence.
There was a stray current log entry at the wrap point. It was an entry from June 21, 2018. ALPHA updated from 02_03_07 to 02_03_08 on that date. This record was from 20:02:30. If you have the complete message log, can you post the activity surrounding that date? Auto-updates can take a few hours to propagate, so that may be when your system updated and restarted.
Anyway, after the wrap, IoTaWatt must find the wrap point after restart, and that algorithm can fail if there are bad records. This is the only bad record in the log. Point is that the bad record could have been there for eleven months but became significant when the wrap detection logic was activated.
If there is some underlying problem with wrap, I think I would have seen it in my own test systems over the past six months. They wrapped every week for months without issue. That said, more and more systems will be wrapping as time passes, and whether it is old damage or something new, I’m glad to have the robust recovery.
Except for the 1.5 hours it took to create the diagnostic file, you should have recovered fully. The history log will support all local inquiries down to 1 minute resolution, and you already have 4 days of new current log at 5 second intervals.
Let me know if you have that June 21 message log.