No, my scenario has nothing to do with inrush current or any load application; the reason I know this is that it happens even when the load is disconnected (it is an attic fan with dedicated switch that can be independently turned off). The fan is only used seasonally, and I developed the automation in the off-season with the load (not the switch) unpowered. That's when I found it sometimes stopped reporting physical paddle presses (though it otherwise worked normally; not a mesh issue). An airgap cycle restores normal operation for weeks. I only rely on the physical switch state reports to provide an audible warning should someone attempt to operate the fan locally without any exterior windows/doors open; as it otherwise works well I have just decided to live with it.
Not too unusual, either, as the same phenomenon has been reported by other users (in ST community). It may be related to the firmware level of the switch; it is one of those early ones that is unable to report double-tap events when used with a compatible driver. Again, what I observed was related to reporting physical paddle presses, not a failure to report a state change due to a digital event.
I just took the dog for a walk and had an idea of what happened. I just got back and looked at the on off times in the log I posted above and this is indeed true.
The pump was running for not 1hr, but 24+ hours.
My system performance had been degrading for several days until the night of the 6th when my "goodnight" command failed totally. It wouldn't shut anything off. I got up, rebooted the hubitat hub and rebooted my home assistant box, shut the light off by hand and went back to bed. This is when the pump got stuck on, and I didn't notice it was running continuously until the going to bed at 1:04 the next evening, when I heard it running.
The times in the logs above confirm this to be the case. The "on" command comes at 12:09 02-06 and the off at 01:04 02-07.
Damnit.
My new RM failsafe should prevent this in that it doesn't just try to shut off once when the state changes, it attempt to shut the pump off every 4 minutes that it's on.
The other thing I wonder about is why my HA rule didn't shut off, but then I just tested that too. This is an external tankless heater, and yesterday/last night was the coldest day of the year so far. It seems that I'm not pumping enough heated water through the return loop to raise the temperature of the water at my sensor above my threshold, when mixed with cold water at the input to the heater at the temperature that the input water was last night.
I guess as a silver lining, I kept the pipes from freezing, though I have some heat tape for that too so it's not really a concern. My pump is a 3 speed pump and I've been running it on the middle speed. That seemed enough, but I think the high speed setting is probably better given this experience.
