I appreciate all the time you guys have spent troubleshooting this issue. If this is the result of the fix things are looking much better:
Since we switched over to pump 2 did the messages with no timestamp start occuring Saturday 2019-09-13 but before that the pump 2 messages all had timestamps?
When I was at the boat 2019-09-15 the bilge was dry enough that the water wasn’t activating the float switch until I dumped some water in the bilge around 3pm after having listened for the pump for an hour or so.
No need to thank us, it’s a very good edge case for making sure we can get out as much reliable information as possible when communications are intermittent.
On the timestamps question, yes, it does look like things were timestamped much more regularly form Sept 2-12, although still a couple of missing timestamps.
We have just pushed into production some more tweaks to the plotting code and a lot of changes to the alerting code. Basically, the code will now only pay attention to pairs of signals (a pump on signal followed by a pump off signal) and will ignore events outside of that pattern. This is still not perfect, and we would like to add some higher level heuristics, but it should at least help avoid the “Your pump has run for 344755 seconds” alerts.
