Min/max/mean all coincide for extended time? 🐛

[danieltsoukup]

Seems like a weird pattern when for longer time periods on this sensor the min/max/mean all match exactly.

https://dashboard.tracket.info/locations/896325

@tcsullivan what do you think, could this be an issue with the sensor? Could this happen naturally?

[tcsullivan]

Seems like an issue with the sensor's timekeeping. Since it depends on the network, having an intermittent connection could make it unpredictable; my guess is that only one decibel reading got into each of these data points before the device decided that 5 minutes had elapsed. For example, in some cases I have seen the device's time report zero (seconds since epoch) when WiFi is lost or initially unavailable.

The best solution would be to keep time independently of the network, tracking date/time in a local variable and only synchronizing if connected to WiFi. This would ensure we have fairly accurate timestamps if the device if offline for a longer period.

For programmers: see here for how we can move a local timestamp forward after an initial sync. mktime will fix the date and time if we just increment the minutes.