Twelve year old bug (XBee)

[Vadim Tkachenko]

Hello lurkers,

From what I understand, not many are going to be impacted, but I
thought I'd publish this anyway, because I'm speechless about how a
bug this bad could exist for such a long time without ever manifesting
itself until a configuration change triggered it... and even though
it's intermittent, the whole system randomly going nuts every few days
is not something anyone could not notice, nor tolerate.

The bug: https://github.com/home-climate-control/dz/issues/241

The circumstances: As luck would have it, it appears that the mix of
XBee and WiFi devices, and the physical layout of the XBee network,
made it so stable, there was never an XBee timeout since 2010.
However, a short while ago, nudged by the power outage and MQTT
subsystem reconfiguration, I removed two XBee sensors that were no
longer used, and this apparently made the XBee mesh unstable (even
though they were leaf nodes) - but only when some other, not yet
identified factors, are in effect.

The good: I don't believe this bug exists in the reactive code base
(still, will verify and write test cases).
The bad: this needs to be fixed pronto, it's pretty serious.
The ugly: it manifested itself in the code base that was slated for
deprecation, so it's sunk time :) Oh well.

--vt