Knowing what happened.
WHAT TO DISTINGUISH
Telemetry: the measurements the system exists to collect
Diagnostics: information about the device's own health
WHY SEPARATE THEM
They have different destinations, retention and urgency.
WHAT DIAGNOSTIC TELEMETRY SHOULD INCLUDE
Restart count and reason Free memory Signal quality Battery state Error counts by type Firmware and configuration versions
WHY SIGNAL QUALITY SPECIFICALLY
It explains both power consumption and lost messages.
WHAT TO SEND IT WITH
Regular messages, appended, rather than separately.
WHY
Separate messages double the transmission cost.
WHAT LOCAL LOGGING SUITS
Detail retrievable on request or on site.
WHAT TO BE CAREFUL WITH
Log storage filling Logging consuming power Logs that are never read
WHAT TO IMPLEMENT
Adjustable verbosity, raised temporarily when investigating.
WHY TEMPORARILY
Detailed logging cannot run continuously on a constrained device.
WHAT TO NEVER LOG
Credentials Personal data, unnecessarily
WHAT TO ENSURE
That logs survive a restart.
WHY
The most valuable log entry is the one before the crash.
