conflict between watchdog and update deployment

Example of a problem encountered when deploying an update

• Poor network connection resulting in a significant delay when pulling the Docker image, leading to a Failure

• Cause (upon checking the workstation) -> the relay switch was activated by the Pi Zero -> probably because the service was down for too long

What approach should be considered to address this issue? What robust strategy should be adopted (we want to delay the Pi Zero but ensure it reboots at some point, as it acts as a backup for the Pi!) ?

Either the Pi Zero should check the Pi’s status by means other than querying Docker (which is currently already a limitation as it only monitors Docker)
or, during the deployment process, should we set the failure counter to a higher level to allow sufficient time for the update to be deployed?

Happy to discuss it !

Start by tracing the watchdog's Docker-based Pi status check, the update deployment path, and the failure counter behavior described in the issue. Compare the timing of a delayed Docker image pull with the Pi Zero relay activation, then define a strategy that delays the backup response while still guaranteeing a reboot; the issue does not name files or tests to run.