The Zigbee2mqtt integration has an “Enable/Disable” button that stops the containers, but does not clean up either database variables or files on disk. There is no way to start from scratch without manual intervention. The idea is to add a “Reset” button on the setup page, with confirmation (like on matterbridge/node-red), that performs a full reset: disconnection, deletion of variables, deletion of files and containers.
For what it’s worth, it looks perfect to me. It’s very clear!
Regarding Gladys + backup, if we restore a version with an old dongle, is there any risk? Or conversely, is it an additional safety net because we could restore in case of a mistake?
That’s something to test, I don’t know how it behaves. In the cases we’ve seen on the forum, it was a different situation: dongle change with files already on the disk.
In the case of a Gladys Plus restore, we don’t restore the Zigbee2mqtt files
For me, we back up the entire contents of the zigbee2mqtt folder (so configuration.yaml, database.db and coordinator_backup.json).
But the restore checks whether a new configuration is present. If you restore a backup made with an old dongle (while a new dongle is plugged in and z2m is not initialized), you will also need to use that button.
I took the opportunity to strengthen the integration’s robustness, because there were instabilities that became even more apparent with this new feature.
Notably the stopped-container bug: until now, in Gladys, if the Zigbee2mqtt container was already stopped, a 304 error was issued and we never proceeded to remove the container. It’s fixed for all integrations
I tested many cases locally on a mini-PC, and it works really well!
Tested on an instance with a dongle but no paired devices. It works well, the states are correct afterwards and the folders/variables/containers are properly deleted