High Load and "Node has not completed discovery" after Controller Re-initialization

[FranzSchi]

Description

Hello! I recently migrated my coordinator from an EM21 to an EM24. The initial migration went smoothly, but I encountered significant performance issues after deciding to "re-initialize" the controller to ensure a clean setup.

The Problem

After triggering the re-initialization, I observed the following:

• System Load: Average jumped to 2.X, compared to the usual 0.2–0.3.

• Discovery Issues: A large number of devices (primarily battery-powered) remained in the status: "node has not completed discovery".

• Unresponsiveness: The entire system became unresponsive for approximately 2 hours.

• eventually resolved this by restoring the userdata directory from a backup, which returned everything to a stable state.

Questions & Feedback

I have a few questions regarding the "Node has not completed discovery" state:

Technical Meaning: What exactly does this state imply for battery-powered devices during a re-initialization?

Next Steps: In this situation, is the best practice to manually trigger a "scan," or is it better to wait for the devices to wake up naturally?

Suggested Improvement

To improve the user experience, would it be possible to update the error message?

If the system knows a device is waiting for a wake-up interval, perhaps the message could provide a hint on how to resolve it (e.g., "Node has not completed discovery. Please wake the device manually or wait for the next check-in."). This would help users understand if the system is hanging or simply waiting for battery-operated nodes to check in.

Thank you for your hard work on this project!

[cdjackson]

Technical Meaning: What exactly does this state imply for battery-powered devices during a re-initialization?

It means that the initialisation wasn't complete and generally the device won't work. So the various bits of configuration that the binding sends to the device so that it works properly, was not completed.

Next Steps: In this situation, is the best practice to manually trigger a "scan," or is it better to wait for the devices to wake up naturally?

If this is a battery device, then it must be woken up to allow config to complete. You will have to go to the Thing configuration, select Advanced, and tick the Reinitialise Node box. Click save, and wake up the device.

To improve the user experience, would it be possible to update the error message?

To what exactly? I mean we can't write long error messages, and I'm not clear what else makes more sense. I'm open to suggestions.

[FranzSchi]

I recommend adding a new sub-section under [When things don't appear to be working]
(https://www.openhab.org/addons/bindings/zigbee/%23when-things-dont-appear-to-be-working) or right after the Thing Configuration section.

Suggested Text to Add:

Known issues after reinitialization of the Controller

Battery powered devices spend most of their time in a "sleep" state to conserve energy. This can occasionally lead to an incomplete initialization where the device appears in openHAB but does not report data or respond to commands.

Symptoms:

• The Thing status stays in OFFLINE - "Node has not completed discovery".
• How to fix: To complete the configuration, the device must be awake while the binding receives its settings.
• Enable Show Advanced in the Thing configuration.
• Select the initialize Device checkbox and click Save.
• Physically wake the device (refer to the manufacturer's manual, typically a short press of the pairing/action button).

The binding will now push the "binding and reporting" configuration. You may need to wake the device several times if it is a complex sensor with many clusters.

One other symptom can be that all devices appear as "gone" this can be resolved by doing an initial "scan" for new devices.