In EmberZNet stack v6.1.0.
The Coordinator (ZC) with the Concentrator Plugin will broadcast an MTORR (many-to-one route request) periodically. This should update all the one hop routes back to the ZC. Then the ZC broadcasts an Address Discovery to find the missing node. This missing node will send a unicast response to the ZC using the newly created many-to-one-route.
When the source route or many-to-one route fails due to some route segment problem, a Route Error notification is forwarded back to the concentrator by the neighbors of the device who had the problem relaying. This Route Error is received by the concentrator, and the emberIncomingRouteErrorHandler is triggered by the stack to make the application take notice.
Our application framework code's Concentrator Utilities Plugin handles this stack-level callback by incrementing a count of the route errors seen since the last MTORR. If the number of route errors exceeds the EMBER_AF_PLUGIN_CONCENTRATOR_ROUTE_ERROR_THRESHOLD (set by the user in the UI configuration dialog for Concentrator Utilities Plugin), a new MTORR is initiated to reconstruct all of the many-to-one routes into the concentrator and have all routers send a new Route Record (to inform concentrator of the latest source route data) on their NEXT unicast to the concentrator. So the routes will repair themselves automatically (via the plugin). More information can be found in: documentation/UG1033-AppDevFundamentals-DesignChoices.pdf section 2.4 Route Establishment.
EMberZNet SDK 6.7.3 update:
The "Concentrator Support" plugin us found under the Utility category of plugins and has the following options:
Useful related KBAs:
2. Refer the routing section of this KBA. https://www.silabs.com/community/wireless/zigbee-and-thread/knowledge-base.entry.html/2014/03/24/guidelines_for_large-pgj4