How it Works
You can configure the heartbeat capability at the interface-level, UPF profile group-level, or both. The interface-level configuration is mandatory. If the interface-level configuration is unavailable, then the heartbeat parameters get configured with the default values. The profile-level configuration overrides the interface-level configuration.
The heartbeat feature is also extended to achieve high-availability for the Node Manager.
Interface and profile-level heartbeat
The SMF-UPF interaction to detect the UPF path failure using the heartbeat messages involves the following steps:
-
The SMF sends a heartbeat request message to the discovered UPF instances or profile groups based on the configured schedule.
-
If the UPF instance or profile is alive, it sends a heartbeat response to the SMF indicating that it is operational. In case the UPF does not send a heartbeat response, then SMF retransmits the heartbeat request based on the configured interval and the number of permitted attempts.
-
After the configured count of heartbeat message reattempts is exhausted and the SMF does not receive a response from UPF, then SMF starts 'Network requested PDU Session Release' procedure for the subscribers that are associated with that UPF.
Heartbeat and high-availability in Node Manager
Each UPF instance is associated with a primary and secondary Node Manager. The secondary Node Manager acts as a standby system on which the primary manager fails over. The primary Node Manager is responsible for the IP allocation and managing the association-specific messages such as association create, update, or delete request. To achieve uninterrupted access to the UPF and ensure a high-availability environment, the following interactions occur:
-
When the Node Manager goes down or reinstated, it updates its status to all the mapped UPFs.
-
In case the primary Node Manager is down, the secondary manager is notified. The secondary manager sends a heartbeat request to the UPF node to determine if the node is alive.
-
When the secondary Node Manager is informed that the primary Node Manager is available, the secondary manager suspends the heartbeat timers and retransmission for the UPF node that is managed by the primary Node Manager.