Feature Description
The SMF supports the DPR and it handles the received DPR from server. The DPR is sent to a peer to inform its intentions to shut down the transport connection. When the peer node sends the DPR to another node, the node validates the DPR for all required AVPs. If the validation isn’t successful, the node sends the DPA with the error to the peer node.
If the validation is successful:
-
The SMF node waits till the drain time before it sends any DPA to the peer node. DrainTime is the time interval between receiving a DPR by diameter endpoint and sending a DPA response. No new requests are initiated by the diameter-ep to the peer during this interval. However any messages received from the peer are processed during this interval. Drain time is a configurable parameter. It’s recommended to configure the drain time value less than the timeout value for inbound messages. It’s advised that no new request to be sent to that peer during this time, but in case any requests comes from the peer, it must be responded.
-
Regardless of whether any messages have previously been forwarded to the client and are waiting for a response, transmit DPA successfully. The peer node sends the DPA with success.
If a server wants to disconnect itself from a peer SMF node, it sends DPR with disconnect cause AVP.
-
If disconnect cause is REBOOTING, defer connecting back based on Tc Timer. Tc timer is a configurable parameter.
-
The Reboot time is the time duration after which connection is reattempted to peer from which DPR is received with disconnect cause “REBOOTING”. This interval starts after sending the DPA response and connection teardown.
-
-
If the disconnect cause is BUSY or DO_NOT_WANT_TO_TALK_TO_YOU, defer connecting back for longer duration. Duration is a configurable parameter. As it’s recommended not to connect back in case of BUSY or DO_NOT_WANT_TO_TALK_TO_YOU. So, you can configure negative or zero as value to never connect back.
-
DoNotTalkTime is the time duration after which connection is reattempted to peer from which DPR is received with disconnect cause “DO_NOT_WANT_TO_TALK_TO_YOU”. This interval starts after sending the DPA response and connection teardown.
-
BusyTime is the time duration after which the connection is reattempted to peer from which DPR is received with disconnect cause “BUSY”. This interval starts after sending the DPA response and connection teardown.
-
-
The time duration for reconnecting back is applicable across all pod restarts or pod failover but doesn't apply to site failover.