The GGSN will receive the SGSN restart counters in GTPv1 echo response from the SGSN. When a GGSN detects that a peer SGSN has restarted it shall delete all PDP context(s), MBMS UE context(s), MBMS Bearer context(s) associated with the peer node that failed as well as freeing any internal GGSN resources associated with those PDP context(s), MBMS UE context(s) and MBMS Bearer context(s). The GGSN may optionally perform other implementation specific actions such as messages to clear other external resources (e.g. PCC messages). If the GGSN needs to send a request for IP-CAN Session Modification procedure towards a PCRF which is known to have restarted since the IP-CAN session establishment, the GGSN may discard the request and may tear down all the PDP context(s) associated with the PDP address of the IP-CAN session, based on operator policy, by initiating a PDP Context Deactivation procedure towards the SGSN with the cause set to "Reactivation requested". This leads the UE to initiate PDP Context Activation procedure for the same APN. Emergency sessions should not be torn down.

When the GGSN receives a mobile terminated PDU for which no valid PDP context exists the GGSN discards the received PDU and may also return an appropriate Error message depending on the protocol used. No further actions are performed by the GGSN. Alternatively, if the GGSN has static PDP information about the PDP address, the GGSN may try to deliver the PDU by initiating the Network-Requested PDP Context Activation procedure (see TS 23.060).

When the GGSN receives a tunnel PDU for which no PDP context exists it discards the tunnel PDU and sends an Error indication message to the originating SGSN. The SGSN deactivates the PDP context and sends an Error indication to the MS. The MS may then re-activate the PDP context.