We are currently investigating an issue in an OSSV2 environment using EOSGS where authentication recovery fails after EOS token expiration.
After approximately one hour of gameplay, the EOS session token expires. In non-EAS configurations, this is handled correctly via OnPendingAuthExpiration, allowing the system to re-authenticate and recover without issue.
However, in EAS-enabled environments, when the token expires, FAuthEOSGS::HandleConnectAuthNotifyExpirationImplOp is triggered. After this point, the recovery flow fails during:
FAuthEOSGS::InitializeConnectLoginRecoveryTimer → FAuthEOSGS::ConnectLoginRecoveryImplOp
because AccountInfoEOS->LoginStatus is not in ELoginStatus::LoggedInReducedFunctionality, causing the recovery process to be treated as invalid.
In parallel, the network layer continues processing:
UIpNetDriver::TickDispatch → FPacketIterator → FSocketEOS::RecvFrom → EOS_P2P_ReceivePacket
At this stage, EOS_P2P_ReceivePacket returns EOS_EResult::EOS_InvalidUser, which leads to SE_EINVAL being set in the socket subsystem.
As a result, authentication recovery does not proceed correctly, and the P2P receive path may block or stall the game thread.
We suspect there is a mismatch between EOSGS authentication state transitions and EOS P2P user validity handling during token expiration and recovery.
We have considered potential workarounds such as:
Adjusting login state earlier during auth expiration handling
Relaxing login status validation in recovery flow
Treating EOS_InvalidUser as non-fatal in FSocketEOS::RecvFrom
However, these may introduce unintended side effects in networking and authentication layers.
We would greatly appreciate guidance on the correct or intended approach for handling EOSGS token expiration recovery in this scenario, or whether there is a recommended fix or configuration we may be missing.
Thank you very much for your support.