Offline-to-Online Synchronization

In this lesson, you will learn how to design a "replay buffer" mechanism that ensures clients recover missed messages seamlessly after a network blackout.

Bridging the Gap: The Replay Buffer

When a client loses its connection, the SignalR Hub loses track of its real-time state. To achieve offline-to-online synchronization, we cannot rely on the volatile connection alone. Instead, we must implement a sequence-tracked replay buffer.

The core idea is to treat messages as an ordered stream. Each message is assigned a monotonically increasing SequenceId. When a client reconnects, it reports the LastReceivedSequenceId it successfully processed. The server then compares this ID against its historical message store and pushes only the missing delta.

Implementation Strategy

Instead of sending messages into a void, maintain a short-lived circular buffer or a dedicated table in your data store (e.g., Redis or SQL) that archives the last N messages.

public async Task Reconnect(long lastReceivedSequenceId)
{
    var context = Context.UserIdentifier;
    // Fetch only messages that occurred after the client's last known point
    var missedMessages = await _messageStore
        .GetMessagesAfterAsync(lastReceivedSequenceId);

    foreach (var msg in missedMessages)
    {
        await Clients.Caller.SendAsync("ReceiveMessage", msg);
    }
}

Critical Considerations

1. Idempotency: Because the client might have received a message but crashed before acknowledging it, your client-side logic must be idempotent. If a client receives a message it already has, it should simply overwrite or ignore it rather than duplicating the record in the UI.
2. Buffer Window: You must decide how long to keep history. If a client is offline for an hour, replaying every message might be impractical. Implement a "hard reset" threshold: if the lastReceivedSequenceId is too old (i.e., outside the buffer range), force the client to perform a full state snapshot fetch instead of a delta replay.
3. Ordering: Ensure the server pushes the missed messages in strict chronological order to prevent UI race conditions during the reconciliation phase.

Key Takeaway: Offline synchronization is effectively a state reconciliation process where the client provides its "last known index," allowing the server to replay only the missing delta from a persistent buffer.