fix daemon reconect error

[kerthcet]

What this PR does / why we need it

Which issue(s) this PR fixes

Fixes #39

Special notes for your reviewer

Does this PR introduce a user-facing change?

[Copilot]

Pull request overview

This PR adds an application-level heartbeat acknowledgement to improve daemon/controller resiliency when daemons are reaped from the server registry while the underlying WebSocket/TCP connection remains alive, and to help the daemon detect a controller that is connected but no longer processing messages.

Changes:

• Extend the shared protocol with Message::HeartbeatAck { success, reason }.
• Update the server to record heartbeats via a unit-testable helper and to ACK every heartbeat (success/failure).
• Update the daemon to track heartbeat ACKs, re-register in-place on NACK, and reconnect if ACKs stop arriving; add end-to-end tests.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 2 comments.

File	Description
server/src/server.rs	Adds handle_heartbeat helper, sends HeartbeatAck on heartbeats, updates reaper commentary, and adds unit tests for reap/re-register scenarios.
sandd/src/main.rs	Tracks last heartbeat ACK, reconnects on ACK silence, re-registers on rejected heartbeats, and adds WebSocket-level tests for recovery behavior.
protocol/src/lib.rs	Introduces the HeartbeatAck message variant and documents intended behavior.

Suppressed comments (2)

server/src/server.rs:371

• The comment says a reaped daemon’s “next heartbeat re-registers it”, but the heartbeat itself is rejected and the daemon re-registers by sending a new Register. Clarifying this avoids suggesting the server reconstructs state automatically.

        // dead. Reaping a daemon whose socket is still open no longer orphans it: its
        // next heartbeat re-registers it (see handle_heartbeat), so a false reap costs
        // one heartbeat interval of invisibility rather than lasting until the socket
        // breaks. Detection is ~30-35s vs the old ~90-120s; clean disconnects are still

sandd/src/main.rs:491

• The ack-timeout reconnect logic assumes the controller will send HeartbeatAck. If a daemon is upgraded before the controller, acks will never arrive and the daemon will disconnect/reconnect every ~ack_timeout even though the connection is otherwise healthy. Consider only enforcing the ack-timeout after at least one HeartbeatAck has been observed on the connection (or add a feature/compat negotiation) to preserve backwards compatibility.

            _ = ack_check.tick() => {
                let silent_for = last_ack.elapsed();
                if silent_for >= ack_timeout {
                    warn!(
                        "No heartbeat ack for {}s (controller connected but unresponsive); reconnecting",

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[kerthcet]

/retest

[kerthcet]

/lgtm
/approve
/kind bug

[InftyAI-Agent]

Approved: PR has both lgtm and approved labels