web ui: retry binding on AddrInUse during restart races

operator hit 'Address already in use (os error 98)' on a harness
restart — the new harness raced the old socket's release. add a
bind_with_retry helper that backs off (250ms doubling, capped at
2s, 12 tries ≈ 22s total) on AddrInUse before giving up. applied
to both the per-agent web UI and the hive-c0re dashboard.

proper fix would be SO_REUSEADDR via socket2 but retry covers the
TIME_WAIT case fine and keeps the dep count down. Other bind errors
still fail immediately (port permission, fd exhaustion).

hive-ag3nt/src/web_ui.rs

@@ -83,9 +83,7 @@ pub async fn serve(
         .route("/api/compact", post(post_compact))
         .with_state(state);
     let addr = SocketAddr::from(([0, 0, 0, 0], port));
-    let listener = tokio::net::TcpListener::bind(addr)
+    let listener = bind_with_retry(addr, "web UI").await?;
-        .await
-        .with_context(|| format!("bind web UI on port {port}"))?;
     tracing::info!(%port, "web UI listening");
     axum::serve(listener, app).await?;
     Ok(())
@@ -95,6 +93,33 @@ pub async fn serve(
 // Static assets + state snapshot
 // ---------------------------------------------------------------------------
 
+/// Bind a TCP listener, retrying on `AddrInUse` for up to ~20s.
+/// nspawn restarts can race the previous harness's socket release;
+/// without retry the new harness fails to bind and systemd just
+/// keeps restarting it. `SO_REUSEADDR` would be the proper fix but
+/// would require socket2; retry is good enough here.
+async fn bind_with_retry(addr: SocketAddr, label: &str) -> Result<tokio::net::TcpListener> {
+    let mut delay_ms = 250u64;
+    let mut attempts = 0u32;
+    loop {
+        match tokio::net::TcpListener::bind(addr).await {
+            Ok(l) => return Ok(l),
+            Err(e) if e.kind() == std::io::ErrorKind::AddrInUse && attempts < 12 => {
+                tracing::warn!(
+                    %addr, attempt = attempts + 1,
+                    "{label}: AddrInUse, retrying in {delay_ms}ms"
+                );
+                tokio::time::sleep(std::time::Duration::from_millis(delay_ms)).await;
+                attempts += 1;
+                delay_ms = (delay_ms * 2).min(2000);
+            }
+            Err(e) => {
+                return Err(e).with_context(|| format!("bind {label} on {addr}"));
+            }
+        }
+    }
+}
+
 async fn serve_index() -> impl IntoResponse {
     (
         [("content-type", "text/html; charset=utf-8")],

hive-c0re/src/dashboard.rs

@@ -57,9 +57,7 @@ pub async fn serve(port: u16, coord: Arc<Coordinator>) -> Result<()> {
         .route("/messages/stream", get(messages_stream))
         .with_state(AppState { coord });
     let addr = SocketAddr::from(([0, 0, 0, 0], port));
-    let listener = tokio::net::TcpListener::bind(addr)
+    let listener = bind_with_retry(addr).await?;
-        .await
-        .with_context(|| format!("bind dashboard on port {port}"))?;
     tracing::info!(%port, "dashboard listening");
     axum::serve(listener, app).await?;
     Ok(())
@@ -73,6 +71,30 @@ pub async fn serve(port: u16, coord: Arc<Coordinator>) -> Result<()> {
 // `/messages/stream` for broker traffic.
 // ---------------------------------------------------------------------------
 
+/// Retry-on-AddrInUse bind. Same shape as the per-agent variant —
+/// hive-c0re restarts also race the previous process's socket release.
+async fn bind_with_retry(addr: SocketAddr) -> Result<tokio::net::TcpListener> {
+    let mut delay_ms = 250u64;
+    let mut attempts = 0u32;
+    loop {
+        match tokio::net::TcpListener::bind(addr).await {
+            Ok(l) => return Ok(l),
+            Err(e) if e.kind() == std::io::ErrorKind::AddrInUse && attempts < 12 => {
+                tracing::warn!(
+                    %addr, attempt = attempts + 1,
+                    "dashboard: AddrInUse, retrying in {delay_ms}ms"
+                );
+                tokio::time::sleep(std::time::Duration::from_millis(delay_ms)).await;
+                attempts += 1;
+                delay_ms = (delay_ms * 2).min(2000);
+            }
+            Err(e) => {
+                return Err(e).with_context(|| format!("bind dashboard on {addr}"));
+            }
+        }
+    }
+}
+
 async fn serve_index() -> impl IntoResponse {
     Html(include_str!("../assets/index.html"))
 }