bugfix batch

Author: jaberjaber23

Cargo.lock

@@ -18,7 +18,7 @@ members = [
 ]
 
 [workspace.package]
-version = "0.3.16"
+version = "0.3.17"
 edition = "2021"
 license = "Apache-2.0 OR MIT"
 repository = "https://github.com/RightNow-AI/openfang"
@@ -122,6 +122,9 @@ argon2 = "0.5"
 # Lightweight regex
 regex-lite = "0.1"
 
+# Socket options (SO_REUSEADDR)
+socket2 = "0.5"
+
 # Zip archive extraction
 zip = { version = "2", default-features = false, features = ["deflate"] }
 

Cargo.toml

@@ -33,6 +33,7 @@ governor = { workspace = true }
 tokio-stream = { workspace = true }
 subtle = { workspace = true }
 base64 = { workspace = true }
+socket2 = { workspace = true }
 
 [dev-dependencies]
 tokio-test = { workspace = true }

crates/openfang-api/Cargo.toml

@@ -740,7 +740,8 @@ pub async fn run_daemon(
         if info_path.exists() {
             if let Ok(existing) = std::fs::read_to_string(info_path) {
                 if let Ok(info) = serde_json::from_str::<DaemonInfo>(&existing) {
-                    if is_process_alive(info.pid) {
+                    // PID alive AND the health endpoint responds → truly running
+                    if is_process_alive(info.pid) && is_daemon_responding(&info.listen_addr) {
                         return Err(format!(
                             "Another daemon (PID {}) is already running at {}",
                             info.pid, info.listen_addr
@@ -749,7 +750,8 @@ pub async fn run_daemon(
                     }
                 }
             }
-            // Stale PID file, remove it
+            // Stale PID file (process dead or different process reused PID), remove it
+            info!("Removing stale daemon info file");
             let _ = std::fs::remove_file(info_path);
         }
 
@@ -771,7 +773,22 @@ pub async fn run_daemon(
     info!("WebChat UI available at http://{addr}/",);
     info!("WebSocket endpoint: ws://{addr}/api/agents/{{id}}/ws",);
 
-    let listener = tokio::net::TcpListener::bind(addr).await?;
+    // Use SO_REUSEADDR to allow binding immediately after reboot (avoids TIME_WAIT).
+    let socket = socket2::Socket::new(
+        if addr.is_ipv4() {
+            socket2::Domain::IPV4
+        } else {
+            socket2::Domain::IPV6
+        },
+        socket2::Type::STREAM,
+        None,
+    )?;
+    socket.set_reuse_address(true)?;
+    socket.set_nonblocking(true)?;
+    socket.bind(&addr.into())?;
+    socket.listen(1024)?;
+    let listener =
+        tokio::net::TcpListener::from_std(std::net::TcpListener::from(socket))?;
 
     // Run server with graceful shutdown.
     // SECURITY: `into_make_service_with_connect_info` injects the peer
@@ -891,3 +908,26 @@ fn is_process_alive(pid: u32) -> bool {
         false
     }
 }
+
+/// Check if an OpenFang daemon is actually responding at the given address.
+/// This avoids false positives where a different process reused the same PID
+/// after a system reboot.
+fn is_daemon_responding(addr: &str) -> bool {
+    // Quick TCP connect check — don't make a full HTTP request to avoid delays
+    let addr_only = addr
+        .strip_prefix("http://")
+        .or_else(|| addr.strip_prefix("https://"))
+        .unwrap_or(addr);
+    if let Ok(sock_addr) = addr_only.parse::<std::net::SocketAddr>() {
+        std::net::TcpStream::connect_timeout(
+            &sock_addr,
+            std::time::Duration::from_millis(500),
+        )
+        .is_ok()
+    } else {
+        // Fallback: try connecting to hostname
+        std::net::TcpStream::connect(addr_only)
+            .map(|_| true)
+            .unwrap_or(false)
+    }
+}

crates/openfang-api/src/server.rs

@@ -3080,7 +3080,7 @@ mod tests {
     #[test]
     fn test_catalog_has_providers() {
         let catalog = ModelCatalog::new();
-        assert_eq!(catalog.list_providers().len(), 31);
+        assert_eq!(catalog.list_providers().len(), 33);
     }
 
     #[test]