From 35d9245b2fec24ad2c168868bc951d0eba508e2d Mon Sep 17 00:00:00 2001
From: Binbin Zhang <binbin36520@gmail.com>
Date: Thu, 3 Sep 2026 00:37:48 +0800
Subject: [PATCH] fix: skip unhealthy workers in proxy GET endpoints

select_first_worker returned the first registered worker without checking
its health state, so /v1/models and other proxy GETs returned 500 when the
first worker was temporarily down. Filter for a healthy worker instead;
fall back to the existing error when none are healthy.

Fixes #135

Signed-off-by: Binbin Zhang <binbin36520@gmail.com>
---
 src/routers/http/router.rs | 33 +++++++++++++++++++++++++++++----
 1 file changed, 29 insertions(+), 4 deletions(-)

diff --git a/src/routers/http/router.rs b/src/routers/http/router.rs
index f68e4b08..d8c72abe 100644
--- a/src/routers/http/router.rs
+++ b/src/routers/http/router.rs
@@ -338,11 +338,12 @@ impl Router {
     }
 
     fn select_first_worker(&self) -> Result<String, String> {
+        // Prefer a healthy worker so proxy GETs (e.g. /v1/models) don't 500
+        // when the first registered worker is temporarily down.
         let workers = self.worker_registry.get_all();
-        if workers.is_empty() {
-            Err("No workers are available".to_string())
-        } else {
-            Ok(workers[0].url().to_string())
+        match workers.into_iter().find(|w| w.is_healthy()) {
+            Some(worker) => Ok(worker.url().to_string()),
+            None => Err("No healthy workers are available".to_string()),
         }
     }
 
@@ -1828,6 +1829,30 @@ mod tests {
         assert!(url == "http://worker1:8080" || url == "http://worker2:8080");
     }
 
+    #[test]
+    fn test_select_first_worker_skips_unhealthy() {
+        let router = create_test_regular_router();
+        // Mark worker1 unhealthy; select_first_worker must not return it.
+        for w in router.worker_registry.get_all() {
+            if w.url() == "http://worker1:8080" {
+                w.set_healthy(false);
+            }
+        }
+        let result = router.select_first_worker();
+        assert!(result.is_ok());
+        assert_eq!(result.unwrap(), "http://worker2:8080");
+    }
+
+    #[test]
+    fn test_select_first_worker_all_unhealthy() {
+        let router = create_test_regular_router();
+        for w in router.worker_registry.get_all() {
+            w.set_healthy(false);
+        }
+        let result = router.select_first_worker();
+        assert!(result.is_err());
+    }
+
     #[tokio::test]
     async fn test_wait_for_healthy_workers_empty_list() {
         // Empty list will return error immediately