diff --git a/lazyllm/tools/infer_service/client.py b/lazyllm/tools/infer_service/client.py
index a7f97e601..47e9a870b 100644
--- a/lazyllm/tools/infer_service/client.py
+++ b/lazyllm/tools/infer_service/client.py
@@ -120,9 +120,6 @@ def get_infra_handle(self, token, job_id):
         response.raise_for_status()
         response = response.json()
         base_model, url, deploy_method = response['base_model'], response['url'], response['deploy_method']
-        lazyllm.LOG.warning(base_model)
-        lazyllm.LOG.warning(url)
-        lazyllm.LOG.warning(deploy_method)
         if self.uniform_status(response['status']) != 'Ready':
             raise RuntimeError(f'Job {job_id} is not running now')
         if not (deployer := getattr(lazyllm.deploy, deploy_method, None)):
diff --git a/lazyllm/tools/infer_service/serve.py b/lazyllm/tools/infer_service/serve.py
index 6082c2fd5..e8c262793 100644
--- a/lazyllm/tools/infer_service/serve.py
+++ b/lazyllm/tools/infer_service/serve.py
@@ -40,6 +40,7 @@ def _update_status(self, token, job_id):
         # Ready to Infer
         if Status[status] == Status.Running and m._url:
             update['status'] = 'Ready'
+            update['url'] = m._url
 
         # Some tasks cannot obtain the storage path when they are just started
         if not log_path:
diff --git a/tests/advanced_tests/standard_test/test_engine.py b/tests/advanced_tests/standard_test/test_engine.py
index e433459c9..eba362d46 100644
--- a/tests/advanced_tests/standard_test/test_engine.py
+++ b/tests/advanced_tests/standard_test/test_engine.py
@@ -192,3 +192,26 @@ def test_engine_train_serve(self):
 
         res = engine.local_model_get_training_cost(token, job_id)
         assert res > 15
+
+    def test_engine_infer_server(self):
+        token = '123'
+        engine = LightEngine()
+        engine.launch_localllm_infer_service()
+        jobid, status = engine.deploy_model(token, 'internlm2-chat-7b')
+        engine.infer_client.wait_ready(token, jobid)
+        r = engine.get_infra_handle(token, jobid)
+        assert isinstance(r, lazyllm.TrainableModule) and r._impl._get_deploy_tasks.flag
+        assert '你好' in r('请重复下面一句话：你好')
+
+        nodes = [dict(id='0', kind='SharedLLM', name='m1', args=dict(
+            llm=jobid, local=False, token=token, stream=True, prompt=dict(
+                system='请根据输入帮我计算，不要反问和发挥', user='输入: {query} \n, 答案:')))]
+        gid = engine.start(nodes)
+        assert '2' in engine.run(gid, '1 + 1 = ?')
+
+        engine.stop(gid)
+        nodes = [dict(id='1', kind='OnlineLLM', name='m1', args=dict(
+            source='lazyllm', base_model=jobid, token=token, stream=True, prompt=dict(
+                system='请根据输入帮我计算，不要反问和发挥', user='输入: {query} \n, 答案:')))]
+        gid = engine.start(nodes)
+        assert '2' in engine.run(gid, '1 + 1 = ?')