part 4 附带：vm.boot/vm.create 500 记响应体

_boot_vm 只留 status 导致间歇 500 根因无法定位（两轮已提，
未改）。_api_call 的两个调用点现在都把响应体（截断到 500
字符单行）放进 ChvFailure 消息：vm_create_failed 与
vm_boot_failed 均含 status 与 body。body 是 VMM 的报错文本，
不含秘密；超长截断并标明。
新测试：test_wake_vm_boot_failed_records_body、
test_wake_vm_create_failed_records_body、
test_wake_vm_boot_failed_body_truncated。

cloud/chv/executor.py

@@ -1227,6 +1227,19 @@ class CloudHypervisorExecutor(Executor):
         return normalize_api_result(
             self.api_client(socket_path, method, path, payload))
 
+    @staticmethod
+    def _truncate_api_body(body, limit=500):
+        """把 API 响应体压成单行短串（放进失败消息里定位用）。
+        vm.create/vm.boot 的错误体是 VMM 的报错文本，不含秘密；
+        超长即截断并标明，避免日志被巨体淹没。"""
+        if body is None:
+            return ""
+        text = body if isinstance(body, str) else repr(body)
+        text = " ".join(text.split())
+        if len(text) > limit:
+            return text[:limit] + "…（已截断）"
+        return text
+
     def _call_runner(self, cmd, timeout=None):
         """runner 调用（共享 helper 在 cloud/sandbox.py）：
         timeout 给定时优先传 timeout，假 runner 只接受 cmd 时
@@ -1869,13 +1882,14 @@ class CloudHypervisorExecutor(Executor):
         self._launch_api_only(task)
         self.wait_api_ready(task)
         self._check_api_socket_permissions(socket_path)
-        status, _ = self._api_call(
+        status, body = self._api_call(
             socket_path, "PUT", "/vm.create",
             self.build_vm_config(task_disk, seed_disk=seed_disk))
         if status not in (200, 204):
             raise ChvFailure(
                 "vm_create_failed",
-                "vm.create 失败：status=%s" % status)
+                "vm.create 失败：status=%s body=%s"
+                % (status, self._truncate_api_body(body)))
         self._emit("vm_create", task,
                    "vm=%s disk=%s" % (name, task_disk))
         if net_entries:
@@ -1886,11 +1900,12 @@ class CloudHypervisorExecutor(Executor):
                     "net_registry_write_failed",
                     "vm=%s net 登记写失败，拒绝启动（缺登记会"
                     "漏判 tap/mac/ip 冲突）" % name)
-        status, _ = self._api_call(socket_path, "PUT", "/vm.boot")
+        status, body = self._api_call(socket_path, "PUT", "/vm.boot")
         if status not in (200, 204):
             raise ChvFailure(
                 "vm_boot_failed",
-                "vm.boot 失败：status=%s" % status)
+                "vm.boot 失败：status=%s body=%s"
+                % (status, self._truncate_api_body(body)))
         self._emit("vm_boot", task, "vm=%s" % name)
         self.wait_guest_ready(task)
         # V-9 移植：密钥已投递时等 guest-env 对 SSH 用户可读再报

cloud/tests/test_chv_closeup.py

@@ -213,6 +213,42 @@ class WakeFailureTest(CloseupCase):
         with self.assertRaises(ChvFailure) as ctx:
             ex.wake_vm(self._task())
         self.assertEqual(ctx.exception.kind, "vm_boot_failed")
+
+    def test_wake_vm_boot_failed_records_body(self):
+        # part 4：vm.boot 500 只留 status 即无法定位——body 要进
+        # 失败消息（截断后，见 _truncate_api_body）。
+        def api(socket_path, method, path, payload=None):
+            if method == "PUT" and path == "/vm.boot":
+                return (500, {"error": "boom-detail-500"})
+            return (200, {})
+        ex = self.make_exec(api_client=api)
+        with self.assertRaises(ChvFailure) as ctx:
+            ex.wake_vm(self._task())
+        self.assertEqual(ctx.exception.kind, "vm_boot_failed")
+        self.assertIn("boom-detail-500", str(ctx.exception))
+        self.assertIn("status=500", str(ctx.exception))
+
+    def test_wake_vm_create_failed_records_body(self):
+        def api(socket_path, method, path, payload=None):
+            if method == "PUT" and path == "/vm.create":
+                return (500, "No bootitem named default")
+            return (200, {})
+        ex = self.make_exec(api_client=api)
+        with self.assertRaises(ChvFailure) as ctx:
+            ex.wake_vm(self._task())
+        self.assertEqual(ctx.exception.kind, "vm_create_failed")
+        self.assertIn("No bootitem named default", str(ctx.exception))
+
+    def test_wake_vm_boot_failed_body_truncated(self):
+        def api(socket_path, method, path, payload=None):
+            if method == "PUT" and path == "/vm.boot":
+                return (500, {"error": "x" * 2000})
+            return (200, {})
+        ex = self.make_exec(api_client=api)
+        with self.assertRaises(ChvFailure) as ctx:
+            ex.wake_vm(self._task())
+        self.assertIn("已截断", str(ctx.exception))
+        self.assertLess(len(str(ctx.exception)), 1200)
         self.assertIn("vm_failed", self._audit_events())
         self.assertEqual(ex._vms, {})