feat: 新增工具插件、系统日志、workflow配置及前端优化

1. 新增工具插件(edit_file, python_executor, search_file, shell_executor, write_file)
2. 新增系统事件日志模块和API
3. 新增workflow配置文件和详情API
4. 前端增加SSE、错误边界、设置引导等组件
5. 优化认证加密、速率限制、配置加载等工具模块
6. 删除废弃的cluster和health API
7. 补充单元测试和集成测试

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-06-03 07:34:43 +00:00
parent f04fef916f
commit a53ffebe0e
57 changed files with 2804 additions and 271 deletions
+46 -2
View File
@@ -11,6 +11,7 @@
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import time
import ray
from functools import lru_cache
@@ -47,14 +48,57 @@ def clear_actor_cache():
_get_cached_actor_handle.cache_clear()
def ray_actor_hook(*actor_names: str):
def wait_for_actor(
actor_name: str, *, timeout: float = 10.0, interval: float = 0.5
):
"""阻塞等待某个 actor 就绪,返回其句柄。
用于"启动期 / ray task 入口刚拉起"这类场景——被依赖的 actor 可能还没注册。
在 ``timeout`` 内按 ``interval`` 轮询 ``ray.get_actor``;拿到就立即返回,
超时则抛带清晰上下文的 ``TimeoutError``(而不是裸 ``ValueError``)。
Args:
actor_name: actor 注册名
timeout: 最长等待秒数;``<=0`` 表示只试一次(等价于直接取句柄)
interval: 轮询间隔秒数
Raises:
TimeoutError: 超时仍未就绪。原始异常通过 ``raise ... from`` 链保留。
"""
deadline = time.monotonic() + max(timeout, 0.0)
last_err: Exception | None = None
while True:
try:
return _get_cached_actor_handle(actor_name)
except Exception as e: # ray.get_actor 失败一般是 ValueError
last_err = e
# 失败不能让 lru_cache 留下脏数据(异常本身不会被缓存,
# 但若底层换实现,这里清一次更稳妥)
if time.monotonic() >= deadline:
raise TimeoutError(
f"等待 actor {actor_name!r} 就绪超时({timeout}s):{last_err}"
) from last_err
time.sleep(interval)
def ray_actor_hook(*actor_names: str, timeout: float = 0.0, interval: float = 0.5):
"""按名字批量取出 Ray Actor 句柄,组装成一个 ``ActorList`` 返回。
例:``actors = ray_actor_hook("postgres_database", "global_state_machine")``
随后即可用 ``actors.postgres_database`` 拿到对应句柄。
Args:
timeout: ``>0`` 时对每个 actor 走 ``wait_for_actor`` 等待就绪(启动期用);
缺省 ``0`` 保持原"快速失败"语义——actor 不在立即抛异常。
interval: 等待轮询间隔,仅在 ``timeout>0`` 时生效。
"""
actor_list = ActorList()
for actor_name in actor_names:
handle = _get_cached_actor_handle(actor_name)
if timeout > 0:
handle = wait_for_actor(
actor_name, timeout=timeout, interval=interval
)
else:
handle = _get_cached_actor_handle(actor_name)
setattr(actor_list, actor_name, handle)
return actor_list