bump version to v0.7.3 (#3416)

* bump version to v0.7.3

* add rayexecutor release timeout (#3403)

* add release timeout

* change log level

---------

Co-authored-by: q yao <[email protected]>

Author: lvhan028

docs/en/get_started/installation.md

@@ -23,7 +23,7 @@ pip install lmdeploy
 The default prebuilt package is compiled on **CUDA 12**. If CUDA 11+ (>=11.3) is required, you can install lmdeploy by:
 
 ```shell
-export LMDEPLOY_VERSION=0.7.2.post1
+export LMDEPLOY_VERSION=0.7.3
 export PYTHON_VERSION=38
 pip install https://github.com/InternLM/lmdeploy/releases/download/v${LMDEPLOY_VERSION}/lmdeploy-${LMDEPLOY_VERSION}+cu118-cp${PYTHON_VERSION}-cp${PYTHON_VERSION}-manylinux2014_x86_64.whl --extra-index-url https://download.pytorch.org/whl/cu118
 ```

docs/zh_cn/get_started/installation.md

@@ -23,7 +23,7 @@ pip install lmdeploy
 默认的预构建包是在 **CUDA 12** 上编译的。如果需要 CUDA 11+ (>=11.3)，你可以使用以下命令安装 lmdeploy：
 
 ```shell
-export LMDEPLOY_VERSION=0.7.2.post1
+export LMDEPLOY_VERSION=0.7.3
 export PYTHON_VERSION=38
 pip install https://github.com/InternLM/lmdeploy/releases/download/v${LMDEPLOY_VERSION}/lmdeploy-${LMDEPLOY_VERSION}+cu118-cp${PYTHON_VERSION}-cp${PYTHON_VERSION}-manylinux2014_x86_64.whl --extra-index-url https://download.pytorch.org/whl/cu118
 ```

lmdeploy/pytorch/engine/executor/ray_executor.py

@@ -230,13 +230,17 @@ def __init__(self,
             logger.info('Warming up distribute environment, this might take long time, please waiting...')
             ray.get([worker.warmup_dist.remote() for worker in self.workers])
 
-    def collective_rpc(self, method: str, args: Tuple[Any] = None, kwargs: Dict[str, Any] = None):
+    def collective_rpc(self,
+                       method: str,
+                       args: Tuple[Any] = None,
+                       kwargs: Dict[str, Any] = None,
+                       timeout: float = None):
         """collective rpc."""
         if args is None:
             args = list()
         if kwargs is None:
             kwargs = dict()
-        return ray.get([getattr(worker, method).remote(*args, **kwargs) for worker in self.workers])
+        return ray.get([getattr(worker, method).remote(*args, **kwargs) for worker in self.workers], timeout=timeout)
 
     def build_model(self):
         """build model."""
@@ -293,9 +297,21 @@ def stop(self):
 
     def release(self):
         """release."""
-        self.collective_rpc('release')
-        for worker in self.workers:
-            ray.kill(worker)
+        if self.dp == 1:
+            try:
+                self.collective_rpc('release', timeout=5.0)
+                logger.debug('RayExecutor workers released.')
+            except ray.exceptions.GetTimeoutError:
+                logger.info('Ray release timeout.')
+
+            try:
+                self.collective_rpc('exit')
+                logger.debug('RayExecutor workers exited.')
+            except ray.exceptions.RayActorError as e:
+                logger.debug(f'ray actor exit: {e}')
+        else:
+            [ray.kill(worker) for worker in self.workers]
+
         ray.util.remove_placement_group(self.placement_group)
 
     def _compile_dag(self):

lmdeploy/version.py

@@ -1,7 +1,7 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 from typing import Tuple
 
-__version__ = '0.7.2.post1'
+__version__ = '0.7.3'
 short_version = __version__