fix(os): fix FileNotFoundError in storage_manager (#455)

* use rank0 to makedirs

* use try-except to handle file error

* fix ci

internlm/utils/storage_manager.py

@@ -662,8 +662,11 @@ class LocalClient(StorageClient):
     def sync_upload_fileobj(fp: str, saved_obj=None, **kwargs):
         assert saved_obj is not None
         fp_dirname = os.path.dirname(fp)
-        if not os.path.exists(fp_dirname):
-            os.makedirs(fp_dirname, exist_ok=True)
+        try:
+            if not os.path.exists(fp_dirname):
+                os.makedirs(fp_dirname, exist_ok=True)
+        except FileNotFoundError:
+            pass
         torch.save(saved_obj, fp, **kwargs)
 
     @staticmethod