bugfix

slime/ray/rollout.py

@@ -283,7 +283,7 @@ def recover(self):
         release_handles = []
         updatable_new_engines = []
         non_updatable_groups_engines: list[tuple[str, list]] = []
-        for g, dead_indices in zip(self.engine_groups, dead_per_group, strict=True):
+        for g, dead_indices in zip(self.server_groups, dead_per_group, strict=True):
             logger.info(f"Recovered {g.num_new_engines} dead rollout engines (worker_type={g.worker_type})")
             assert g.num_new_engines == len(dead_indices), "num_new_engines does not match dead_indices length"
             if g.needs_offload and dead_indices:
@@ -331,7 +331,7 @@ def onload_weights(self):
         CPU backup already contains the correct (unchanged) weights.
         """
         handles = []
-        for g in self.engine_groups:
+        for g in self.server_groups:
             if not g.needs_offload:
                 continue
             handles.extend(g.onload(tags=[GPU_MEMORY_TYPE_WEIGHTS]))
@@ -340,7 +340,7 @@ def onload_weights(self):
     def onload_kv(self):
         """Resume KV cache and CUDA graphs for offloaded groups."""
         handles = []
-        for g in self.engine_groups:
+        for g in self.server_groups:
             handles.extend(g.onload(tags=[GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_CUDA_GRAPH]))
         return ray.get(handles) if handles else []
 

tests/test_qwen2.5_0.5B_sglang_config.py

@@ -16,7 +16,7 @@
 SGLANG_CONFIG_YAML = """\
 sglang:
   - name: default
-    engine_groups:
+    server_groups:
       - worker_type: regular
         num_gpus: 4
         num_gpus_per_engine: 2

tests/test_qwen2.5_0.5B_sglang_config_distributed.py

@@ -17,7 +17,7 @@
 SGLANG_CONFIG_YAML = """\
 sglang:
   - name: default
-    engine_groups:
+    server_groups:
       - worker_type: regular
         num_gpus: 2
         num_gpus_per_engine: 2

tests/test_sglang_config_mixed_offload.py

@@ -29,13 +29,13 @@
 sglang:
   - name: actor
     update_weights: true
-    engine_groups:
+    server_groups:
       - worker_type: regular
         num_gpus: 4
         num_gpus_per_engine: 1
   - name: ref
     update_weights: false
-    engine_groups:
+    server_groups:
       - worker_type: regular
         num_gpus: 4
         num_gpus_per_engine: 1

tests/test_sglang_config_mixed_offload_ft.py

@@ -26,13 +26,13 @@
 sglang:
   - name: actor
     update_weights: true
-    engine_groups:
+    server_groups:
       - worker_type: regular
         num_gpus: 4
         num_gpus_per_engine: 1
   - name: ref
     update_weights: false
-    engine_groups:
+    server_groups:
       - worker_type: regular
         num_gpus: 4
         num_gpus_per_engine: 1