end

Signed-off-by: MrZ20 <[email protected]>

Author: MrZ20

.github/workflows/nightly_test_a3.yaml

@@ -74,11 +74,11 @@ jobs:
           - name: multi-node-qwenw8a8-2node-longseq
             config_file_path: Qwen3-235B-W8A8-longseq.yaml
             size: 2
-          - name: multi-node-qwen3-235b-a22b-pd
-            config_file_path: Qwen3-235B-A22B-pd.yaml
+          - name: multi-node-qwen-disagg-pd
+            config_file_path: Qwen3-235B-disagg-pd.yaml
             size: 2
-          - name: multi-node-qwen3-vl-235b-a22b-instruct
-            config_file_path: Qwen3-VL-235B-A22B-Instruct.yaml
+          - name: multi-node-qwen-vl-disagg-pd
+            config_file_path: Qwen3-VL-235B-disagg-pd.yaml
             size: 2
     uses: ./.github/workflows/_e2e_nightly_multi_node.yaml
     with:

…ulti_node/config/Qwen3-235B-A22B-pd.yaml …ti_node/config/Qwen3-235B-disagg-pd.yamltests/e2e/nightly/multi_node/config/Qwen3-235B-A22B-pd.yaml renamed to tests/e2e/nightly/multi_node/config/Qwen3-235B-disagg-pd.yaml tests/e2e/nightly/multi_node/config/Qwen3-235B-A22B-pd.yaml renamed to tests/e2e/nightly/multi_node/config/Qwen3-235B-disagg-pd.yaml

@@ -1,4 +1,4 @@
-test_name: "test Qwen3-235B-A22B pd online"
+test_name: "test Qwen3-235B-A22B disaggregated_prefill"
 model: "Qwen/Qwen3-235B-A22B"
 num_nodes: 2
 npu_per_node: 16

…/config/Qwen3-VL-235B-A22B-Instruct.yaml …node/config/Qwen3-VL-235B-disagg-pd.yamltests/e2e/nightly/multi_node/config/Qwen3-VL-235B-A22B-Instruct.yaml renamed to tests/e2e/nightly/multi_node/config/Qwen3-VL-235B-disagg-pd.yaml tests/e2e/nightly/multi_node/config/Qwen3-VL-235B-A22B-Instruct.yaml renamed to tests/e2e/nightly/multi_node/config/Qwen3-VL-235B-disagg-pd.yaml

@@ -1,4 +1,4 @@
-test_name: "test qwen3-vl-235b-a22b disaggregated_prefill on A3"
+test_name: "test Qwen3-VL-235B-A22B disaggregated_prefill"
 model: "Qwen/Qwen3-VL-235B-A22B-Instruct"
 num_nodes: 2
 npu_per_node: 16
@@ -11,7 +11,6 @@ env_common:
   HCCL_OP_EXPANSION_MODE: "AIV"
   TASK_QUEUE_ENABLE: 1
   PYTORCH_NPU_ALLOC_CONF: expandable_segments:True
-  DISAGGREGATED_PREFILL_PROXY_SCRIPT: "examples/disaggregated_prefill_v1/load_balance_proxy_server_example.py"
 
 disaggregated_prefill:
   enabled: true

vllm_ascend/ascend_config.py

@@ -102,8 +102,12 @@ def __init__(self, vllm_config: "VllmConfig"):
                     decode_tp_size = min(decode_tp_size, num_kv_head)
                     self.pd_head_ratio = prefill_tp_size // decode_tp_size
                 except Exception:
-                    raise AssertionError(
-                        "Can not get num_key_value_heads from model_config")
+                    raise ValueError(
+                        "The text_config extracted from the model config does not have "
+                        "`num_key_value_heads` attribute. This indicates a mismatch "
+                        "between the model config and vLLM's expectations. Please "
+                        "ensure that the model config is compatible with vLLM."
+                    )
 
             if self.pd_tp_ratio == 0:
                 raise AssertionError(