[Config] Raise ValueError instead of assert for DBO all2all backend validation

tests/test_config.py

@@ -2621,3 +2621,28 @@ def test_revision_resolved_when_weights_match_model(mock_resolve):
     assert isinstance(config.revision, ResolvedRevision)
     assert config.revision.resolved == REVISION
     mock_resolve.assert_any_call(model, None, config.hf_token)
+
+
+@pytest.mark.skip_global_cleanup
+@patch("vllm.config.model.get_config")
+@patch("vllm.config.model.resolve_revision", return_value=ResolvedRevision(REVISION))
+def test_microbatching_all2all_backend_validation(mock_resolve, mock_get_config):
+    from transformers import PretrainedConfig
+    hf_config = PretrainedConfig(
+        model_type="qwen3",
+        num_hidden_layers=1,
+        architectures=["Qwen3ForCausalLM"],
+    )
+    mock_get_config.return_value = hf_config
+
+    # Microbatching requires specific all2all backends.
+    # By default, use_ubatching is True when enable_dbo is True,
+    # but all2all_backend defaults to allgather_reducescatter.
+    # This must raise a ValueError.
+    from vllm.config import DeviceConfig
+    with pytest.raises(ValueError, match="Microbatching currently only supports"):
+        VllmConfig(
+            model_config=ModelConfig("Qwen/Qwen3-0.6B"),
+            parallel_config=ParallelConfig(enable_dbo=True),
+            device_config=DeviceConfig(device="cpu"),
+        )

vllm/config/vllm.py

@@ -1740,18 +1740,19 @@ def has_blocked_weights():
 
         if self.parallel_config.use_ubatching:
             a2a_backend = self.parallel_config.all2all_backend
-            assert a2a_backend in [
+            if a2a_backend not in [
                 "deepep_low_latency",
                 "deepep_high_throughput",
                 "nixl_ep",
-            ], (
-                "Microbatching currently only supports the deepep_low_latency, "
-                "deepep_high_throughput, and nixl_ep all2all backends. "
-                f"{a2a_backend} is not supported. To fix use "
-                "--all2all-backend=deepep_low_latency, "
-                "--all2all-backend=deepep_high_throughput, or "
-                "--all2all-backend=nixl_ep and install the matching kernels."
-            )
+            ]:
+                raise ValueError(
+                    "Microbatching currently only supports the deepep_low_latency, "
+                    "deepep_high_throughput, and nixl_ep all2all backends. "
+                    f"{a2a_backend} is not supported. To fix use "
+                    "--all2all-backend=deepep_low_latency, "
+                    "--all2all-backend=deepep_high_throughput, or "
+                    "--all2all-backend=nixl_ep and install the matching kernels."
+                )
 
             if not self.model_config.disable_cascade_attn:
                 self.model_config.disable_cascade_attn = True