vllm - 💡(How to fix) Fix [Bug]: Frequent crashes with gemma4 MTP enabled [1 participants]

Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 25.10 (x86_64)
GCC version                  : (Ubuntu 15.2.0-4ubuntu4) 15.2.0
Clang version                : 20.1.8 (0ubuntu4)
CMake version                : version 3.28.4
Libc version                 : glibc-2.42

==============================
       PyTorch Info
==============================
PyTorch version              : 2.11.0+cu130
Is debug build               : False
CUDA used to build PyTorch   : 13.0
ROCM used to build PyTorch   : N/A
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.13.7 (main, Mar  3 2026, 12:19:54) [GCC 15.2.0] (64-bit runtime)
Python platform              : Linux-6.17.0-8-generic-x86_64-with-glibc2.42

==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : 12.4.131
CUDA_MODULE_LOADING set to   :
GPU models and configuration :
GPU 0: NVIDIA H200 NVL
GPU 1: NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU 2: NVIDIA RTX PRO 6000 Blackwell Server Edition

Nvidia driver version        : 590.48.01
cuDNN version                : Could not collect
HIP runtime version          : N/A
MIOpen runtime version       : N/A
Is XNNPACK available         : True

==============================
          CPU Info
==============================
Architecture:                            x86_64
CPU op-mode(s):                          32-bit, 64-bit
Address sizes:                           52 bits physical, 57 bits virtual
Byte Order:                              Little Endian
CPU(s):                                  192
On-line CPU(s) list:                     0-191
Vendor ID:                               GenuineIntel
Model name:                              Intel(R) Xeon(R) 6747P
CPU family:                              6
Model:                                   173
Thread(s) per core:                      2
Core(s) per socket:                      48
Socket(s):                               2
Stepping:                                1
CPU(s) scaling MHz:                      46%
CPU max MHz:                             3900.0000
CPU min MHz:                             800.0000
BogoMIPS:                                5400.00
Flags:                                   fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cat_l2 cdp_l3 intel_ppin cdp_l2 ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local split_lock_detect user_shstk avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req hfi vnmi avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr ibt amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization:                          VT-x
L1d cache:                               4.5 MiB (96 instances)
L1i cache:                               6 MiB (96 instances)
L2 cache:                                192 MiB (96 instances)
L3 cache:                                576 MiB (2 instances)
NUMA node(s):                            4
NUMA node0 CPU(s):                       0-23,96-119
NUMA node1 CPU(s):                       24-47,120-143
NUMA node2 CPU(s):                       48-71,144-167
NUMA node3 CPU(s):                       72-95,168-191
Vulnerability Gather data sampling:      Not affected
Vulnerability Ghostwrite:                Not affected
Vulnerability Indirect target selection: Not affected
Vulnerability Itlb multihit:             Not affected
Vulnerability L1tf:                      Not affected
Vulnerability Mds:                       Not affected
Vulnerability Meltdown:                  Not affected
Vulnerability Mmio stale data:           Not affected
Vulnerability Old microcode:             Not affected
Vulnerability Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Vulnerable
Vulnerability Spectre v1:                Vulnerable: __user pointer sanitization and usercopy barriers only; no swapgs barriers
Vulnerability Spectre v2:                Vulnerable; IBPB: disabled; STIBP: disabled; PBRSB-eIBRS: Not affected; BHI: Vulnerable
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Vulnerable

==============================
Versions of relevant libraries
==============================
[pip3] flashinfer-python==0.6.8.post1
[pip3] mypy_extensions==1.1.0
[pip3] numpy==2.3.1
[pip3] nvidia-cublas==13.1.0.3
[pip3] nvidia-cuda-cupti==13.0.85
[pip3] nvidia-cuda-nvrtc==13.0.88
[pip3] nvidia-cuda-runtime==13.0.96
[pip3] nvidia-cudnn-cu13==9.19.0.56
[pip3] nvidia-cudnn-frontend==1.18.0
[pip3] nvidia-cufft==12.0.0.61
[pip3] nvidia-cufile==1.15.1.6
[pip3] nvidia-curand==10.4.0.35
[pip3] nvidia-cusolver==12.0.4.66
[pip3] nvidia-cusparse==12.6.3.3
[pip3] nvidia-cusparselt-cu13==0.8.0
[pip3] nvidia-cutlass-dsl==4.5.0
[pip3] nvidia-cutlass-dsl-libs-base==4.5.0
[pip3] nvidia-ml-py==13.595.45
[pip3] nvidia-nccl-cu13==2.28.9
[pip3] nvidia-nvjitlink==13.0.88
[pip3] nvidia-nvshmem-cu13==3.4.5
[pip3] nvidia-nvtx==13.0.85
[pip3] onnx==1.21.0
[pip3] onnxruntime==1.24.4
[pip3] pyzmq==27.1.0
[pip3] torch==2.11.0
[pip3] torch_c_dlpack_ext==0.1.5
[pip3] torchaudio==2.11.0
[pip3] torchvision==0.26.0
[pip3] transformers==5.8.0
[pip3] triton==3.6.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : Could not collect
vLLM Version                 : 0.20.2
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; XPU: Disabled
GPU Topology:
  	GPU0	GPU1	GPU2	NIC0	NIC1	CPU Affinity	NUMA Affinity	GPU NUMA ID
GPU0	 X 	SYS	SYS	NODE	NODE	0-23,96-119	0		N/A
GPU1	SYS	 X 	NODE	SYS	SYS	48-71,144-167	2		N/A
GPU2	SYS	NODE	 X 	SYS	SYS	48-71,144-167	2		N/A
NIC0	NODE	SYS	SYS	 X 	PIX
NIC1	NODE	SYS	SYS	PIX	 X

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

NIC Legend:

  NIC0: mlx5_0
  NIC1: mlx5_1

==============================
     Environment Variables
==============================
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_tjwebb

---

/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [116,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [117,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [118,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [119,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [120,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [121,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [122,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [123,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [124,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [125,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [126,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
/pytorch/aten/src/ATen/native/cuda/Indexing.cu:1515: indexSelectSmallIndex: block: [1,0,0], thread: [127,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
(EngineCore pid=615) Process EngineCore:
ERROR 2026-05-10 22:01:53,336 vllm-router_vllm-31b-h200-1 8d5aszha 0964fb3b-7457-415e-9448-5b9cda258cdc -- Request failed.
Traceback (most recent call last):
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1310, in _handle_errors_and_metrics
    yield _status_code_callback
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 2084, in _wrap_request
    yield status_code_callback
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1640, in handle_request_with_rejection
    raise self._maybe_wrap_grpc_exception(e, request_metadata) from e
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1635, in handle_request_with_rejection
    yield await self._user_callable_wrapper.call_user_method(
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 3657, in call_user_method
    result, _ = await self._call_func_or_gen(
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 3148, in _call_func_or_gen
    result = await result
             ^^^^^^^^^^^^
  File "/app/serve/vllm_app.py", line 440, in chat_completion
    final, duration_s = await self._run(engine_input, sampling, request_id)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/serve/vllm_app.py", line 349, in _run
    async for output in self.engine.generate(prompt, sampling, request_id):
  File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/async_llm.py", line 579, in generate
    out = q.get_nowait() or await q.get()
                            ^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/output_processor.py", line 85, in get
    raise output
  File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/async_llm.py", line 660, in output_handler
    outputs = await engine_core.get_output_async()
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/core_client.py", line 998, in get_output_async
    raise self._format_exception(outputs) from None
vllm.v1.engine.exceptions.EngineDeadError: EngineCore encountered an issue. See stack trace (above) for the root cause.
(EngineCore pid=615) Traceback (most recent call last):
INFO 2026-05-10 22:01:53,337 vllm-router_vllm-31b-h200-1 8d5aszha 0964fb3b-7457-415e-9448-5b9cda258cdc -- CALL chat_completion ERROR 9392442.7ms
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/core.py", line 1140, in run_engine_core
(EngineCore pid=615)     raise e
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/core.py", line 1129, in run_engine_core
(EngineCore pid=615)     engine_core.run_busy_loop()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/core.py", line 1170, in run_busy_loop
(EngineCore pid=615)     self._process_engine_step()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/core.py", line 1209, in _process_engine_step
(EngineCore pid=615)     outputs, model_executed = self.step_fn()
(EngineCore pid=615)                               ^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/core.py", line 473, in step_with_batch_queue
(EngineCore pid=615)     exec_future = self.model_executor.execute_model(
(EngineCore pid=615)                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/executor/uniproc_executor.py", line 114, in execute_model
(EngineCore pid=615)     output.result()
(EngineCore pid=615)   File "/usr/lib/python3.12/concurrent/futures/_base.py", line 449, in result
(EngineCore pid=615)     return self.__get_result()
(EngineCore pid=615)            ^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/lib/python3.12/concurrent/futures/_base.py", line 401, in __get_result
(EngineCore pid=615)     raise self._exception
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/executor/uniproc_executor.py", line 84, in collective_rpc
(EngineCore pid=615)     result = run_method(self.driver_worker, method, args, kwargs)
(EngineCore pid=615)              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/serial_utils.py", line 510, in run_method
(EngineCore pid=615)     return func(*args, **kwargs)
(EngineCore pid=615)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker/worker_base.py", line 337, in execute_model
(EngineCore pid=615)     return self.worker.execute_model(scheduler_output)
(EngineCore pid=615)            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=615)     return func(*args, **kwargs)
(EngineCore pid=615)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker/gpu_worker.py", line 841, in execute_model
(EngineCore pid=615)     output = self.model_runner.execute_model(
(EngineCore pid=615)              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/torch/utils/_contextlib.py", line 124, in decorate_context
(EngineCore pid=615)     return func(*args, **kwargs)
(EngineCore pid=615)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker/gpu_model_runner.py", line 3879, in execute_model
(EngineCore pid=615)     self.synchronize_input_prep(),
(EngineCore pid=615)     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/lib/python3.12/contextlib.py", line 137, in __enter__
(EngineCore pid=615)     return next(self.gen)
(EngineCore pid=615)            ^^^^^^^^^^^^^^
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker/gpu_model_runner.py", line 3543, in synchronize_input_prep
(EngineCore pid=615)     self.prepare_inputs_event.synchronize()
(EngineCore pid=615) torch.AcceleratorError: CUDA error: device-side assert triggered
(EngineCore pid=615) Search for `cudaErrorAssert' in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html for more information.
(EngineCore pid=615) CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
(EngineCore pid=615) For debugging consider passing CUDA_LAUNCH_BLOCKING=1
(EngineCore pid=615) Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.
(EngineCore pid=615) 
(EngineCore pid=615) 
(EngineCore pid=615) During handling of the above exception, another exception occurred:
(EngineCore pid=615) 
(EngineCore pid=615) Traceback (most recent call last):
(EngineCore pid=615)   File "/usr/lib/python3.12/multiprocessing/process.py", line 314, in _bootstrap
(EngineCore pid=615)     self.run()
(EngineCore pid=615)   File "/usr/lib/python3.12/multiprocessing/process.py", line 108, in run
(EngineCore pid=615)     self._target(*self._args, **self._kwargs)
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/core.py", line 1147, in run_engine_core
(EngineCore pid=615)     engine_core.shutdown()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/core.py", line 574, in shutdown
(EngineCore pid=615)     self.model_executor.shutdown()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/executor/uniproc_executor.py", line 137, in shutdown
(EngineCore pid=615)     worker.shutdown()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker/worker_base.py", line 212, in shutdown
(EngineCore pid=615)     self.worker.shutdown()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker/gpu_worker.py", line 1049, in shutdown
(EngineCore pid=615)     model_runner.shutdown()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker/gpu_model_runner.py", line 5978, in shutdown
(EngineCore pid=615)     self._cleanup_profiling_kv_cache()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/vllm/v1/worker/gpu_model_runner.py", line 5986, in _cleanup_profiling_kv_cache
(EngineCore pid=615)     torch.accelerator.synchronize()
(EngineCore pid=615)   File "/usr/local/lib/python3.12/dist-packages/torch/accelerator/__init__.py", line 263, in synchronize
(EngineCore pid=615)     torch._C._accelerator_synchronizeDevice(device_index)
(EngineCore pid=615) torch.AcceleratorError: CUDA error: device-side assert triggered
(EngineCore pid=615) Search for `cudaErrorAssert' in https://docs.nvidia.com/cuda/cuda-runtime-api/group__CUDART__TYPES.html for more information.
(EngineCore pid=615) CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
(EngineCore pid=615) For debugging consider passing CUDA_LAUNCH_BLOCKING=1
(EngineCore pid=615) Compile with `TORCH_USE_CUDA_DSA` to enable device-side assertions.
(EngineCore pid=615) 
[rank0]:[W510 22:01:53.522015150 ProcessGroupNCCL.cpp:1575] Warning: WARNING: destroy_process_group() was not called before program exit, which can leak resources. For more info, please see https://pytorch.org/docs/stable/distributed.html#shutdown (function operator())
ERROR 2026-05-10 22:02:00,377 vllm-router_vllm-31b-h200-1 8d5aszha 5b1e9f25-0480-4763-bea3-ad42350696c9 -- Request failed.
Traceback (most recent call last):
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1310, in _handle_errors_and_metrics
    yield _status_code_callback
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 2084, in _wrap_request
    yield status_code_callback
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1640, in handle_request_with_rejection
    raise self._maybe_wrap_grpc_exception(e, request_metadata) from e
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1635, in handle_request_with_rejection
    yield await self._user_callable_wrapper.call_user_method(
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 3657, in call_user_method
    result, _ = await self._call_func_or_gen(
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 3148, in _call_func_or_gen
    result = await result
             ^^^^^^^^^^^^
  File "/app/serve/vllm_app.py", line 440, in chat_completion
    final, duration_s = await self._run(engine_input, sampling, request_id)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/serve/vllm_app.py", line 349, in _run
    async for output in self.engine.generate(prompt, sampling, request_id):
  File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/async_llm.py", line 559, in generate
    q = await self.add_request(
        ^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/async_llm.py", line 301, in add_request
    raise EngineDeadError()
vllm.v1.engine.exceptions.EngineDeadError: EngineCore encountered an issue. See stack trace (above) for the root cause.
INFO 2026-05-10 22:02:00,378 vllm-router_vllm-31b-h200-1 8d5aszha 5b1e9f25-0480-4763-bea3-ad42350696c9 -- CALL chat_completion ERROR 1.9ms
ERROR 2026-05-10 22:02:01,227 vllm-router_vllm-31b-h200-1 8d5aszha 82e5bd7f-53fc-4c12-9ce9-bb5dc80d945b -- Request failed.
Traceback (most recent call last):
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1310, in _handle_errors_and_metrics
    yield _status_code_callback
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 2084, in _wrap_request
    yield status_code_callback
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1640, in handle_request_with_rejection
    raise self._maybe_wrap_grpc_exception(e, request_metadata) from e
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 1635, in handle_request_with_rejection
    yield await self._user_callable_wrapper.call_user_method(
          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 3657, in call_user_method
    result, _ = await self._call_func_or_gen(
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/ray/serve/_private/replica.py", line 3148, in _call_func_or_gen
    result = await result
             ^^^^^^^^^^^^
  File "/app/serve/vllm_app.py", line 440, in chat_completion
    final, duration_s = await self._run(engine_input, sampling, request_id)
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/serve/vllm_app.py", line 349, in _run
    async for output in self.engine.generate(prompt, sampling, request_id):
  File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/async_llm.py", line 559, in generate
    q = await self.add_request(
        ^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/dist-packages/vllm/v1/engine/async_llm.py", line 301, in add_request
    raise EngineDeadError()
vllm.v1.engine.exceptions.EngineDeadError: EngineCore encountered an issue. See stack trace (above) for the root cause.