vllm: [Bug]: TRACKING ISSUE: `AsyncEngineDeadError`
Your current environment
The output of `python collect_env.py`
🐛 Describe the bug
Recently, we have seen reports of AsyncEngineDeadError, including:
- #5060
- #2000
- #3310
- #3839
- #4000
- #4135
- #4293
- #5443
- #5732
- #5822
- #6190
- #6208
- #6361
- #6421
- #6614
- #6790
- #6969
- #7356
If you see something like the following, please report here:
2024-06-25 12:27:29.905 File "/usr/local/lib/python3.10/dist-packages/vllm/entrypoints/openai/api_server.py", line 84, in health
2024-06-25 12:27:29.905 await openai_serving_chat.engine.check_health()
2024-06-25 12:27:29.905 File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 839, in check_health
2024-06-25 12:27:29.905 raise AsyncEngineDeadError("Background loop is stopped.")
2024-06-25 12:27:29.905 vllm.engine.async_llm_engine.AsyncEngineDeadError: Background loop is stopped.
Key areas we are looking into include:
- logprob usage
- guided regex usage
When reporting an issue, please include a sample request that causes the issue so we can reproduce on our side.
About this issue
- Original URL
- State: open
- Created 2 years ago
- Reactions: 14
- Comments: 23 (1 by maintainers)
Got the same issue across multiple versions of official vllm docker images. I first encountered this issue with version v0.4.3, then 0.5.0, 0.5.3.post1, now 0.5.4.
envs and launch commands
My environment info:
My launch command is: (part of a k8s yaml file)
This error seems ALWAYS occurs under a continuous inference workload, before version v0.5.4, the service can realize its own error status and restart automatically, but with v0.5.4, auto restart doesn’t work.
error log
Then the service does not restart, just stuck here.
question
Thanks!
[Bug] No available block found in 60 second in shm
Your current environment
/path/vllm/vllm/usage/usage_lib.py:19: RuntimeWarning: Failed to read commit hash: No module named ‘vllm.commit_id’ from vllm.version import version as VLLM_VERSION PyTorch version: 2.3.0+cu121 Is debug build: False CUDA used to build PyTorch: 12.1 ROCM used to build PyTorch: N/A OS: Ubuntu 22.04.3 LTS (x86_64) GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0 Clang version: Could not collect CMake version: version 3.30.0 Libc version: glibc-2.35 Python version: 3.10.14 (main, May 6 2024, 19:42:50) [GCC 11.2.0] (64-bit runtime) Python platform: Linux-4.18.0-425.3.1.el8.x86_64-x86_64-with-glibc2.35 Is CUDA available: True CUDA runtime version: 12.2.140 CUDA_MODULE_LOADING set to: LAZY GPU models and configuration: GPU 0: NVIDIA A800-SXM4-80GB GPU 1: NVIDIA A800-SXM4-80GB GPU 2: NVIDIA A800-SXM4-80GB GPU 3: NVIDIA A800-SXM4-80GB GPU 4: NVIDIA A800-SXM4-80GB GPU 5: NVIDIA A800-SXM4-80GB GPU 6: NVIDIA A800-SXM4-80GB GPU 7: NVIDIA A800-SXM4-80GB
Nvidia driver version: 525.60.13
cuDNN version: Probably one of the following: /usr/lib/x86_64-linux-gnu/libcudnn.so.8.9.5 /usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.9.5 /usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.9.5 /usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.9.5 /usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.9.5 /usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.9.5 /usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.9.5
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True CPU: Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Address sizes: 46 bits physical, 57 bits virtual Byte Order: Little Endian CPU(s): 128 On-line CPU(s) list: 0-127 Vendor ID: GenuineIntel Model name: Intel® Xeon® Platinum 8358 CPU @ 2.60GHz CPU family: 6 Model: 106 Thread(s) per core: 2 Core(s) per socket: 32 Socket(s): 2 Stepping: 6 CPU max MHz: 3400.0000 CPU min MHz: 800.0000 BogoMIPS: 5200.00
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 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 invpcid_single intel_ppin ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi 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 wbnoinvd dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid fsrm md_clear pconfig flush_l1d arch_capabilities
Virtualization: VT-x L1d cache: 3 MiB (64 instances) L1i cache: 2 MiB (64 instances) L2 cache: 80 MiB (64 instances) L3 cache: 96 MiB (2 instances) NUMA node(s): 1 NUMA node0 CPU(s): 0-127 Vulnerability Itlb multihit: Not affected Vulnerability L1tf: Not affected Vulnerability Mds: Not affected Vulnerability Meltdown: Not affected Vulnerability Mmio stale data: Vulnerable: Clear CPU buffers attempted, no microcode; SMT vulnerable Vulnerability Retbleed: Not affected Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Enhanced IBRS, IBPB conditional, RSB filling, PBRSB-eIBRS SW sequence Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected Versions of relevant libraries: [pip3] numpy==1.26.4 [pip3] nvidia-nccl-cu12==2.20.5 [pip3] torch==2.3.0 [pip3] torchvision==0.18.0 [pip3] transformers==4.42.4 [pip3] triton==2.3.0 [conda] numpy 1.26.4 pypi_0 pypi [conda] nvidia-nccl-cu12 2.20.5 pypi_0 pypi [conda] torch 2.3.0 pypi_0 pypi [conda] torchvision 0.18.0 pypi_0 pypi [conda] transformers 4.42.4 pypi_0 pypi [conda] triton 2.3.0 pypi_0 pypi ROCM Version: Could not collect Neuron SDK Version: N/A vLLM Version: 0.5.2 vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0 GPU1 GPU2 GPU3 GPU4 GPU5 GPU6 GPU7 mlx5_0 mlx5_1 mlx5_4 mlx5_5 CPU Affinity NUMA Affinity GPU0 X NV8 NV8 NV8 NV8 NV8 NV8 NV8 PXB SYS SYS SYS 0-127 N/A GPU1 NV8 X NV8 NV8 NV8 NV8 NV8 NV8 PXB SYS SYS SYS 0-127 N/A GPU2 NV8 NV8 X NV8 NV8 NV8 NV8 NV8 SYS PXB SYS SYS 0-127 N/A GPU3 NV8 NV8 NV8 X NV8 NV8 NV8 NV8 SYS PXB SYS SYS 0-127 N/A GPU4 NV8 NV8 NV8 NV8 X NV8 NV8 NV8 SYS SYS PXB SYS 0-127 N/A GPU5 NV8 NV8 NV8 NV8 NV8 X NV8 NV8 SYS SYS PXB SYS 0-127 N/A GPU6 NV8 NV8 NV8 NV8 NV8 NV8 X NV8 SYS SYS SYS PXB 0-127 N/A GPU7 NV8 NV8 NV8 NV8 NV8 NV8 NV8 X SYS SYS SYS PXB 0-127 N/A mlx5_0 PXB PXB SYS SYS SYS SYS SYS SYS X SYS SYS SYS mlx5_1 SYS SYS PXB PXB SYS SYS SYS SYS SYS X SYS SYS mlx5_4 SYS SYS SYS SYS PXB PXB SYS SYS SYS SYS X SYS mlx5_5 SYS SYS SYS SYS SYS SYS PXB PXB SYS SYS SYS 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
To support our model, I built it from source code.
🐛 Describe the bug
#6614.
I printed out the rank information at the warning and found that one of the GPUs was stuck (a total of 4 GPUs for tensor parallelism). This bug is highly reproducible, especially when running models above 70B(like qwen2)and encountering a large number of requests.
I think it’s because there are requests still in the queue, but since the process is stuck, no tokens have been generated, which eventually caused the server to crash.
after server dead
Environment: A800 *8 Nvidia driver 535.129.01 -> 535.169.07 or 535.183.06
I experienced the issue too.
My setup:
The issue seems to happen when I send around 30 parallel requests each (few) seconds. The prompt is about 1400-tokens-long.
A possible fix method can be found in #issuecomment-2745110814
@titu1994 @fengyizhu @doYrobot @tuobay @etiennebonnafoux
Here is mine
VLLM serve with the command
The Input
The error