vllm icon indicating copy to clipboard operation
vllm copied to clipboard

Published 20 hours ago • verified publisher icon vllm-project

[Bug]: GPU Placement Group Creation Error in Multi-Node Setup with vLLM

Open jerilkuriakose opened this issue 1 week ago • 3 comments

Your current environment

The output of `python collect_env.py` 
INFO 02-17 00:50:40 __init__.py:190] Automatically detected platform cuda.
Collecting environment information...
PyTorch version: 2.5.1+cu124
Is debug build: False
CUDA used to build PyTorch: 12.4
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 3.31.4
Libc version: glibc-2.35

Python version: 3.12.9 (main, Feb  5 2025, 08:49:00) [GCC 11.4.0] (64-bit runtime)
Python platform: Linux-5.15.0-1078-azure-x86_64-with-glibc2.35
Is CUDA available: True
CUDA runtime version: 12.1.105
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration: GPU 0: NVIDIA H100 NVL
Nvidia driver version: 550.127.08
cuDNN version: Could not collect
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:                        48 bits physical, 48 bits virtual
Byte Order:                           Little Endian
CPU(s):                               40
On-line CPU(s) list:                  0-39
Vendor ID:                            AuthenticAMD
Model name:                           AMD EPYC 9V84 96-Core Processor
CPU family:                           25
Model:                                17
Thread(s) per core:                   1
Core(s) per socket:                   40
Socket(s):                            1
Stepping:                             1
BogoMIPS:                             4800.05
Flags:                                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl tsc_reliable nonstop_tsc cpuid extd_apicid aperfmperf tsc_known_freq pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw topoext perfctr_core invpcid_single vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves avx512_bf16 clzero xsaveerptr rdpru arat avx512vbmi umip avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq rdpid fsrm
Hypervisor vendor:                    Microsoft
Virtualization type:                  full
L1d cache:                            1.3 MiB (40 instances)
L1i cache:                            1.3 MiB (40 instances)
L2 cache:                             40 MiB (40 instances)
L3 cache:                             160 MiB (5 instances)
NUMA node(s):                         1
NUMA node0 CPU(s):                    0-39
Vulnerability Gather data sampling:   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 Reg file data sampling: Not affected
Vulnerability Retbleed:               Not affected
Vulnerability Spec rstack overflow:   Mitigation; safe RET, no microcode
Vulnerability Spec store bypass:      Vulnerable
Vulnerability Spectre v1:             Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:             Mitigation; Retpolines; STIBP disabled; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                  Not affected
Vulnerability Tsx async abort:        Not affected

Versions of relevant libraries:
[pip3] numpy==1.26.4
[pip3] nvidia-cublas-cu12==12.4.5.8
[pip3] nvidia-cuda-cupti-cu12==12.4.127
[pip3] nvidia-cuda-nvrtc-cu12==12.4.127
[pip3] nvidia-cuda-runtime-cu12==12.4.127
[pip3] nvidia-cudnn-cu12==9.1.0.70
[pip3] nvidia-cufft-cu12==11.2.1.3
[pip3] nvidia-curand-cu12==10.3.5.147
[pip3] nvidia-cusolver-cu12==11.6.1.9
[pip3] nvidia-cusparse-cu12==12.3.1.170
[pip3] nvidia-ml-py==12.570.86
[pip3] nvidia-nccl-cu12==2.21.5
[pip3] nvidia-nvjitlink-cu12==12.4.127
[pip3] nvidia-nvtx-cu12==12.4.127
[pip3] pyzmq==26.2.1
[pip3] torch==2.5.1
[pip3] torchaudio==2.5.1
[pip3] torchvision==0.20.1
[pip3] transformers==4.48.2
[pip3] triton==3.1.0
[conda] Could not collect
ROCM Version: Could not collect
Neuron SDK Version: N/A
vLLM Version: 0.7.2
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0	NIC0	CPU Affinity	NUMA Affinity	GPU NUMA ID
GPU0	 X 	SYS	0-39	0		N/A
NIC0	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

NIC Legend:

  NIC0: mlx5_0

NVIDIA_VISIBLE_DEVICES=GPU-720b9a86-454f-1c5d-2916-ee394fb397a1
NVIDIA_REQUIRE_CUDA=cuda>=12.1 brand=tesla,driver>=470,driver<471 brand=unknown,driver>=470,driver<471 brand=nvidia,driver>=470,driver<471 brand=nvidiartx,driver>=470,driver<471 brand=geforce,driver>=470,driver<471 brand=geforcertx,driver>=470,driver<471 brand=quadro,driver>=470,driver<471 brand=quadrortx,driver>=470,driver<471 brand=titan,driver>=470,driver<471 brand=titanrtx,driver>=470,driver<471 brand=tesla,driver>=525,driver<526 brand=unknown,driver>=525,driver<526 brand=nvidia,driver>=525,driver<526 brand=nvidiartx,driver>=525,driver<526 brand=geforce,driver>=525,driver<526 brand=geforcertx,driver>=525,driver<526 brand=quadro,driver>=525,driver<526 brand=quadrortx,driver>=525,driver<526 brand=titan,driver>=525,driver<526 brand=titanrtx,driver>=525,driver<526
NCCL_VERSION=2.17.1-1
NVIDIA_DRIVER_CAPABILITIES=compute,utility
NCCL_P2P_LEVEL=NVL
NCCL_DEBUG=INFO
NCCL_IB_HCA=mlx5
NVIDIA_PRODUCT_NAME=CUDA
VLLM_USAGE_SOURCE=production-docker-image
NVIDIA_CUDA_END_OF_LIFE=1
CUDA_VERSION=12.1.0
CUDA_DEVICE_MAX_CONNECTIONS=1
LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64
VLLM_HOST_IP=10.61.0.31
NCCL_CUMEM_ENABLE=0
TORCHINDUCTOR_COMPILE_THREADS=1
CUDA_MODULE_LOADING=LAZY

🐛 Describe the bug

Issue Description

We're running into an issue while trying to run vLLM with multiple GPUs (16) in a Kubernetes environment. Based on the Ray dashboard, all 16 GPUs are visible and ALIVE, but we're getting placement group creation errors.

Environment

  • Running in Kubernetes
  • 16 GPUs across multiple nodes
  • Using Ray for distributed setup
  • DeepSeek model with tensor parallel size 16

Command Used

# For head node
ray start --head --port=6379 --block
vllm serve /models/DeepSeek-R1 \
    --tensor-parallel-size 16 \
    --trust-remote-code

# For worker nodes
ray start --address=${DEEPSEEK_HEAD_HOST}:6379 --block

Error Message

Waiting for creating a placement group of specs for 70 seconds... 
Error: No available node types can fulfill resource request {'node:10.61.0.27': 0.001, 'GPU': 1.0}

What We've Checked

  1. VLLM_HOST_IP is correctly set to respective node IPs
  2. Ray status shows all nodes are active. Here's the full output:
======== Autoscaler status: 2025-02-16 07:20:29.943586 ========
Node status ---------------------------------------------------------------
Active:
1 node_5b4c1d6ab97c55faf64f3224ac9b6faea81ebb0ae69e178975c25ebc
1 node_3f73b2c67f34864c90a1de6b4dbc5e2c768f69c942fc998bba45d88f
1 node_7141dbe867cde1531b536562d8e263c2f1cf8fec0f524c74441c1dbf
1 node_5af1da785cca84ea07fda287d856b4f4dfe21ad58db096bca0bf8cad
1 node_ee91de95745c6a3e13c7b1a44c808fe72d0481da0dca0a37f7e49b06
1 node_804192a5060a466a7e2ea7e87462939a00ca0a06e3f11f6d754e545a
1 node_c34fa4dd425a6a2318f8a7a7ca8f3bac2f9cdf2f5b95ef98e42b69e7
1 node_0d82d80055d65bc06fb2d2540ecd6bcaef3e2dd8820f6e668934a386
1 node_69c91c067af5a41f9d25abed4dc8d41ec95b90949ac357616c35acd4
1 node_1697bb7c4cda15748b21ba7f92dd8d962437010d2389f80a30764165
1 node_df0a162714bc3d166903077ee66552014834818062019a0d1fe67e9b
1 node_a735dc7333532fcc11e36fd14936f070d8ab5cbb88d8fbe35491c89f
1 node_46d575d1910717b4a10e2068f863943fb94c880425470b0e29ed0a14
1 node_3894767246def2895d50cab222be279647db6e4e6c8835f5010a3bc9
1 node_379acf612df6caaee73110ece8521f7ea836eee2669e690736251a1d
1 node_4831907df968564ac2da68c437860a7dc7021b277a5cb3560dddf85c

Pending: (no pending nodes)
Recent failures: (no failures)

Resources ---------------------------------------------------------------
Usage: 0.0/640.0 CPU
       0.0/16.0 GPU
       0B/4.76TiB memory
       0B/149.01GiB object_store_memory

Demands: {'node:10.61.0.27': 0.001, 'GPU': 1.0} * 1, {'GPU': 1.0} * 15 (PACK): 1+ pending placement groups
  1. Ray dashboard shows all GPUs as ALIVE (screenshots attached)
  2. Resources show 16 GPUs available

Screenshots

Screenshot of Ray Dashboard showing GPU status Screenshot of Job Details showing the error

Question

What might be causing this placement group creation failure despite having all GPUs available? Are we missing any specific configuration for multi-node GPU setup?

Additional Context

We've verified that:

  • All nodes can see each other
  • The Ray cluster is properly formed
  • GPUs are properly detected by Ray
  • VLLM_HOST_IP environment variable is correctly set on all nodes

Any guidance would be greatly appreciated.

Before submitting a new issue...

  • [x] Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions.

jerilkuriakose avatar Feb 17 '25 08:02 jerilkuriakose