[Bug]: vllm slows down after a long run

Your current environment

Collecting environment information...
/data/miniconda3_new/envs/vllm/lib/python3.10/site-packages/transformers/utils/hub.py:124: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.
  warnings.warn(
PyTorch version: 2.1.2+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.2 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: Could not collect
CMake version: version 3.28.0
Libc version: glibc-2.35

Python version: 3.10.0 | packaged by conda-forge | (default, Nov 20 2021, 02:24:10) [GCC 9.4.0] (64-bit runtime)
Python platform: Linux-6.5.0-21-generic-x86_64-with-glibc2.35
Is CUDA available: True
CUDA runtime version: Could not collect
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration: 
GPU 0: NVIDIA GeForce RTX 3090
GPU 1: NVIDIA GeForce RTX 3090
GPU 2: NVIDIA GeForce RTX 3090
GPU 3: NVIDIA GeForce RTX 3090

Nvidia driver version: 525.147.05
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
架构：                              x86_64
CPU 运行模式：                      32-bit, 64-bit
Address sizes:                      46 bits physical, 48 bits virtual
字节序：                            Little Endian
CPU:                                64
在线 CPU 列表：                     0-63
厂商 ID：                           GenuineIntel
型号名称：                          Intel(R) Xeon(R) Gold 5218 CPU @ 2.30GHz
CPU 系列：                          6
型号：                              85
每个核的线程数：                    2
每个座的核数：                      16
座：                                2
步进：                              7
CPU 最大 MHz：                      3900.0000
CPU 最小 MHz：                      1000.0000
BogoMIPS：                          4600.00
标记：                              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 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 cdp_l3 invpcid_single intel_ppin ssbd mba ibrs ibpb stibp ibrs_enhanced tpr_shadow flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts vnmi pku ospke avx512_vnni md_clear flush_l1d arch_capabilities
虚拟化：                            VT-x
L1d 缓存：                          1 MiB (32 instances)
L1i 缓存：                          1 MiB (32 instances)
L2 缓存：                           32 MiB (32 instances)
L3 缓存：                           44 MiB (2 instances)
NUMA 节点：                         2
NUMA 节点0 CPU：                    0-15,32-47
NUMA 节点1 CPU：                    16-31,48-63
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit:        KVM: Mitigation: VMX disabled
Vulnerability L1tf:                 Not affected
Vulnerability Mds:                  Not affected
Vulnerability Meltdown:             Not affected
Vulnerability Mmio stale data:      Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Retbleed:             Mitigation; Enhanced IBRS
Vulnerability Spec rstack overflow: 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 / Automatic IBRS, IBPB conditional, RSB filling, PBRSB-eIBRS SW sequence
Vulnerability Srbds:                Not affected
Vulnerability Tsx async abort:      Mitigation; TSX disabled

Versions of relevant libraries:
[pip3] mypy-extensions==1.0.0
[pip3] numpy==1.24.4
[pip3] onnx==1.15.0
[pip3] onnxruntime==1.15.1
[pip3] rapidocr-onnxruntime==1.3.8
[pip3] torch==2.1.2
[pip3] torchaudio==2.1.2
[pip3] torchvision==0.16.2
[pip3] triton==2.1.0
[conda] numpy                     1.24.4                   pypi_0    pypi
[conda] torch                     2.1.2                    pypi_0    pypi
[conda] torchaudio                2.1.2                    pypi_0    pypi
[conda] torchvision               0.16.2                   pypi_0    pypi
[conda] triton                    2.1.0                    pypi_0    pypiROCM Version: Could not collect
Neuron SDK Version: N/A
vLLM Version: 0.2.6
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0    GPU1    GPU2    GPU3    NIC0    NIC1    CPU Affinity    NUMA Affinity
GPU0     X  PIX NODE    NODE    NODE    NODE    0-15,32-47  0
GPU1    PIX  X  NODE    NODE    NODE    NODE    0-15,32-47  0
GPU2    NODE    NODE     X  PIX NODE    NODE    0-15,32-47  0
GPU3    NODE    NODE    PIX  X  NODE    NODE    0-15,32-47  0
NIC0    NODE    NODE    NODE    NODE     X  PIX     
NIC1    NODE    NODE    NODE    NODE    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
🐛 Describe the bug

vllm slows down after a long run about 12 hour at the begin : 6e7e68ebd53ed866ec9bde2fe698882a after 12 hour: 95cb08242ec6f27b37aa702af9fc9a5c
then i reboot vllm: speed recovery, but i loss some data
vllm-project / vllm

[Bug]: vllm slows down after a long run #3520

Your current environment

🐛 Describe the bug
