[Bug]: VLLM_ALL2ALL_BACKEND=naive hangs/crashes on multi nodes when serving DeepSeekV3 #23448
Description
Your current environment
The output of python collect_env.py
--2025-08-22 16:28:07-- https://raw.githubusercontent.com/vllm-project/vllm/main/vllm/collect_env.py
Resolving raw.githubusercontent.com (raw.githubusercontent.com)... 185.199.111.133, 185.199.108.133, 185.199.110.133, ...
Connecting to raw.githubusercontent.com (raw.githubusercontent.com)|185.199.111.133|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 28526 (28K) [text/plain]
Saving to: ‘collect_env.py’
collect_env.py 100%[====================================================================================>] 27.86K --.-KB/s in 0.003s
2025-08-22 16:28:07 (9.01 MB/s) - ‘collect_env.py’ saved [28526/28526]
(base) ray@k-64de8b3f752920001:~$ python collect_env.py
Collecting environment information...
==============================
System Info
==============================
OS : Ubuntu 22.04.5 LTS (x86_64)
GCC version : (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version : Could not collect
CMake version : version 4.1.0
Libc version : glibc-2.35
==============================
PyTorch Info
==============================
PyTorch version : 2.7.1+cu128
Is debug build : False
CUDA used to build PyTorch : 12.8
ROCM used to build PyTorch : N/A
==============================
Python Environment
==============================
Python version : 3.11.11 | packaged by conda-forge | (main, Mar 3 2025, 20:43:55) [GCC 13.3.0] (64-bit runtime)
Python platform : Linux-5.15.0-136-generic-x86_64-with-glibc2.35
==============================
CUDA / GPU Info
==============================
Is CUDA available : True
CUDA runtime version : Could not collect
CUDA_MODULE_LOADING set to : LAZY
GPU models and configuration :
GPU 0: NVIDIA H200
GPU 1: NVIDIA H200
GPU 2: NVIDIA H200
GPU 3: NVIDIA H200
GPU 4: NVIDIA H200
GPU 5: NVIDIA H200
GPU 6: NVIDIA H200
GPU 7: NVIDIA H200
Nvidia driver version : 550.144.03
cuDNN version : Probably one of the following:
/usr/lib/x86_64-linux-gnu/libcudnn.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_adv.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_cnn.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_precompiled.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_engines_runtime_compiled.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_graph.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_heuristic.so.9.8.0
/usr/lib/x86_64-linux-gnu/libcudnn_ops.so.9.8.0
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: 43 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(R) Xeon(R) Platinum 8468
CPU family: 6
Model: 143
Thread(s) per core: 2
Core(s) per socket: 32
Socket(s): 2
Stepping: 8
BogoMIPS: 4200.00
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid tsc_known_freq pni pclmulqdq dtes64 ssse3 fma cx16 pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch cpuid_fault invpcid_single ssbd ibrs ibpb stibp ibrs_enhanced fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves avx_vnni avx512_bf16 wbnoinvd arat avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b fsrm md_clear serialize tsxldtrk amx_bf16 avx512_fp16 amx_tile amx_int8 arch_capabilities
Hypervisor vendor: KVM
Virtualization type: full
L1d cache: 4 MiB (128 instances)
L1i cache: 4 MiB (128 instances)
L2 cache: 256 MiB (64 instances)
L3 cache: 32 MiB (2 instances)
NUMA node(s): 2
NUMA node0 CPU(s): 0-63
NUMA node1 CPU(s): 64-127
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: Unknown: No mitigations
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Not affected
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
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; BHI SW loop, KVM SW loop
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Mitigation; TSX disabled
==============================
Versions of relevant libraries
==============================
[pip3] numpy==1.26.4
[pip3] nvidia-cublas-cu12==12.8.3.14
[pip3] nvidia-cuda-cupti-cu12==12.8.57
[pip3] nvidia-cuda-nvrtc-cu12==12.8.61
[pip3] nvidia-cuda-runtime-cu12==12.8.57
[pip3] nvidia-cudnn-cu12==9.7.1.26
[pip3] nvidia-cufft-cu12==11.3.3.41
[pip3] nvidia-cufile-cu12==1.13.0.11
[pip3] nvidia-curand-cu12==10.3.9.55
[pip3] nvidia-cusolver-cu12==11.7.2.55
[pip3] nvidia-cusparse-cu12==12.5.7.53
[pip3] nvidia-cusparselt-cu12==0.6.3
[pip3] nvidia-nccl-cu12==2.26.2
[pip3] nvidia-nvjitlink-cu12==12.8.61
[pip3] nvidia-nvtx-cu12==12.8.55
[pip3] pyzmq==26.0.3
[pip3] torch==2.7.1+cu128
[pip3] torchaudio==2.7.1+cu128
[pip3] torchvision==0.22.1+cu128
[pip3] transformers==4.53.2
[pip3] triton==3.3.1
[conda] numpy 1.26.4 pypi_0 pypi
[conda] nvidia-cublas-cu12 12.8.3.14 pypi_0 pypi
[conda] nvidia-cuda-cupti-cu12 12.8.57 pypi_0 pypi
[conda] nvidia-cuda-nvrtc-cu12 12.8.61 pypi_0 pypi
[conda] nvidia-cuda-runtime-cu12 12.8.57 pypi_0 pypi
[conda] nvidia-cudnn-cu12 9.7.1.26 pypi_0 pypi
[conda] nvidia-cufft-cu12 11.3.3.41 pypi_0 pypi
[conda] nvidia-cufile-cu12 1.13.0.11 pypi_0 pypi
[conda] nvidia-curand-cu12 10.3.9.55 pypi_0 pypi
[conda] nvidia-cusolver-cu12 11.7.2.55 pypi_0 pypi
[conda] nvidia-cusparse-cu12 12.5.7.53 pypi_0 pypi
[conda] nvidia-cusparselt-cu12 0.6.3 pypi_0 pypi
[conda] nvidia-nccl-cu12 2.26.2 pypi_0 pypi
[conda] nvidia-nvjitlink-cu12 12.8.61 pypi_0 pypi
[conda] nvidia-nvtx-cu12 12.8.55 pypi_0 pypi
[conda] pyzmq 26.0.3 pypi_0 pypi
[conda] torch 2.7.1+cu128 pypi_0 pypi
[conda] torchaudio 2.7.1+cu128 pypi_0 pypi
[conda] torchvision 0.22.1+cu128 pypi_0 pypi
[conda] transformers 4.53.2 pypi_0 pypi
[conda] triton 3.3.1 pypi_0 pypi
==============================
vLLM Info
==============================
ROCM Version : Could not collect
Neuron SDK Version : N/A
vLLM Version : 0.10.1.dev383+g302962e80 (git sha: 302962e80)
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0 GPU1 GPU2 GPU3 GPU4 GPU5 GPU6 GPU7 NIC0 NIC1 NIC2 NIC3 NIC4 NIC5 NIC6 NIC7 CPU Affinity NUMA Affinity GPU NUMA ID
GPU0 X NV18 NV18 NV18 NV18 NV18 NV18 NV18 SYS SYS SYS SYS PIX PHB PHB PHB 0-63 0 N/A
GPU1 NV18 X NV18 NV18 NV18 NV18 NV18 NV18 SYS SYS SYS SYS PHB PIX PHB PHB 0-63 0 N/A
GPU2 NV18 NV18 X NV18 NV18 NV18 NV18 NV18 SYS SYS SYS SYS PHB PHB PIX PHB 0-63 0 N/A
GPU3 NV18 NV18 NV18 X NV18 NV18 NV18 NV18 SYS SYS SYS SYS PHB PHB PHB PIX 0-63 0 N/A
GPU4 NV18 NV18 NV18 NV18 X NV18 NV18 NV18 PIX PHB PHB PHB SYS SYS SYS SYS 64-127 1 N/A
GPU5 NV18 NV18 NV18 NV18 NV18 X NV18 NV18 PHB PIX PHB PHB SYS SYS SYS SYS 64-127 1 N/A
GPU6 NV18 NV18 NV18 NV18 NV18 NV18 X NV18 PHB PHB PIX PHB SYS SYS SYS SYS 64-127 1 N/A
GPU7 NV18 NV18 NV18 NV18 NV18 NV18 NV18 X PHB PHB PHB PIX SYS SYS SYS SYS 64-127 1 N/A
NIC0 SYS SYS SYS SYS PIX PHB PHB PHB X PHB PHB PHB SYS SYS SYS SYS
NIC1 SYS SYS SYS SYS PHB PIX PHB PHB PHB X PHB PHB SYS SYS SYS SYS
NIC2 SYS SYS SYS SYS PHB PHB PIX PHB PHB PHB X PHB SYS SYS SYS SYS
NIC3 SYS SYS SYS SYS PHB PHB PHB PIX PHB PHB PHB X SYS SYS SYS SYS
NIC4 PIX PHB PHB PHB SYS SYS SYS SYS SYS SYS SYS SYS X PHB PHB PHB
NIC5 PHB PIX PHB PHB SYS SYS SYS SYS SYS SYS SYS SYS PHB X PHB PHB
NIC6 PHB PHB PIX PHB SYS SYS SYS SYS SYS SYS SYS SYS PHB PHB X PHB
NIC7 PHB PHB PHB PIX SYS SYS SYS SYS SYS SYS SYS SYS PHB PHB PHB 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
NIC2: mlx5_2
NIC3: mlx5_3
NIC4: mlx5_4
NIC5: mlx5_5
NIC6: mlx5_6
NIC7: mlx5_7
==============================
Environment Variables
==============================
NVIDIA_VISIBLE_DEVICES=GPU-f74dac9a-eadb-edbc-bea5-2f5357148056,GPU-d79f9bd9-d555-ae25-e080-9719708693be,GPU-ce1a007d-9c4a-3d33-0e3c-ff58c002714e,GPU-4f2c67d4-be0f-afcd-97c2-a5ae211919ae,GPU-871b06be-51ee-dfd1-2c9f-f15bef6f6612,GPU-6e1afb47-0032-bebb-d4f1-8acfbf5d6089,GPU-c4684fa1-73a8-1013-97bb-af2917b98fec,GPU-dcfe463c-3b25-dbe2-3f70-636b9ad77ef4
NVIDIA_REQUIRE_CUDA=cuda>=12.8 brand=unknown,driver>=470,driver<471 brand=grid,driver>=470,driver<471 brand=tesla,driver>=470,driver<471 brand=nvidia,driver>=470,driver<471 brand=quadro,driver>=470,driver<471 brand=quadrortx,driver>=470,driver<471 brand=nvidiartx,driver>=470,driver<471 brand=vapps,driver>=470,driver<471 brand=vpc,driver>=470,driver<471 brand=vcs,driver>=470,driver<471 brand=vws,driver>=470,driver<471 brand=cloudgaming,driver>=470,driver<471 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<536 brand=nvidia,driver>=535,driver<536 brand=quadro,driver>=535,driver<536 brand=quadrortx,driver>=535,driver<536 brand=nvidiartx,driver>=535,driver<536 brand=vapps,driver>=535,driver<536 brand=vpc,driver>=535,driver<536 brand=vcs,driver>=535,driver<536 brand=vws,driver>=535,driver<536 brand=cloudgaming,driver>=535,driver<536 brand=unknown,driver>=550,driver<551 brand=grid,driver>=550,driver<551 brand=tesla,driver>=550,driver<551 brand=nvidia,driver>=550,driver<551 brand=quadro,driver>=550,driver<551 brand=quadrortx,driver>=550,driver<551 brand=nvidiartx,driver>=550,driver<551 brand=vapps,driver>=550,driver<551 brand=vpc,driver>=550,driver<551 brand=vcs,driver>=550,driver<551 brand=vws,driver>=550,driver<551 brand=cloudgaming,driver>=550,driver<551 brand=unknown,driver>=560,driver<561 brand=grid,driver>=560,driver<561 brand=tesla,driver>=560,driver<561 brand=nvidia,driver>=560,driver<561 brand=quadro,driver>=560,driver<561 brand=quadrortx,driver>=560,driver<561 brand=nvidiartx,driver>=560,driver<561 brand=vapps,driver>=560,driver<561 brand=vpc,driver>=560,driver<561 brand=vcs,driver>=560,driver<561 brand=vws,driver>=560,driver<561 brand=cloudgaming,driver>=560,driver<561 brand=unknown,driver>=565,driver<566 brand=grid,driver>=565,driver<566 brand=tesla,driver>=565,driver<566 brand=nvidia,driver>=565,driver<566 brand=quadro,driver>=565,driver<566 brand=quadrortx,driver>=565,driver<566 brand=nvidiartx,driver>=565,driver<566 brand=vapps,driver>=565,driver<566 brand=vpc,driver>=565,driver<566 brand=vcs,driver>=565,driver<566 brand=vws,driver>=565,driver<566 brand=cloudgaming,driver>=565,driver<566
NCCL_VERSION=2.25.1-1
NCCL_SOCKET_IFNAME=^lo,docker,veth,tailscale,anyscale
NVIDIA_DRIVER_CAPABILITIES=compute,utility
NVIDIA_PRODUCT_NAME=CUDA
NVIDIA_MOFED=enabled
CUDA_VERSION=12.8.1
LD_LIBRARY_PATH=/usr/local/nvshmem/lib:/usr/local/cuda/lib64:/usr/local/ucx/lib:/usr/local/nixl/lib/x86_64-linux-gnu:/usr/local/nvidia/lib64
NCCL_CUMEM_ENABLE=0
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
CUDA_MODULE_LOADING=LAZY
🐛 Describe the bug
VLLM_ALL2ALL_BACKEND=naive (the default) hangs/crashes on multi nodes when serving DeepSeekV3
We are aware that PPLX and DeepEP kernels can be used. However, these kernels cannot be used if IB is not available for multinode.
To reproduce:
# Node 0 (with ip address 10.96.94.252)
vllm serve /mnt/.cache/huggingface/hub/models--deepseek-ai--DeepSeek-V3-0324/snapshots/e9b33add76883f293d6bf61f6bd89b497e80e335 --data-parallel-size 16 --data-parallel-size-local 8 --data-parallel-address 10.96.94.252 --data-parallel-rpc-port 13345 --enable-expert-parallel
# Node 1
vllm serve /mnt/.cache/huggingface/hub/models--deepseek-ai--DeepSeek-V3-0324/snapshots/e9b33add76883f293d6bf61f6bd89b497e80e335 --headless --data-parallel-size 16 --data-parallel-size-local 8 --data-parallel-start-rank 8 --data-parallel-address 10.96.94.252 --data-parallel-rpc-port 13345 --enable-expert-parallel
Execution hangs and the node dies (possibly due to OOM, see analysis with ray below) during CUDA graph capture. And the following errors are logged:
Error logs
[rank0]:[E822 16:31:16.131931664 ProcessGroupNCCL.cpp:632] [Rank 0] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600009 milliseconds before timing out.
[rank0]:[E822 16:31:16.132547721 ProcessGroupNCCL.cpp:2271] [PG ID 4 PG GUID 67 Rank 0] failure detected by watchdog at work sequence id: 5730 PG status: last enqueued work: 6464, last completed work: 5729
[rank0]:[E822 16:31:16.132564125 ProcessGroupNCCL.cpp:670] Stack trace of the failed collective not found, potentially because FlightRecorder is disabled. You can enable it by setting TORCH_NCCL_TRACE_BUFFER_SIZE to a non-zero value.
[rank0]:[E822 16:31:16.132605556 ProcessGroupNCCL.cpp:2106] [PG ID 4 PG GUID 67 Rank 0] First PG on this rank to signal dumping.
[rank3]:[E822 16:31:16.135435629 ProcessGroupNCCL.cpp:632] [Rank 3] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600008 milliseconds before timing out.
[rank3]:[E822 16:31:16.137270546 ProcessGroupNCCL.cpp:2271] [PG ID 4 PG GUID 67 Rank 3] failure detected by watchdog at work sequence id: 5729 PG status: last enqueued work: 6464, last completed work: 5728
[rank3]:[E822 16:31:16.137312219 ProcessGroupNCCL.cpp:670] Stack trace of the failed collective not found, potentially because FlightRecorder is disabled. You can enable it by setting TORCH_NCCL_TRACE_BUFFER_SIZE to a non-zero value.
[rank3]:[E822 16:31:16.137390694 ProcessGroupNCCL.cpp:2106] [PG ID 4 PG GUID 67 Rank 3] First PG on this rank to signal dumping.
[rank2]:[E822 16:31:16.183580558 ProcessGroupNCCL.cpp:632] [Rank 2] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600062 milliseconds before timing out.
[rank2]:[E822 16:31:16.184825116 ProcessGroupNCCL.cpp:2271] [PG ID 4 PG GUID 67 Rank 2] failure detected by watchdog at work sequence id: 5729 PG status: last enqueued work: 6464, last completed work: 5728
[rank2]:[E822 16:31:16.184844565 ProcessGroupNCCL.cpp:670] Stack trace of the failed collective not found, potentially because FlightRecorder is disabled. You can enable it by setting TORCH_NCCL_TRACE_BUFFER_SIZE to a non-zero value.
[rank2]:[E822 16:31:16.184890195 ProcessGroupNCCL.cpp:2106] [PG ID 4 PG GUID 67 Rank 2] First PG on this rank to signal dumping.
[rank6]:[E822 16:31:16.186150403 ProcessGroupNCCL.cpp:632] [Rank 6] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600059 milliseconds before timing out.
[rank6]:[E822 16:31:16.186671749 ProcessGroupNCCL.cpp:2271] [PG ID 4 PG GUID 67 Rank 6] failure detected by watchdog at work sequence id: 5730 PG status: last enqueued work: 6464, last completed work: 5729
...
[rank4]:[E822 16:31:16.294173405 ProcessGroupNCCL.cpp:1746] [PG ID 0 PG GUID 0 Rank 4] Received a dump signal due to a collective timeout from this local rank and we will try our best to dump the debug info. Last enqueued NCCL work: -1, last completed NCCL work: -1.This is most likely caused by incorrect usages of collectives, e.g., wrong sizes used across ranks, the order of collectives is not same for all ranks or the scheduled collective, for some reason, didn't run. Additionally, this can be caused by GIL deadlock or other reasons such as network errors or bugs in the communications library (e.g. NCCL), etc.
[rank4]:[E822 16:31:16.294344990 ProcessGroupNCCL.cpp:1536] [PG ID 0 PG GUID 0 Rank 4] ProcessGroupNCCL preparing to dump debug info. Include stack trace: 1
[rank6]:[E822 16:31:16.352075294 ProcessGroupNCCL.cpp:1746] [PG ID 0 PG GUID 0 Rank 6] Received a dump signal due to a collective timeout from this local rank and we will try our best to dump the debug info. Last enqueued NCCL work: -1, last completed NCCL work: -1.This is most likely caused by incorrect usages of collectives, e.g., wrong sizes used across ranks, the order of collectives is not same for all ranks or the scheduled collective, for some reason, didn't run. Additionally, this can be caused by GIL deadlock or other reasons such as network errors or bugs in the communications library (e.g. NCCL), etc.
[rank7]:[E822 16:31:16.352086293 ProcessGroupNCCL.cpp:1746] [PG ID 0 PG GUID 0 Rank 7] Received a dump signal due to a collective timeout from this local rank and we will try our best to dump the debug info. Last enqueued NCCL work: -1, last completed NCCL work: -1.This is most likely caused by incorrect usages of collectives, e.g., wrong sizes used across ranks, the order of collectives is not same for all ranks or the scheduled collective, for some reason, didn't run. Additionally, this can be caused by GIL deadlock or other reasons such as network errors or bugs in the communications library (e.g. NCCL), etc.
[rank6]:[E822 16:31:16.352234789 ProcessGroupNCCL.cpp:1536] [PG ID 0 PG GUID 0 Rank 6] ProcessGroupNCCL preparing to dump debug info. Include stack trace: 1
[rank7]:[E822 16:31:16.352287277 ProcessGroupNCCL.cpp:1536] [PG ID 0 PG GUID 0 Rank 7] ProcessGroupNCCL preparing to dump debug info. Include stack trace: 1
[rank5]:[E822 16:31:16.359866596 ProcessGroupNCCL.cpp:1746] [PG ID 0 PG GUID 0 Rank 5] Received a dump signal due to a collective timeout from this local rank and we will try our best to dump the debug info. Last enqueued NCCL work: -1, last completed NCCL work: -1.This is most likely caused by incorrect usages of collectives, e.g., wrong sizes used across ranks, the order of collectives is not same for all ranks or the scheduled collective, for some reason, didn't run. Additionally, this can be caused by GIL deadlock or other reasons such as network errors or bugs in the communications library (e.g. NCCL), etc.
[rank5]:[E822 16:31:16.360106993 ProcessGroupNCCL.cpp:1536] [PG ID 0 PG GUID 0 Rank 5] ProcessGroupNCCL preparing to dump debug info. Include stack trace: 1
[rank1]:[E822 16:31:16.378507092 ProcessGroupNCCL.cpp:1746] [PG ID 0 PG GUID 0 Rank 1] Received a dump signal due to a collective timeout from this local rank and we will try our best to dump the debug info. Last enqueued NCCL work: -1, last completed NCCL work: -1.This is most likely caused by incorrect usages of collectives, e.g., wrong sizes used across ranks, the order of collectives is not same for all ranks or the scheduled collective, for some reason, didn't run. Additionally, this can be caused by GIL deadlock or other reasons such as network errors or bugs in the communications library (e.g. NCCL), etc.
[rank1]:[E822 16:31:16.378667092 ProcessGroupNCCL.cpp:1536] [PG ID 0 PG GUID 0 Rank 1] ProcessGroupNCCL preparing to dump debug info. Include stack trace: 1
[rank0]:[E822 16:31:17.943532381 ProcessGroupNCCL.cpp:1746] [PG ID 0 PG GUID 0 Rank 0] Received a dump signal due to a collective timeout from this local rank and we will try our best to dump the debug info. Last enqueued NCCL work: -1, last completed NCCL work: -1.This is most likely caused by incorrect usages of collectives, e.g., wrong sizes used across ranks, the order of collectives is not same for all ranks or the scheduled collective, for some reason, didn't run. Additionally, this can be caused by GIL deadlock or other reasons such as network errors or bugs in the communications library (e.g. NCCL), etc.
[rank0]:[E822 16:31:17.943716558 ProcessGroupNCCL.cpp:1536] [PG ID 0 PG GUID 0 Rank 0] ProcessGroupNCCL preparing to dump debug info. Include stack trace: 1
[rank0]:[W822 16:32:16.133271453 ProcessGroupNCCL.cpp:2114] [PG ID 4 PG GUID 67 Rank 0] timed out after waiting for 60000ms flight recorder dumps to finish.
[rank0]:[E822 16:32:16.133405860 ProcessGroupNCCL.cpp:684] [Rank 0] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data.
[rank0]:[E822 16:32:16.133441408 ProcessGroupNCCL.cpp:698] [Rank 0] To avoid data inconsistency, we are taking the entire process down.
[rank0]:[E822 16:32:16.134535739 ProcessGroupNCCL.cpp:1899] [PG ID 4 PG GUID 67 Rank 0] Process group watchdog thread terminated with exception: [Rank 0] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600009 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f4e3d5785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f4dd3150b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f4dd31528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f4dd3153fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f4e3f02b198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f4e3f1d4ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f4e3f266850 in /lib/x86_64-linux-gnu/libc.so.6)
terminate called after throwing an instance of 'c10::DistBackendError'
what(): [PG ID 4 PG GUID 67 Rank 0] Process group watchdog thread terminated with exception: [Rank 0] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600009 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f4e3d5785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f4dd3150b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f4dd31528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f4dd3153fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f4e3f02b198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f4e3f1d4ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f4e3f266850 in /lib/x86_64-linux-gnu/libc.so.6)
Exception raised from ncclCommWatchdog at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1905 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f4e3d5785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: + 0xcc7b9e (0x7f4dd3122b9e in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: + 0x9165ed (0x7f4dd2d715ed in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: + 0xd8198 (0x7f4e3f02b198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #4: + 0x94ac3 (0x7f4e3f1d4ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #5: + 0x126850 (0x7f4e3f266850 in /lib/x86_64-linux-gnu/libc.so.6)
[rank3]:[W822 16:32:16.141492595 ProcessGroupNCCL.cpp:2114] [PG ID 4 PG GUID 67 Rank 3] timed out after waiting for 60000ms flight recorder dumps to finish.
[rank3]:[E822 16:32:16.141588378 ProcessGroupNCCL.cpp:684] [Rank 3] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data.
[rank3]:[E822 16:32:16.141616295 ProcessGroupNCCL.cpp:698] [Rank 3] To avoid data inconsistency, we are taking the entire process down.
[rank3]:[E822 16:32:16.142546155 ProcessGroupNCCL.cpp:1899] [PG ID 4 PG GUID 67 Rank 3] Process group watchdog thread terminated with exception: [Rank 3] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600008 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f06313785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f05c6f50b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f05c6f528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f05c6f53fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f0632e4d198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f0632ff6ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f0633088850 in /lib/x86_64-linux-gnu/libc.so.6)
terminate called after throwing an instance of 'c10::DistBackendError'
what(): [PG ID 4 PG GUID 67 Rank 3] Process group watchdog thread terminated with exception: [Rank 3] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600008 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f06313785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f05c6f50b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f05c6f528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f05c6f53fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f0632e4d198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f0632ff6ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f0633088850 in /lib/x86_64-linux-gnu/libc.so.6)
Exception raised from ncclCommWatchdog at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1905 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f06313785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: + 0xcc7b9e (0x7f05c6f22b9e in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: + 0x9165ed (0x7f05c6b715ed in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: + 0xd8198 (0x7f0632e4d198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #4: + 0x94ac3 (0x7f0632ff6ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #5: + 0x126850 (0x7f0633088850 in /lib/x86_64-linux-gnu/libc.so.6)
[rank2]:[W822 16:32:16.185054825 ProcessGroupNCCL.cpp:2114] [PG ID 4 PG GUID 67 Rank 2] timed out after waiting for 60000ms flight recorder dumps to finish.
[rank2]:[E822 16:32:16.185150063 ProcessGroupNCCL.cpp:684] [Rank 2] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data.
[rank2]:[E822 16:32:16.185180420 ProcessGroupNCCL.cpp:698] [Rank 2] To avoid data inconsistency, we are taking the entire process down.
[rank2]:[E822 16:32:16.186239608 ProcessGroupNCCL.cpp:1899] [PG ID 4 PG GUID 67 Rank 2] Process group watchdog thread terminated with exception: [Rank 2] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600062 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f4ef0f785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f4e86b50b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f4e86b528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f4e86b53fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f4ef2a92198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f4ef2c3bac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f4ef2ccd850 in /lib/x86_64-linux-gnu/libc.so.6)
terminate called after throwing an instance of 'c10::DistBackendError'
what(): [PG ID 4 PG GUID 67 Rank 2] Process group watchdog thread terminated with exception: [Rank 2] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600062 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f4ef0f785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f4e86b50b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f4e86b528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f4e86b53fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f4ef2a92198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f4ef2c3bac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f4ef2ccd850 in /lib/x86_64-linux-gnu/libc.so.6)
Exception raised from ncclCommWatchdog at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1905 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f4ef0f785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: + 0xcc7b9e (0x7f4e86b22b9e in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: + 0x9165ed (0x7f4e867715ed in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: + 0xd8198 (0x7f4ef2a92198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #4: + 0x94ac3 (0x7f4ef2c3bac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #5: + 0x126850 (0x7f4ef2ccd850 in /lib/x86_64-linux-gnu/libc.so.6)
[rank6]:[W822 16:32:16.186838621 ProcessGroupNCCL.cpp:2114] [PG ID 4 PG GUID 67 Rank 6] timed out after waiting for 60000ms flight recorder dumps to finish.
[rank6]:[E822 16:32:16.186918796 ProcessGroupNCCL.cpp:684] [Rank 6] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data.
[rank6]:[E822 16:32:16.186939495 ProcessGroupNCCL.cpp:698] [Rank 6] To avoid data inconsistency, we are taking the entire process down.
[rank6]:[E822 16:32:16.187872149 ProcessGroupNCCL.cpp:1899] [PG ID 4 PG GUID 67 Rank 6] Process group watchdog thread terminated with exception: [Rank 6] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600059 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7fe2189785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7fe1ae550b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7fe1ae5528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7fe1ae553fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7fe21a4ac198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7fe21a655ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7fe21a6e7850 in /lib/x86_64-linux-gnu/libc.so.6)
terminate called after throwing an instance of 'c10::DistBackendError'
what(): [PG ID 4 PG GUID 67 Rank 6] Process group watchdog thread terminated with exception: [Rank 6] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600059 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7fe2189785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7fe1ae550b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7fe1ae5528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7fe1ae553fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7fe21a4ac198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7fe21a655ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7fe21a6e7850 in /lib/x86_64-linux-gnu/libc.so.6)
Exception raised from ncclCommWatchdog at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1905 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7fe2189785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: + 0xcc7b9e (0x7fe1ae522b9e in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: + 0x9165ed (0x7fe1ae1715ed in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: + 0xd8198 (0x7fe21a4ac198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #4: + 0x94ac3 (0x7fe21a655ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #5: + 0x126850 (0x7fe21a6e7850 in /lib/x86_64-linux-gnu/libc.so.6)
[rank1]:[W822 16:32:16.194225886 ProcessGroupNCCL.cpp:2114] [PG ID 4 PG GUID 67 Rank 1] timed out after waiting for 60000ms flight recorder dumps to finish.
[rank1]:[E822 16:32:16.194306669 ProcessGroupNCCL.cpp:684] [Rank 1] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data.
[rank1]:[E822 16:32:16.194326563 ProcessGroupNCCL.cpp:698] [Rank 1] To avoid data inconsistency, we are taking the entire process down.
[rank1]:[E822 16:32:16.195210048 ProcessGroupNCCL.cpp:1899] [PG ID 4 PG GUID 67 Rank 1] Process group watchdog thread terminated with exception: [Rank 1] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600073 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f5b9ed785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f5b34950b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f5b349528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f5b34953fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f5ba0862198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f5ba0a0bac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f5ba0a9d850 in /lib/x86_64-linux-gnu/libc.so.6)
terminate called after throwing an instance of 'c10::DistBackendError'
what(): [PG ID 4 PG GUID 67 Rank 1] Process group watchdog thread terminated with exception: [Rank 1] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600073 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f5b9ed785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f5b34950b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f5b349528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f5b34953fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f5ba0862198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f5ba0a0bac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f5ba0a9d850 in /lib/x86_64-linux-gnu/libc.so.6)
Exception raised from ncclCommWatchdog at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1905 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f5b9ed785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: + 0xcc7b9e (0x7f5b34922b9e in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: + 0x9165ed (0x7f5b345715ed in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: + 0xd8198 (0x7f5ba0862198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #4: + 0x94ac3 (0x7f5ba0a0bac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #5: + 0x126850 (0x7f5ba0a9d850 in /lib/x86_64-linux-gnu/libc.so.6)
[rank7]:[W822 16:32:16.202389755 ProcessGroupNCCL.cpp:2114] [PG ID 4 PG GUID 67 Rank 7] timed out after waiting for 60000ms flight recorder dumps to finish.
[rank7]:[E822 16:32:16.202485674 ProcessGroupNCCL.cpp:684] [Rank 7] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data.
[rank7]:[E822 16:32:16.202508468 ProcessGroupNCCL.cpp:698] [Rank 7] To avoid data inconsistency, we are taking the entire process down.
[rank7]:[E822 16:32:16.203612683 ProcessGroupNCCL.cpp:1899] [PG ID 4 PG GUID 67 Rank 7] Process group watchdog thread terminated with exception: [Rank 7] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600080 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f21c33d35e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f21c4150b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f21c41528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f21c4153fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f2230002198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f22301abac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f223023d850 in /lib/x86_64-linux-gnu/libc.so.6)
terminate called after throwing an instance of 'c10::DistBackendError'
what(): [PG ID 4 PG GUID 67 Rank 7] Process group watchdog thread terminated with exception: [Rank 7] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600080 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f21c33d35e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f21c4150b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f21c41528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f21c4153fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f2230002198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f22301abac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f223023d850 in /lib/x86_64-linux-gnu/libc.so.6)
Exception raised from ncclCommWatchdog at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1905 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f21c33d35e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: + 0xcc7b9e (0x7f21c4122b9e in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: + 0x9165ed (0x7f21c3d715ed in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: + 0xd8198 (0x7f2230002198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #4: + 0x94ac3 (0x7f22301abac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #5: + 0x126850 (0x7f223023d850 in /lib/x86_64-linux-gnu/libc.so.6)
[rank4]:[W822 16:32:16.211333174 ProcessGroupNCCL.cpp:2114] [PG ID 4 PG GUID 67 Rank 4] timed out after waiting for 60000ms flight recorder dumps to finish.
[rank4]:[E822 16:32:16.211410920 ProcessGroupNCCL.cpp:684] [Rank 4] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data.
[rank4]:[E822 16:32:16.211431467 ProcessGroupNCCL.cpp:698] [Rank 4] To avoid data inconsistency, we are taking the entire process down.
[rank4]:[E822 16:32:16.212335158 ProcessGroupNCCL.cpp:1899] [PG ID 4 PG GUID 67 Rank 4] Process group watchdog thread terminated with exception: [Rank 4] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600086 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7eff44d785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7efeda950b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7efeda9528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7efeda953fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7eff46900198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7eff46aa9ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7eff46b3b850 in /lib/x86_64-linux-gnu/libc.so.6)
terminate called after throwing an instance of 'c10::DistBackendError'
what(): [PG ID 4 PG GUID 67 Rank 4] Process group watchdog thread terminated with exception: [Rank 4] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5729, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600086 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7eff44d785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7efeda950b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7efeda9528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7efeda953fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7eff46900198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7eff46aa9ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7eff46b3b850 in /lib/x86_64-linux-gnu/libc.so.6)
Exception raised from ncclCommWatchdog at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1905 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7eff44d785e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: + 0xcc7b9e (0x7efeda922b9e in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: + 0x9165ed (0x7efeda5715ed in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: + 0xd8198 (0x7eff46900198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #4: + 0x94ac3 (0x7eff46aa9ac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #5: + 0x126850 (0x7eff46b3b850 in /lib/x86_64-linux-gnu/libc.so.6)
[rank5]:[W822 16:32:16.225133377 ProcessGroupNCCL.cpp:2114] [PG ID 4 PG GUID 67 Rank 5] timed out after waiting for 60000ms flight recorder dumps to finish.
[rank5]:[E822 16:32:16.225240851 ProcessGroupNCCL.cpp:684] [Rank 5] Some NCCL operations have failed or timed out. Due to the asynchronous nature of CUDA kernels, subsequent GPU operations might run on corrupted/incomplete data.
[rank5]:[E822 16:32:16.225279655 ProcessGroupNCCL.cpp:698] [Rank 5] To avoid data inconsistency, we are taking the entire process down.
[rank5]:[E822 16:32:16.226571535 ProcessGroupNCCL.cpp:1899] [PG ID 4 PG GUID 67 Rank 5] Process group watchdog thread terminated with exception: [Rank 5] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600099 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f9073bd35e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f9074950b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f90749528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f9074953fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f90e0786198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f90e092fac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f90e09c1850 in /lib/x86_64-linux-gnu/libc.so.6)
terminate called after throwing an instance of 'c10::DistBackendError'
what(): [PG ID 4 PG GUID 67 Rank 5] Process group watchdog thread terminated with exception: [Rank 5] Watchdog caught collective operation timeout: WorkNCCL(SeqNum=5730, OpType=BROADCAST, NumelIn=3612672, NumelOut=3612672, Timeout(ms)=600000) ran for 600099 milliseconds before timing out.
Exception raised from checkTimeout at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:635 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f9073bd35e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: c10d::ProcessGroupNCCL::WorkNCCL::checkTimeout(std::optional<std::chrono::duration<long, std::ratio<1l, 1000l> > >) + 0x23d (0x7f9074950b4d in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: c10d::ProcessGroupNCCL::watchdogHandler() + 0xc80 (0x7f90749528d0 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x14d (0x7f9074953fdd in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #4: + 0xd8198 (0x7f90e0786198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #5: + 0x94ac3 (0x7f90e092fac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #6: + 0x126850 (0x7f90e09c1850 in /lib/x86_64-linux-gnu/libc.so.6)
Exception raised from ncclCommWatchdog at /pytorch/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1905 (most recent call first):
frame #0: c10::Error::Error(c10::SourceLocation, std::__cxx11::basic_string<char, std::char_traits, std::allocator >) + 0x98 (0x7f9073bd35e8 in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libc10.so)
frame #1: + 0xcc7b9e (0x7f9074922b9e in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #2: + 0x9165ed (0x7f90745715ed in /home/ray/anaconda3/lib/python3.11/site-packages/torch/lib/libtorch_cuda.so)
frame #3: + 0xd8198 (0x7f90e0786198 in /home/ray/anaconda3/bin/../lib/libstdc++.so.6)
frame #4: + 0x94ac3 (0x7f90e092fac3 in /lib/x86_64-linux-gnu/libc.so.6)
frame #5: + 0x126850 (0x7f90e09c1850 in /lib/x86_64-linux-gnu/libc.so.6)
If testing with --distributed-executor-backend ray, hanging would also cause CPU memory OOM, and ray OOM killer would detect it and raise OOM error.
Memory profiling indicates NCCL network proxy accumulated a large amount of memory.
Before submitting a new issue...
- 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.