OSError: [Errno 9] Bad file descriptor: '.../python3.10/site-packages/torch/nested'

[josephcappadona]

🐛 Describe the bug

Traceback (most recent call last):
  File "/home/joe/micromamba/lib/python3.10/multiprocessing/queues.py", line 244, in _feed
    obj = _ForkingPickler.dumps(obj)
  File "/home/joe/micromamba/lib/python3.10/multiprocessing/reduction.py", line 51, in dumps
    cls(buf, protocol).dump(obj)
  File "/home/joe/micromamba/lib/python3.10/site-packages/torch/multiprocessing/reductions.py", line 296, in reduce_tensor
    from torch.nested._internal.nested_tensor import NestedTensor
  File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1002, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 945, in _find_spec
  File "<frozen importlib._bootstrap_external>", line 1439, in find_spec
  File "<frozen importlib._bootstrap_external>", line 1411, in _get_spec
  File "<frozen importlib._bootstrap_external>", line 1548, in find_spec
  File "<frozen importlib._bootstrap_external>", line 1591, in _fill_cache
OSError: [Errno 9] Bad file descriptor: '/home/joe/micromamba/lib/python3.10/site-packages/torch/nested'
Traceback (most recent call last):
  File "/home/joe/micromamba/lib/python3.10/multiprocessing/queues.py", line 244, in _feed
    obj = _ForkingPickler.dumps(obj)
  File "/home/joe/micromamba/lib/python3.10/multiprocessing/reduction.py", line 51, in dumps
    cls(buf, protocol).dump(obj)
  File "/home/joe/micromamba/lib/python3.10/site-packages/torch/multiprocessing/reductions.py", line 296, in reduce_tensor
    from torch.nested._internal.nested_tensor import NestedTensor
  File "<frozen importlib._bootstrap>", line 1027, in _find_and_load
  File "<frozen importlib._bootstrap>", line 1002, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 945, in _find_spec
  File "<frozen importlib._bootstrap_external>", line 1439, in find_spec
  File "<frozen importlib._bootstrap_external>", line 1411, in _get_spec
  File "<frozen importlib._bootstrap_external>", line 1548, in find_spec
  File "<frozen importlib._bootstrap_external>", line 1591, in _fill_cache
OSError: [Errno 9] Bad file descriptor: '/home/joe/micromamba/lib/python3.10/site-packages/torch/nested'

I'm not sure what is happening here. I think it is happening around when a dataloader is being instantiated. Any help would be appreciated. Very strangely, this doesn't happen on two other machines with (as far as I can tell) identical environments.

Versions

Collecting environment information...
PyTorch version: 2.4.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.4 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: Could not collect
CMake version: Could not collect
Libc version: glibc-2.35

Python version: 3.10.14 | packaged by conda-forge | (main, Mar 20 2024, 12:45:18) [GCC 12.3.0] (64-bit runtime)
Python platform: Linux-5.15.0-94-generic-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 H100 80GB HBM3
GPU 1: NVIDIA H100 80GB HBM3
GPU 2: NVIDIA H100 80GB HBM3
GPU 3: NVIDIA H100 80GB HBM3
GPU 4: NVIDIA H100 80GB HBM3
GPU 5: NVIDIA H100 80GB HBM3
GPU 6: NVIDIA H100 80GB HBM3
GPU 7: NVIDIA H100 80GB HBM3

Nvidia driver version: 550.90.07
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:                      46 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) PLATINUM 8558
CPU family:                         6
Model:                              207
Thread(s) per core:                 2
Core(s) per socket:                 48
Socket(s):                          2
Stepping:                           2
Frequency boost:                    enabled
CPU max MHz:                        2101.0000
CPU min MHz:                        800.0000
BogoMIPS:                           4200.00
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 invpcid_single intel_ppin cdp_l2 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 avx_vnni avx512_bf16 wbnoinvd dtherm ida arat pln pts avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg tme avx512_vpopcntdq la57 rdpid bus_lock_detect cldemote movdiri movdir64b enqcmd fsrm md_clear serialize tsxldtrk pconfig arch_lbr amx_bf16 avx512_fp16 amx_tile amx_int8 flush_l1d arch_capabilities
Virtualization:                     VT-x
L1d cache:                          4.5 MiB (96 instances)
L1i cache:                          3 MiB (96 instances)
L2 cache:                           192 MiB (96 instances)
L3 cache:                           520 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 Itlb multihit:        Not affected
Vulnerability L1tf:                 Not affected
Vulnerability Mds:                  Not affected
Vulnerability Meltdown:             Not affected
Vulnerability Mmio stale data:      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 IBRS, IBPB conditional, RSB filling, PBRSB-eIBRS SW sequence
Vulnerability Srbds:                Not affected
Vulnerability Tsx async abort:      Not affected

Versions of relevant libraries:
[pip3] botorch==0.11.3
[pip3] gpytorch==1.12
[pip3] numpy==1.23.5
[pip3] pytorchvideo==0.1.5
[pip3] torch==2.4.0
[pip3] torchmetrics==1.4.0.post0
[pip3] torchvision==0.19.0
[pip3] triton==3.0.0
[conda] Could not collect

[josephcappadona]

After commenting out torch.multiprocessing.set_sharing_strategy('file_system') which I had in my code, this is the traceback I'm now getting:

[rank2]: Traceback (most recent call last):
[rank2]:   File "/home/joe/micromamba/lib/python3.10/multiprocessing/resource_sharer.py", line 138, in _serve
[rank2]:     with self._listener.accept() as conn:
[rank2]:   File "/home/joe/micromamba/lib/python3.10/multiprocessing/connection.py", line 463, in accept
[rank2]:     c = self._listener.accept()
[rank2]:   File "/home/joe/micromamba/lib/python3.10/multiprocessing/connection.py", line 609, in accept
[rank2]:     s, self._last_accepted = self._socket.accept()
[rank2]:   File "/home/joe/micromamba/lib/python3.10/socket.py", line 293, in accept
[rank2]:     fd, addr = self._accept()
[rank2]: OSError: [Errno 9] Bad file descriptor

[josephcappadona]

The root cause seemed to be the image files that my Dataset was loading. I'm not sure what was wrong with them, as I was able to open all of the files independently using PIL and only when they were being loaded in a multi-worker setting did they cause this error. If anyone from PyTorch could fill me in on why this might have happened, that would be great -- but I will mark this as resolved.