(IMPORT FAILED): /mnt/wf/ComfyUI/custom_nodes/ComfyUI-SUPIR

[mike-rowley]

I am getting an Import Failed on loading the comfy server, I didnt find anything that appeared to match from current or past issues.

I have tried to fix Supir, uninstalled and reinstalled it, I reran pip install -r requirements.txt and updated comfy to the latest and it still fails to load. Any suggestions?

Here is the traceback:

ComfyUI Revision: 2169 [0fecfd2b] | Released on '2024-05-09'

Traceback (most recent call last): File "/mnt/wf/ComfyUI/nodes.py", line 1867, in load_custom_node module_spec.loader.exec_module(module) File "", line 883, in exec_module File "", line 241, in _call_with_frames_removed File "/mnt/wf/ComfyUI/custom_nodes/ComfyUI-SUPIR/init.py", line 1, in from .nodes import SUPIR_Upscale File "/mnt/wf/ComfyUI/custom_nodes/ComfyUI-SUPIR/nodes.py", line 11, in from .sgm.util import instantiate_from_config File "/mnt/wf/ComfyUI/custom_nodes/ComfyUI-SUPIR/sgm/init.py", line 1, in from .models import AutoencodingEngine, DiffusionEngine File "/mnt/wf/ComfyUI/custom_nodes/ComfyUI-SUPIR/sgm/models/init.py", line 1, in from .autoencoder import AutoencodingEngine File "/mnt/wf/ComfyUI/custom_nodes/ComfyUI-SUPIR/sgm/models/autoencoder.py", line 6, in import pytorch_lightning as pl File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/pytorch_lightning/init.py", line 27, in from pytorch_lightning.callbacks import Callback # noqa: E402 File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/pytorch_lightning/callbacks/init.py", line 14, in from pytorch_lightning.callbacks.batch_size_finder import BatchSizeFinder File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/pytorch_lightning/callbacks/batch_size_finder.py", line 26, in from pytorch_lightning.callbacks.callback import Callback File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/pytorch_lightning/callbacks/callback.py", line 22, in from pytorch_lightning.utilities.types import STEP_OUTPUT File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/pytorch_lightning/utilities/types.py", line 41, in from torchmetrics import Metric File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/init.py", line 22, in from torchmetrics import functional # noqa: E402 File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/functional/init.py", line 14, in from torchmetrics.functional.audio._deprecated import _permutation_invariant_training as permutation_invariant_training File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/functional/audio/init.py", line 14, in from torchmetrics.functional.audio.pit import permutation_invariant_training, pit_permutate File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/functional/audio/pit.py", line 22, in from torchmetrics.utilities import rank_zero_warn File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/utilities/init.py", line 14, in from torchmetrics.utilities.checks import check_forward_full_state_property File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/utilities/checks.py", line 25, in from torchmetrics.metric import Metric File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/metric.py", line 30, in from torchmetrics.utilities.data import ( File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/utilities/data.py", line 22, in from torchmetrics.utilities.imports import _TORCH_GREATER_EQUAL_1_12, _TORCH_GREATER_EQUAL_1_13, _XLA_AVAILABLE File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchmetrics/utilities/imports.py", line 51, in _TORCHAUDIO_GREATER_EQUAL_0_10: Optional[bool] = compare_version("torchaudio", operator.ge, "0.10.0") File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/lightning_utilities/core/imports.py", line 77, in compare_version pkg = importlib.import_module(package) File "/usr/lib/python3.10/importlib/init.py", line 126, in import_module return _bootstrap._gcd_import(name[level:], package, level) File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchaudio/init.py", line 1, in from . import ( # noqa: F401 File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchaudio/_extension/init.py", line 45, in _load_lib("libtorchaudio") File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchaudio/_extension/utils.py", line 64, in _load_lib torch.ops.load_library(path) File "/mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torch/_ops.py", line 1032, in load_library ctypes.CDLL(path) File "/usr/lib/python3.10/ctypes/init.py", line 374, in init self._handle = _dlopen(self._name, mode) OSError: /mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchaudio/lib/libtorchaudio.so: undefined symbol: _ZN2at4_ops10zeros_like4callERKNS_6TensorEN3c108optionalINS5_10ScalarTypeEEENS6_INS5_6LayoutEEENS6_INS5_6DeviceEEENS6_IbEENS6_INS5_12MemoryFormatEEE

Cannot import /mnt/wf/ComfyUI/custom_nodes/ComfyUI-SUPIR module for custom nodes: /mnt/wf/ComfyUI/venv/lib/python3.10/site-packages/torchaudio/lib/libtorchaudio.so: undefined symbol: _ZN2at4_ops10zeros_like4callERKNS_6TensorEN3c108optionalINS5_10ScalarTypeEEENS6_INS5_6LayoutEEENS6_INS5_6DeviceEEENS6_IbEENS6_INS5_12MemoryFormatEEE

[kijai]

Something about pytorch-lightning, possible you have some other node that uses older version.. I'm not forcing a version as it can lead to such issues with other nodes, you could check if you have pytorch-lightning>=2.2.1 at least, and force install the specific version that is known to work: pip install pytorch-lightning==2.2.1

What torch version are you running?

[mike-rowley]

I ran the lightning pip command and everything is updated to 2.2.1 and I get the same error loading the comfy server

Pytorch version is '2.3.0+cu121'

Thanks for the assistance in sorting this out 🙏

[kijai]

I ran the lightning pip command and everything is updated to 2.2.1 and I get the same error loading the comfy server

Pytorch version is '2.3.0+cu121'

Thanks for the assistance in sorting this out 🙏

Just checking, did you run the command in the python virtual environment of your install? As in activated it first with the venv/Scripts/Activate

[mike-rowley]

Yes I ran the pip command in the env for comfy

[mike-rowley]

I was able to resolve my issue from this post https://github.com/pytorch/audio/issues/62

pip install -U torch torchaudio --no-cache-dir

Apparently the -U is the important fix, after this command and a ComfyUI restart, Supir loads again.