ComfyUI No Module Named Error: When It Is Safe to Ignore

If ComfyUI prints No module named ..., do not assume the whole install is broken.

In many real startup logs, the missing package belongs to one custom node or optional backend. ComfyUI may still open normally, and only the workflow that uses that node will fail. The right move is to classify the warning before installing packages.

Telemetry snapshot queried on April 17, 2026: 5,159 events across 76 installs for comfyui_fatal_error_detected. This was the largest telemetry bucket in that snapshot, but most events behaved like plugin-scoped import noise rather than full launcher crashes.

Quick Answer

What the Log Looks Like

A non-fatal plugin warning often looks like this:

IMPORT FAILED: ComfyUI-SomeCustomNode
ModuleNotFoundError: No module named 'somepackage'

or:

Cannot import custom nodes from ComfyUI-SomePlugin:
No module named 'onnxruntime'

or:

Found optional backend triton: {'available': False, 'unavailable_reason': "ImportError: No module named 'triton'"}

The key question is not whether the log says fatal. The key question is whether ComfyUI became ready and which plugin actually tried to import the missing module.

Step 1: Check Whether ComfyUI Is Actually Running

Look for the normal ready line:

To see the GUI go to: http://127.0.0.1:8188

If the browser UI opens and a simple workflow runs, treat the missing module as a plugin-specific warning. You can still fix it, but you do not need to rebuild the whole environment first.

If ComfyUI never reaches the ready URL, use ComfyUI Startup Failed? How to Diagnose and Recover Faster before installing more packages.

Step 2: Find the Plugin and the Exact Missing Module

Search the startup log for IMPORT FAILED, ModuleNotFoundError, or No module named.

Capture three pieces of information:

If the UI opens, you can also check ComfyUI's import failure endpoint:

http://127.0.0.1:8188/v2/customnode/import_fail_info_bulk

That endpoint is useful because it groups import failures by custom node instead of forcing you to read the whole terminal log.

Step 3: Match the Missing Module to the Right Fix

Do not run random pip install commands from system Python. Use the Python environment that starts ComfyUI.

For the GitHub Windows portable package, open a terminal in the extracted portable root: the folder that contains the run_*.bat files, python_embeded, and ComfyUI.

That usually means:

.\python_embeded\python.exe -s -m pip install <package-name>

If that path is not found, you are probably inside the ComfyUI\ subfolder. Move one folder up and run the command again.

For ComfyUI Desktop or another managed launcher, do not copy the portable python_embeded command blindly. Use the app's terminal or environment tools so the package installs into the Python that actually launches ComfyUI.

For a manual virtual environment, activate the environment first, then use:

python -m pip install <package-name>

Common ComfyUI missing-module paths:

Step 4: Decide Whether to Ignore, Repair, or Rebuild

Ignore it when

• ComfyUI is ready
• your workflow does not use the failed plugin
• the warning is for an optional backend with a known fallback
• the missing package belongs to a feature you do not plan to use

Example: a Triton capability warning is often harmless when you are not using SageAttention, compiled kernels, or a Triton-backed video workflow.

Repair it when

• the workflow you need shows red nodes
• the failing plugin is required for your output
• the error maps to one known package
• pip check does not show a wider dependency conflict

Make one narrow repair, restart ComfyUI, then retest the same workflow.

Rebuild or escalate when

• multiple unrelated plugins fail at the same time
• pip check reports conflicts around torch, numpy, opencv-python, pillow, or transformers
• a previous repair changed CUDA, PyTorch, or Python package versions
• ComfyUI never becomes ready after the warning

At that point, the problem is no longer a single missing module. Use ComfyUI Dependency Conflicts or How to Fix ComfyUI Plugin Import Failed Errors.

What Not To Do

• Do not reinstall ComfyUI only because one optional plugin import failed.
• Do not install packages into system Python when ComfyUI uses embedded Python or a virtual environment.
• Do not repair every warning in the log before testing whether the workflow actually needs it.
• Do not run a plugin's requirements.txt blindly on a fragile environment; read it first for core package downgrades.
• Do not mix fixes from different Python versions, CUDA versions, or portable packages.

How Wonderful Launcher Helps

Wonderful Launcher is useful when the log has become hard to reason about:

• it keeps ComfyUI installs separated
• it makes the startup log easier to collect
• it helps preserve models and workflows before risky repairs
• it gives you a safer path than repeated command-line package changes

If ComfyUI is running and only one unused plugin is noisy, you probably do not need a repair session. If several plugin imports fail or a business-critical workflow is blocked, start from the exact missing module and fix one dependency path at a time.

Related Guides

• How to Fix ComfyUI Plugin Import Failed Errors
• ComfyUI Dependency Conflicts
• InsightFace Missing in ComfyUI
• ONNX / ONNXRuntime Missing in ComfyUI
• OpenCV cv2 Missing in ComfyUI
• ComfyUI-GGUF Missing
• llama_cpp Missing in ComfyUI
• piexif Missing in ComfyUI
• pkg_resources Missing in ComfyUI
• SageAttention Missing in ComfyUI
• Nunchaku Missing in ComfyUI
• Triton Missing or Unavailable in ComfyUI

Source References

• ComfyUI troubleshooting overview
• ComfyUI custom node troubleshooting guide
• ComfyUI model troubleshooting guide
• ComfyUI Manager overview