ComfyUI Node
Legion: Configuration
A ComfyUI node in Legion with 1 input and 1 output.
Legion: Configuration
- legion_config
◄config_yaml# Legion Configuration
# ---------------------------
# This is the default template for a worker's configuration.
# You can edit this file to change what appears in the "Legion: Configuration"
# node when you first create it.
comfyui:
# 'type': The type of worker. For now, only 'local_process' is supported. 'docker' will be added in the future.
type: local_process
# 'port': The port on which the worker will listen.
# 'auto' will use the automatic port management defined in the global config.yaml.
port: auto
# 'paths': Paths used by the worker.
paths:
# 'comfyui_path': Path to the main ComfyUI root directory (required to launch a local worker).
# Leave empty to attempt auto-detection.
comfyui_path:
# 'python_executable': Path to the python executable to use to run comfyui (required to launch a local worker).
# Leave empty to attempt auto-detection.
python_executable:
# 'custom_nodes_template': The name of the template folder in 'runtime/ComfyUIs'
# to be used for this worker's custom nodes.
custom_nodes_template: "OnlyCPU"
# 'execution': Settings related to the execution environment.
execution:
# 'startup_timeout': Maximum time to wait for this specific worker to start (in seconds)
# Leave empty to use the global default from config.yaml (usually 300s)
# Override this if this particular worker needs more/less time
# Examples:
# - Fast setup (few nodes): 30
# - Standard setup: 120
# - Heavy setup (many nodes): 600
startup_timeout:
# 'dry_run': set this to true if you just want to test the input/output mechanics, results will be
# simulated, external comfyui execution will not be run
dry_run: false
# 'asynch': set this to true if you just want to execute the external workflow in an asynch mode,
# you need either "Legion: Join Campaign" or "Legion: Join All Campaigns" node/s to get the results (output_X)
asynch: false
# 'extra_args': use this to pass additional command line arguments to the launch of the external ComfyUI
# Leave empty if you don't need extra arguments
# Examples: "--gpu-only --preview-method auto" or ["--gpu-only", "--preview-method", "auto"]
extra_args:
# 'env_vars': use this to set environment variables for the worker process
# Useful for GPU selection, compilation settings, etc.
# Examples:
# CUDA_VISIBLE_DEVICES: "1" # Use only GPU 1
# TORCH_COMPILE_DISABLE: "1" # Disable torch compilation
# PYTORCH_CUDA_ALLOC_CONF: "max_split_size_mb:512"
env_vars:
# CUDA_VISIBLE_DEVICES: "0"
# TORCH_COMPILE_DISABLE: "1"
# 'workflow': path on disk of the workflow to run in the other comfyui instance
workflow: plain_face_restore_api.json►
CategoryLegion
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| config_yaml | STRING | # Legion Configuration # --------------------------- # This is the default template for a worker's configuration. # You can edit this file to change what appears in the "Legion: Configuration" # node when you first create it. comfyui: # 'type': The type of worker. For now, only 'local_process' is supported. 'docker' will be added in the future. type: local_process # 'port': The port on which the worker will listen. # 'auto' will use the automatic port management defined in the global config.yaml. port: auto # 'paths': Paths used by the worker. paths: # 'comfyui_path': Path to the main ComfyUI root directory (required to launch a local worker). # Leave empty to attempt auto-detection. comfyui_path: # 'python_executable': Path to the python executable to use to run comfyui (required to launch a local worker). # Leave empty to attempt auto-detection. python_executable: # 'custom_nodes_template': The name of the template folder in 'runtime/ComfyUIs' # to be used for this worker's custom nodes. custom_nodes_template: "OnlyCPU" # 'execution': Settings related to the execution environment. execution: # 'startup_timeout': Maximum time to wait for this specific worker to start (in seconds) # Leave empty to use the global default from config.yaml (usually 300s) # Override this if this particular worker needs more/less time # Examples: # - Fast setup (few nodes): 30 # - Standard setup: 120 # - Heavy setup (many nodes): 600 startup_timeout: # 'dry_run': set this to true if you just want to test the input/output mechanics, results will be # simulated, external comfyui execution will not be run dry_run: false # 'asynch': set this to true if you just want to execute the external workflow in an asynch mode, # you need either "Legion: Join Campaign" or "Legion: Join All Campaigns" node/s to get the results (output_X) asynch: false # 'extra_args': use this to pass additional command line arguments to the launch of the external ComfyUI # Leave empty if you don't need extra arguments # Examples: "--gpu-only --preview-method auto" or ["--gpu-only", "--preview-method", "auto"] extra_args: # 'env_vars': use this to set environment variables for the worker process # Useful for GPU selection, compilation settings, etc. # Examples: # CUDA_VISIBLE_DEVICES: "1" # Use only GPU 1 # TORCH_COMPILE_DISABLE: "1" # Disable torch compilation # PYTORCH_CUDA_ALLOC_CONF: "max_split_size_mb:512" env_vars: # CUDA_VISIBLE_DEVICES: "0" # TORCH_COMPILE_DISABLE: "1" # 'workflow': path on disk of the workflow to run in the other comfyui instance workflow: plain_face_restore_api.json | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| legion_config | LEGION_CONFIG | — |