Newsy.co

ComfyUI Notebook for RunPod

This post is related to the ComfyUI Notebook that I have created for RunPod and is available for purchase on our Shop for a small fee. This fee helps support the blog and channel to keep creating the wonderful content for you.

The notebook is made by us to get to started with using ComfyUI in under 5 mins on RunPod. This idea came to me when I got tired of many ComfyUI images built by the community taking too long to start up (some take up to 30mins) which wastes the GPU time which you are paying for on RunPod. It also downloads and install models that are pre-packaged so I don’t have control of what models are loaded by default.

As such I designed this notebook for my initial use but decided to share it with our community (as a paid product) to help generate some revenue to support the various costs that I incurr in writing and recording this content for you.

Get started

Once you purchase the ComfyUI Notebook for RunPod from our Shop you will have a ZIP file which will create the notebook (.ipynb file). You will need to upload this to RunPod’s Jupyter Lab and launch it.

Launch any RunPod instance that is has RunPod Pytorch 2.x.x in the name, these are official pods built by the RunPod team so guaranteed to work. Edit and override the setting below: Container Disk, Volume Disk and Expose HTTP Ports

For SD1.5/SDXL Models I suggest you set: Container Disk to 30 GB and Volume Disk to 50GB. However for Flux I would suggest you set the Volume Disk to 100GB.

For all cases make sure you add the port 8188 in Expose HTTP Ports separated by the Comma (as shown above in the screenshot).

  • 8888 – this is the port where the Jupyter Lab runs
  • 8188 – this is the port that ComfyUI will run on

Click on Set Override and launch the Pod via Deploy On-Demand.

Your Pod should start in 30 seconds or less and you will see the Connect button. I have done a full walk-through of this notebook that you can watch in the below video.

Troubleshooting

Common Issues

  • OSError: [Errno 98] error while attempting to bind on address (‘0.0.0.0’, 8188): address already in use. You might see this message when you try to launch ComfyUI, if you do this means that ComyUI is still running and didn’t stop successfully.
    • Solution – you need to Force Terminate ComfyUI as explained below.
  • Accidentally closed Terminal that was running ComfyUI. Not to worry the process is still running in the background you just can’t see it.
    • Solution – If you want to see the logs you can use tail command in a new terminal or Force Terminate ComfyUI and launch again.
  • Cannot download Flux Models – you get an error when you run that cell in Notebook.
    • Solution – make sure the Hugging face token is entered in the first cell marked “Must Run”. Enter the Token there and Run that cell before trying to run the download cells in the Notebook for Flux models. You can get a new token via https://huggingface.co/settings/tokens

Force Terminate ComfyUI

In the event your ComfyUI is not responding you need to KILL the process. Yes that’s actually the command in Linux, called Kill.

First need to identify what Process ID (PID) ComfyUI is running under. To do this, in a new Terminal run the command: ps -fA | egrep "python|PID", this will give you a list of processes. You are looking for python /workspace/ComfyUI/main.py --listen.

In the below screenshot the PID is 449. So next command we run is kill -9 449 each time replace the 449 with whatever the PID you get from the previous command.

You can then run the command ps -fA | egrep "python|PID" again, which will confirm that ComfyUI is not running.