ComfyUI Manager button doesn't show
Hi, I just started my first RunPod, installed the ComfyUI from template and tried to install Manager for custom nodes, but the button doestn't appear. Even if I fully stop the Pod and launch it again. I installed the Manager using this:
cd /workspace/ComfyUI/custom_nodes git clone https://github.com/ltdrdata/ComfyUI-ManagerAny idea what to do?
GitHub
GitHub - ltdrdata/ComfyUI-Manager
Contribute to ltdrdata/ComfyUI-Manager development by creating an account on GitHub.
91 Replies
Which ComfyUI template are you using?
runpod/stable-diffusion:comfy-ui-3.0.2
this one
You may want to try the community one with Jupyter, other people also had issues with the RunPod official one.
I tried the community one with Jupyter, but it got stuck here 😦 any ideas?
Not sure, you can ask the dev
figured it out, the storage was full
Ah cool, @RobBalla is the developer of the template if you have any further issues.
thanks 🙂 just maybe one last question - how can I add some nwe checkpoint?
Use the terminal, you can't use Jupyter because firstly Jupyter is bad at handling large files and secondly it has a bug where any folder called
checkpoints
is inaccessible.
Most models on CivitAI seem to need authentication these days so you can either add headers to your wget command to add your CivitAI API key, or else download locally and use runpodctl
to transfer the files to your pod.GitHub
GitHub - runpod/runpodctl: 🧰 | RunPod CLI for pod management
🧰 | RunPod CLI for pod management. Contribute to runpod/runpodctl development by creating an account on GitHub.
I was trying to use the terminal, but I can't get to the file.. 😕
Linux is case sensitive, it has to match exactly, ie.
ComfyUI
.still nothing
Try this:
I know now..
there was the / before ComfyUI..
Yes exactly
got there 🙂 thanks for helping 🙂
but I would have bet there need to be the / before the folder name.. but for some reason it doesn't work here with that 😕
No it doesn't work like that, nothing before the folder name if you are changing directory into a sub-directory of the directory you are already in.
/
in Linux is the very top level (root) of your filesystem, it is very unlikely that you would ever need to use it for anything.tehre is this command in the manual here with the / before the folder name so I thought it should be always there.. https://civitai.com/articles/3304/a-guide-to-diving-into-ai-image-creation-on-comfyui-with-gpu-rentals
A Guide to Diving into AI Image Creation on ComfyUI with GPU Rental...
The information provided in this article may include references to various websites, resources, or tools that are mentioned based on the personal p...
I don't see where its doing that anywhere.
well doesn't matter now 🙂 thanks for your help today 🙂
Hi, I am having issues with the ComfyUI manager not showing up in any of the PODS I have tried. Any new ideas?
You are probably using a template that doesn't have the manager installed.
Did you try this one?
https://discord.com/channels/912829806415085598/1144401747100582008
I know its pre-installed in this template but not sure about others.
Yes! I have used this template but it is still not showing up. I have deleted and re-installed the manager myself but still nothing.
Do you know if there is anyone I can reach out to who will be willing to check this?
Maybe you installed a custom node that broke it
I'll check if it works for me
Yeah, it was working and then I did install some nodes and thereafter it was gone
I’m using my network volume to run the pod
Yeah then a custom node definitely broke it
Check the comfyui log file
Ok I will check, thanks. But will this problem occur again with the custom nodes? I need those nodes for my workflow.
There are probably some conflicting dependencies or something, its impossible to know without checking the log.
Thanks! Will check those logs
Hi! Hope you are well! I have sourced the ComfyUI log which will hopefully shed some light on the missing manager button
this is it
I don't see any errors there, don't know whats up with "Stopped server" though
I have just deployed the above mentioned comfy template but i cant even connect to it. The log shows that the container is ready but its just the clock page on the referred port.
Im gonna terminate and try another template. Really annoying that none of them work.
its probably still syncing to workspace when that happens
check container logs like the 502 page says
i already checked - said container was ready
Then check comfyui logs
Are you using network storage?
yes
this is the latest log
Traceback (most recent call last):
File "/workspace/ComfyUI/main.py", line 73, in <module>
import comfy.utils
File "/workspace/ComfyUI/comfy/utils.py", line 5, in <module>
import safetensors.torch
ModuleNotFoundError: No module named 'safetensors'
Did you perhaps add
DISABLE_SYNC
environment variable to your pod?OK, I have now deployed the ai-dock - comfyui and it is working properly! The manager button is also there.
Happy days! not sure why the other templates were giving issues.
Thanks for your help! Much appreciated!
Yeah thats strange, seems like some venv bug
i will stick to this template from now on. Thanks
hi - i deployed this docker and it does not show all the ports properly but can use the web terminal
Which docker?
Stable Diffusion Kohya ComfyUI InvokeAI Ultimate
it loads, and can go into the folders properly
Probably syncing to /workspace, check logs
but then it only shows port 3000 and 8888
logs look fine
what should I be looking for in the log
Oh thats a RunPod bug unfortunately
All ports should show, just not be available if there is an issue
if they aren't showing, its a RunPod bug
yes so do i just wait for runpod ? or is there an http address i can use
to see if the ports are open
so nothing I can do?
theres a specific format for it
https://{POD_ID}-{INTERNAL_PORT}.proxy.runpod.net
try to fill in the {}, dont actually use the {} just replace it ..
internal pod is your portIf the ports are not showing, I suggest terminating the pod and creating a new one.
ok ill do that - what is the easiest way to copy over exact workspace from old pod to new?
Runpodctl, rsyc, syncing using cloud sync
ok so no way to actually move workspaces pod to pod easily
have to download and reupload
Yeah, but if you have any idea or feature requests for runpod put it on #🧐|feedback
i think they are working on that kind of feature too
after setting Comfy UI with a new Network volume, deploying a new pod & the initial install, it work wonders and didnt have issues.
When starting a new pod from an existing network volume, i am facing issues with missing model, nodes and most important, the model manager not appearing.
Does anyone has still issues with the ComfyUI when starting a new pod?
two attempts with different pods and network volumes
manager still not showing
You are probably using RunPod ComfyUI which keeps venv on container disk.
Don't use the RunPod ComfyUI template, there are others that sync properly.
Looks like you need to install cv2. Does this app have requirements.txt file? If so, add opencv-python to it.
ComfyUI manager installs everything but it gets lost if stuff is installed to container disk and pod is restarted.
Its better to use templates that keep the venv in /workspace and not the container disk.
I think all of them use /workspace besides the "RunPod ComfyUI" one.
RunPod ComfyUI is broken
thank for your messages! it makes sense what you are saying and it was alongside my thoughts.
Is there a template you can recommend?
Madiator2011v2
PeerTube
Better ComfyUI Template Guide | [28.07.2024]
Guide how to use my RunPod ComfyUI template
thanks! will try
is it possible it takes more than 10 min to load the port to connect?
I can help you. please feel free to reach out.
hey! yes that would be good!
trying now with madiator template, but takes some time to load the port
thank you! yes please
im now trying a new pod and restarting the process
with this template
23 min and port still not loaded.. am i missing something?
do you mind taking a look at the logs and see if im missing a trick?
@LupaLuna if you use network storage first sync takes time
Oh I see you have broken custom nodes
Go to port 7777 and there you have notebook to fix custom nodes
30 min in and the port opened
Next boots should be faster
is it possible nodes are broken because i install the with a diff template and py version 10.3 (think it was that one)
There is notebook that can fix nodes for you then
It reinstall python dependencies that might be missing
ok will try
Just need to make sure to read and run first two cells
i might have made the mistake of choosing the older py when trying to run the first cell... can i undo that?
Running cells with 'Python 3.10.12' requires the ipykernel package.
Run the following command to install 'ipykernel' into the Python environment.
Command: '/bin/python3.10 -m pip install ipykernel -U --user --force-reinstall
There is no Pip installer available in the selected environment.
Pro tip remove venv it will recreate fresh one my image is based on py3.11
oh! where can do that?
all venv folder?
Port 7777 has code server
hey! tried to login again this morning and 40 min ins and the port is not loaded yet. How long does it take you to load?
you have deleted venv? If so it would resync from scratch
tried with a new volume too, but taking quite some time
most nodes are saying they cant update when I click try again it does nothing and comes back to same thing
Ui bug cause cloudflare timeouts