-
-
Notifications
You must be signed in to change notification settings - Fork 237
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Bug] ComfyUI restarting / not working correctly #8
Comments
This may be because I didn't include a section to enable GPU. A fairly big oversight that I'll rectify soon. Could you try appending the following to the compose file
|
Added the suggested configuration, did not seem to work sadly. |
Thanks for the update. The ComfyUI process is ending too quickly. I need to change the supervisor startsecs to 5 (build/COPY_ROOT/etc/supervisor/supervisord/conf.d/comfyui.conf) to ensure its properly marked as a failure and maybe we will get some more useful information about why it's exiting. I haven't come across this bug previously. I'll start a rebuild shortly - Sorry for the inconvenience |
Tried building the image after changing the startsecs to 5, getting a build error: Build log
Based on the logs, it was building fine. |
What is your build environment? I have completed a successful local build prior to pushing changes on an intel i5, 4gb ram |
Docker version 24.0.7, build afdd53b4e3 Hardware: OS: Latest nVidia vulkan dev driver installed (through https://github.com/Frogging-Family/nvidia-all, works on other nVidia docker containers running pytorch) Maybe I am overseeing something for building the image correctly? |
I have no idea, unfortunately. To debug you can set +x on the build scripts to find exactly where the failure occurs. I only build on Ubuntu (same as the GitHub runners) and there are no issues. I love Arch - Long time user, but it may have an unstable package causing trouble... Which one is a mystery but could also be causing the runtime problems. Pushing a new build now. |
Hi @robballantyne, thanks for the awesome work! I think, I'm running into the same (or at least similar issue). Logs
I already activated the GPU, as mentioned in #8 (comment). I'm using an Nvidia GPU on Manjaro linux (basically arch). |
I have re-worked some of the base images and an update is imminent. Hopefully it will address these issues as there is a new docker-compose.yaml and a slightly modified build process. The new release includes some extra features (API requests to s3 storage which I need to be certain works properly before I commit) so it'll be early next week. It may be worth investigating any difference in how Arch configures docker Vs Ubuntu. I use Mint and Ubuntu exclusively to stay close to my production environment and there are no build problems here. I'd like it to work on Arch. I used it for many years after I gave up on Slackware so I'm very fond of it but it killed my productivity. If I do get some time I'll see what I can do to investigate but I can't give any ETA on that if the new release is still problematic. |
I just tested the current |
I have a similar issue on such an AMD system, also with always encountering endless reconnects of ComfyUI,
However errors are still here and reconnections are still being occured, even already with 16GB UMA slightly more than required for properly functioning,
And finally, how to "download" for local SDiff models in a config file? |
I just pulled this and tried it with latest-rocm, and am experiencing the same issue
|
Arch Linux again? |
Using the shipped docker-compose file, the container does not seem to start correctly.
Container logs spam the following lines over and over:
When trying to access
DOCKERMACHINE:8188
caddy seems to work but the container then spits out the following error:A full log can be found underneath:
Full log
Any way to further diagnose the problem?
The text was updated successfully, but these errors were encountered: