Issue with Runpod task response for Stable Diffusion Webui

[velvet crest]

Issue with Runpod task response for Stable Diffusion Webui

[velvet crest]

Can someone help me pls

"Task completed with output: {'images': ['B64StringHereType error occurred: unsupported operand type(s) for +: 'NoneType' and 'int'""

This is how the error is being sent. It is part of the B64 String for some reason.

[bright nacelle]

/run has a rate limit of 1000 requests per second.
/runsync has a rate limit of 2000 requests per second.
https://docs.runpod.io/docs/serverless-endpoint-urls

[velvet crest]

Oh, I dont think I ever reached that limit. I dont think I even crossed 1 request per second

[bright nacelle]

Upgrade to latest version of the RunPod SDK and see if it resolves the issue.

[velvet crest]

Okayy one minute

Wait, where do I do that? On their site?

[bright nacelle]

No. In your docker image.

[velvet crest]

Where can I check the latest version

My docker image uses

runpod==0.9.4

[bright nacelle]

Thats an ancient version, 1.5.2 is the latest.

pip install runpod>=1.5.2

You can check rhe latest version here:
https://github.com/runpod/runpod-python/releases

[velvet crest]

Thank you

[velvet crest]

I recharged $25 and it still did not work

[bright nacelle]

What are you running on your endpoint?

[velvet crest]

Stable Diffusion a1111

What is happening is, sometimes, it generates the image

And other times, it returns only part of the B64 String of the output then the error

Lemme show

See, you can see the B64 string but towards the end, the B64 String gets cut and the "Type error occurred: unsupported operand type(s) for +: 'NoneType' and 'int'" gets added in the end

[bright nacelle]

Which A111 worker?

[velvet crest]

From my Docker Image

[bright nacelle]

Where did you get it from?

Did you use someone's repo or just start from scratch?

[velvet crest]

RUN git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git &&
cd stable-diffusion-webui &&
git reset --hard 89f9faa63388756314e8a1d96cf86bf5e0663045 &&
pip install -r requirements_versions.txt

I got the Dockerfile from my boss, so I assume he made it.

It was working until Friday Morning

[bright nacelle]

Ask him whats happening or else log a bug in the A1111 repo.

[velvet crest]

It worked for like over 5-6 months, it randomly stopped 😭

[velvet crest]

I asked him, he said he is not sure 😭

[bright nacelle]

Log a bug in A1111 Github repo then

Maybe also use latest commit, maybe there is something wrong with the commit you're using

[velvet crest]

Hmm, I tried everything but it isn't working. I am also unable to create a new Docker Image for some odd reason, it throws some git bash error

[bright nacelle]

Are you using a Mac?

[velvet crest]

Nope Windows

This is the Error

"ERROR: failed to solve: process "/bin/bash -o pipefail -c cd /stable-diffusion-webui && python cache.py --use-cpu=all --ckpt /stable-diffusion-webui/models/Stable-diffusion/childrensStories_v1SemiReal.safetensors" did not complete successfully: exit code: 1"

[bright nacelle]

Try use https://depot.dev to build your image

[velvet crest]

Oh, it's paid 🥹

[bright nacelle]

It has a free trial or something

[velvet crest]

Ahh I see

My boss said that the Dockerfile might be outdated now

[bright nacelle]

Where did he get it from?

[velvet crest]

He said he got it from a client

Any idea on where I can get an updated SD Webui Dockerfile project that supports runpod?

[bright nacelle]

You may want to try mine:
https://github.com/ashleykleynhans/runpod-worker-a1111

You add your models to a network volume instead of the Docker image.

You also don't need to build the docker image, you can use mine.

[velvet crest]

Lmaooo I was just looking at that

So, this uses Runpod's Network Volume right?

Which kinda eliminates the need for Docker

[bright nacelle]

You still need Docker, everything on RunPod runs off Docker containers.

You just don't put your models into the docker image, you put them onto your network volume.

[velvet crest]

Ahh I see, I asked my boss if we can shift to your approach. Waiting for his reply

[velvet crest]

I have a doubt

Can't I attach it to my Serverless Pod? I mean the Network Volume

[bright nacelle]

You can attach network storage to your serverless endpoint under advanced settings

[velvet crest]

Do I need to create a normal Pod then?

So, I followed the instructions from https://github.com/ashleykleynhans/runpod-worker-a1111/blob/main/docs/installing.md

I did the automatic downloading functions in the Web Terminal of the Pod

@bright nacelleDo I still need the normal Pod if I am going to use your Docker Image in the template?

[bright nacelle]

Yes, you need to start a pod to configure everything and add your models etc, then once you've done that, you can terminate the pod.

[velvet crest]

Ahh okay okay thank you

My god it takes forever to upload the model to the Jupyter book thing

[velvet crest]

@bright nacelle How many GB should the Network Volume be?

I left it at 10 😭

ERROR: Could not install packages due to an OSError: [Errno 122] Disk quota exceeded

[bright nacelle]

10 is way too small, should be at least 50GB

[velvet crest]

50? Damn

[bright nacelle]

50 is only around $3.50 per momth

Mine are all 80GB

[velvet crest]

So, 50 GB Is enough for Basic? What if I want to add my own Model, LoRA, VAE and ADetailer?

[velvet crest]

Huh it's still live 😭

@bright nacelle Does the web terminal time out after a while? 😭

[bright nacelle]

Use SSH or Jupyter if Web Terminal is giving you issues.

[velvet crest]

I am getting errors ;-;

[bright nacelle]

Don't know why it can't write. Did you run all the other commands first?

You can use the auto install script to ensure everything is installed correctly

[velvet crest]

I used auto install only

[bright nacelle]

Something isn't right, you have to do it again.

I've used that script at least 10 times without issues so it works.

[velvet crest]

Yeah I deleted the Volume and Pod and I am restarting everything

[velvet crest]

Yeah, I think it worked

I have a doubt though, is there a faster way to add my Models? Doing it through Jupyter is a bit too slow

[velvet crest]

Okay, it is working. But do you have any idea on how to fix the error "Type error occurred: unsupported operand type(s) for +: 'NoneType' and 'int'" from Stable Diffusion?

[bright nacelle]

Use runpodctl, croc, sftp etc.

[bright nacelle]

No, not sure what you are doing that this is still happening, nobody else is getting this.

[velvet crest]

😭

Can I send you my Python code?

[velvet crest]

I just patiently waited for it to upload 😭

I have been facing this issue since Friday and it randomly started after I updated my code

[velvet crest]

omg I think I figured it out

Omg I did, thank you so much @bright nacelle

It was an error in code

But at least I updated the SD and Runpod packages

How do I close this

!close

[bright nacelle]

Usually there is an Answer Overflow message where you can close it from, but I don't see it for some reason.

[velvet crest]

Ahhh I see, ig it will auto close