-
Notifications
You must be signed in to change notification settings - Fork 231
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
feat: Use Amazon S3 as Stable Diffusion Model Storage #616
base: main
Are you sure you want to change the base?
Conversation
…om/lindarr915/data-on-eks into bottlerocket-cache-container-image
…ata-on-eks into cache-model-from-s3
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@lindarr915 thanks so much for the PR! Please address the comments. Also make sure you run pre-commit run -a
and commit the changes made.
gen-ai/inference/stable-diffusion-rayserve-gpu/ray-serve-stablediffusion-from-s3.yaml
Show resolved
Hide resolved
image: public.ecr.aws/data-on-eks/ray2.11.0-py310-gpu-stablediffusion:latest | ||
imagePullPolicy: IfNotPresent # Ensure the image is always pulled when updated | ||
image: public.ecr.aws/data-on-eks/ray-serve-gpu-stablediffusion:2.33.0-py311-gpu | ||
imagePullPolicy: Always # Ensure the image is always pulled when updated |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why Always
?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
If the image public.ecr.aws/data-on-eks/ray-serve-gpu-stablediffusion:2.33.0-py311-gpu is not up-to-date, then kubelet will pull the image from ECR. Otherwise, it will use the cached image.
https://kubernetes.io/docs/concepts/containers/images/#image-pull-policy
Always
every time the kubelet launches a container, the kubelet queries the container image registry to resolve the name to an image digest. If the kubelet has a container image with that exact digest cached locally, the kubelet uses its cached image; otherwise, the kubelet pulls the image with the resolved digest, and uses that image to launch the container.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thus, if I update the image with the same tag, I can make sure it will use the latest version. If the SHA256 is the same, then kubelet will use the cached image on disk
|
||
## End-to-end Example | ||
|
||
An end-to-end deployment example can be found in [Stable Diffusion on GPU](../gen-ai/inference/GPUs/stablediffusion-gpus). |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Wrong architecture diagram.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
In the page there is no diagram. Would you let me know which architecture diagram you are referring to?
gen-ai/inference/stable-diffusion-rayserve-gpu/download_models.ipynb
Outdated
Show resolved
Hide resolved
- Added a new job to download and save models to Amazon S3. - Created a YAML file for the job configuration. - Defined a persistent volume and persistent volume claim for model storage. - Created a config map for the shell script used in the job. - Configured the job container with the necessary image and command. - Mounted the script and model storage directories in the container.
@askulkarni2 Please review the new commits for the PR. |
This PR has been automatically marked as stale because it has been open 30 days |
I am still working on this PR. |
fix: ray service yaml filename
This PR has been automatically marked as stale because it has been open 30 days |
What does this PR do?
Implement 448
🛑 Please open an issue first to discuss any significant work and flesh out details/direction - we would hate for your time to be wasted.
Consult the CONTRIBUTING guide for submitting pull-requests.
Motivation
More
website/docs
orwebsite/blog
section for this featurepre-commit run -a
with this PR. Link for installing pre-commit locallyFor Moderators
Additional Notes