search

aws / apprunner-roadmap

This is the public roadmap for AWS App Runner.
https://aws.amazon.com/apprunner/
Other
292 stars 13 forks source link

Add a feature in AppRunner service which will allow to increase the ephemeral storage #178

Closed raesquiv360 closed 1 year ago

raesquiv360 commented 1 year ago

Community Note

Tell us about your request What do you want us to build? We would like to be able to increase the ephemeral storage so more than 3 GB can be used.

Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered. For now, we can use ECS service as alternative approach.

Additional context Anything else we should know? As mentioned in https://docs.aws.amazon.com/apprunner/latest/dg/develop.html#develop.considerations increasing the storage limit will allow us to use container images that exceed 3Gb and also have space to store and work with large files

Attachments If you think you might have additional information that you'd like to include via an attachment, please do - we'll take a look. (Remember to remove any personally-identifiable information.)

amitgupta85 commented 1 year ago

What would be a reasonable ephemeral storage that you would like us to support?

jsheld commented 1 year ago

Issue 112 is asking for the same: https://github.com/aws/apprunner-roadmap/issues/112. Closing this issue as it is a duplicate of that one.

asford commented 10 months ago

@jsheld It doesn't appear that #112 is asking for the same feature. Right now the AppRunner ephemeral storage limit, and resulting container image size limit, is well documented at 3gb. However the storage limit is quite lower than:

This is a near blocker for some scientific software applications, as a standard image for python scientific/numeric computing is at or beyond a 3gb image size. For calibration, continuumio/anaconda3 is a representative base image designed to support low-effort scientific python applications using the anaconda ecosystem. The image is ~1.5gb compressed, and ~5gb uncompressed. It launches comfortably in standard fargate environments, but isn't compatible with AppRunner's stated limits.

This is obviously a pretty large image, but it's standard practice for simple projects in data science applications to include a relatively large base image. Given that AppRunner targets a simplified deployment story, it'd be extremely useful to be able to host these classes of simple applications in AppRunner.

msamsami commented 8 months ago

Any updates on this issue? 3 GB is not enough storage for many applications, especially those involved with language models, computer vision, GPU, etc.

I'm getting this error while building my app on App Runner which I assume is because of storage issue:


2023-10-05T14:38:53.808+03:30 | [Build] Downloading nvidia_cusparse_cu12-12.1.0.106-py3-none-manylinux1_x86_64.whl (196.0 MB)
-- | --
| 2023-10-05T14:38:53.808+03:30 | [Build] Collecting nvidia-cuda-nvrtc-cu12==12.1.105
| 2023-10-05T14:38:53.808+03:30 | [Build] Downloading nvidia_cuda_nvrtc_cu12-12.1.105-py3-none-manylinux1_x86_64.whl (23.7 MB)
| 2023-10-05T14:38:53.808+03:30 | [Build] Collecting nvidia-cudnn-cu12==8.9.2.26
| 2023-10-05T14:38:53.808+03:30 | [Build] Downloading nvidia_cudnn_cu12-8.9.2.26-py3-none-manylinux1_x86_64.whl (731.7 MB)
| 2023-10-05T14:38:53.808+03:30 | [Build] /root/.pyenv/pyenv.d/exec/pip-rehash/pip: line 20: 403 Killed "$PYENV_COMMAND_PATH" "$@"
| 2023-10-05T14:39:02.803+03:30 | [AppRunner] Failed to deploy your application source code.
GitNilsKrause commented 6 months ago

Could you please consider increasing the default size of the ephemeral storage, or provide an option to configure its size? This enhancement would significantly boost AWS's appeal in comparison to GCP's App Engine.