Closed Johnz86 closed 2 years ago
Hi @Johnz86 , Thank you for using MONAI Deploy App sdk and raising an issue!
Given the information you provided, it was hard for me to reproduce it. If you don't mind, could you please attach log file in here? I am particularly interested in which commands were executed when creating docker image
Step 1/29 : FROM nvcr.io/nvidia/pytorch:21.07-py3
...
Step 5/29 : LABEL sdk_version="0.3.0+5.gc64296d"
Seeing the following message (having only 15 steps and using nvcr.io/nvidia/pytorch:21.07
instead of nvcr.io/nvidia/pytorch:21.07-py3
) doesn't look correct.
[ 1/15] FROM nvcr.io/nvidia/pytorch:21.07
And I guess executing the following command is supposed to give error because there is no package
command.
monai-deploy apps/simple_imaging_app -t simple_app:latest -l DEBUG
# monai-deploy package apps/simple_imaging_app -t simple_app:latest -l DEBUG
Could you please try upgrade the package version
pip install --upgrade monai-deploy-app-sdk
and following the instruction below works?
FYI, we didn't test on the Windows Operating system. It seems that you are running on Windows (not Windows Subsystem for Linux[WSL]). MONAI Deploy App SDK is officially supporting Ubuntu OS and Windows users can install CUDA on WSL and work inside WSL to use MONAI Deploy App SDK.
Links to documentation provided.
Describe the bug Executing monai-deploy package ends with success. Produced docker image is not available locally and for next monai-deploy run. Invisible failed builds continuously consume disk space and they are not locally available.
Steps/Code to reproduce bug Consider following long run of command, which does not contain any error in output:
Following command fails next.
Expected behavior Failed builds should be visible. Failed builds should not occupy disk space. If the build is succesfull, then following command
Should not fail on not finding image locally.
Environment details (please complete the following information)
Additional context Latest docker is installed. This is locally sucesfull: