container configuration for streamlit app

ysaeed4

Dear Experts,

I am having difficulties in starting Streamlit app. In the container configuration, as shown in the attached image, when I select None, it gives error "Unable to start run : No container config to run Code Studio on".

However when I select container ai_exec gpu, it always say Pod is still missing.

I do not want to run the app using gpu, it can be run with cpus but I am unable to select that option.

Kind regards,

Yawar

ysaeed4

Hi @JordanB
,

Thanks for your help. I have managed to start the streamlit app. I have another question with regards to creating virtual environment because after connecting to my GitHub repository with streamlit app code and requirements.txt, I create virtual environment and sync files with DSS but it always says " ModuleNotFoundError: No module named 'tensorflow'". It seems like it does not install the modules at the correct path as required.

Kind regards,

Yawar

JordanB

Hi @ysaeed4
,

You will need to add a code environment block to your code studio template and rebuild the template (ref: https://knowledge.dataiku.com/latest/code/work-environment/tutorial-first-code-studio.html#configure-definition-settings).

Unfortunately, it will not work to create a virtual environment directly from your code i.e. virtualenv/venv.

Let me know if you have questions!

JordanB

Hi @ysaeed4
,

You will need to select a container configuration for Code Studios/to run streamlit (to fulfill the requirement of running your Code Studio on a Kubernetes Cluster). Note, you can create a cluster node and container config for cpu usage (Administration > Settings > Containerized Execution). Please see the following tutorial for requirements and setup guidance: https://knowledge.dataiku.com/latest/code/work-environment/tutorial-first-code-studio.html