Some streamlit scripts that we use to display our training logs (sbord) and to monitor our servers (sstat). Note that the latter requires monitoring data in some quite specific and undocumented format so it is probably not that re-usable.

Run directly as streamlit run sbord.py -- <path/to/log/folder> or install and run via

conda env create -f environment.yaml
conda activate sbord
sbord <path/to/log/folder>

<path/to/log/folder> defaults to . if not specified.

The sbord executable assumes you are using miniconda3 and that it is located at ~/miniconda3. If you run into problems, adjust those paths in bin/sbord and reinstall using pip install -e ..

To run from anywhere without activating the environment first, add ~/miniconda3/envs/sbord/bin/sbord to your path:

export PATH="${PATH}:~/miniconda3/envs/sbord/bin/sbord"

Same as above. Point first argument to folder containing free_data.csv, process_data.csv and utilization.csv. It's convenient to to

alias sstat="~/miniconda3/envs/sbord/bin/sstat <path>"

To adjust streamlit settings directly, run as

streamlit run sboard/sstat.py --server.port 8080 -- <path/to/queue>

Just a copy of sstat for deployment.