Feat: Deploy microservice to VM

[qianxuege]

Summary

• This PR aims to modify the Dockerfile so that it runs in our VM.

Changes

• Changed original Dockerfile to dev.Dockerfile .
• Created a new Dockerfile that uses gunicorn instead of flask to address the error shown in the screenshot (about using production WSGI server).
• Removed calling ollama server in translator.py since that is being managed by the docker-compose.yml file
• Fixed case check for detected_language

Validation

• In VSCode, open the repo in container
• run docker compose up --build in the terminal
• open the browser it tells you to. For example:

• append /?content=Dies%20ist%20eine%20Nachricht%20auf%20Deutsch to the end of the browser url and reload.
  • successful if can see a response in JSON. The response should look like:
    {"is_english":false,"translated_content":"This is a message in German."}
• run docker compose down to close it afterwards.

Notes for deployment

• Steps to deploy to linux server:
  • connect to campus vpn and ssh [email protected]
  • run ls to see if this repo already exists on the VM.
    • If not, git clone this repo.
  • cd llm-experiment-microservice-jack/
  • git pull
  • docker compose up --build -d to start container in detach mode

[fatbiscuit247]

Looks good, through codespace, I saw the expected LLM response when running docker compose to comfirm LLM was running, and also was able to ssh to our team