notes.txt

### 1 - Deploying product

## Local work
# Build the docker file
docker build --build-arg HF_TOKEN=$HF_TOKEN -t mlops-cs3 .

# run the docker image locally
docker run -p 7860:7860 mlops-cs3

# tag the docker image to docker hub repository
docker tag mlops-cs3 venkateshroshan/mlops-cs3:latest

# push the docker image to docker hub repository
sudo docker push venkateshroshan/mlops-cs3:latest

## VM Server
# login to vm server via ssh
ssh -p 2222 group13@melnibone.wpi.edu

# pull the docker image
docker pull venkateshroshan/mlops-cs3:latest

# run the docker image
docker run -d -p 7860:7860 venkateshroshan/mlops-cs3

# access the application in localhost via ssh tunneling
ssh -L 7860:localhost:7860 -p 2222 group13@melnibone.wpi.edu
http://localhost:7860

-> explanation : Forwards local machine’s port 7860 to port 7860 on the VM and the ssh server is listening on port 2222.

### 2 - Monitoring product
# add prometheus-node-exporter in dockerfile and rebuild the docker image
RUN apt-get install -y prometheus-node-exporter

# exposing ports 
# Prometheus Node Exporter metrics
EXPOSE 9100
# Prometheus Python app metrics
EXPOSE 8000

# Run both the Node Exporter and the Gradio application
CMD ["sh", "-c", "prometheus-node-exporter & python app.py"]

# run docker
docker run -d -p 25560:7860 -p 25561:9100 -p 25562:8000 venkateshroshan/mlops-cs3:latest4

# access the application in localhost via ssh tunneling
ssh -L 25560:localhost:25560 -L 25561:localhost:25561 -L 25562:localhost:25562 -p 2222 group13@melnibone.wpi.edu

Gradio application at http://localhost:25560
Prometheus monitoring at http://localhost:25561
Prometheus metrics endpoint at http://localhost:25562

### 3 - Expose your product using ngrok (or a similar service) 
* Expose your service globally
# add authtoken of ngrok
ngrok authtoken YOUR_AUTH_TOKEN

# expose the port
ngrok http 25560 -> this gives public URL

ngrok http 25561

ngrok http 25562

# to run forever - use nohup
nohup ngrok http --log=log.txt http://localhost:25560 > /dev/null &


# Grafana visualization
docker pull grafana/grafana:latest
docker run -d -p 25563:3000 --name grafana grafana/grafana:latest

# created a prometheus.yml file
global:
  scrape_interval: 5s  # Adjust as needed

scrape_configs:
  - job_name: 'node-exporter'
    static_configs:
      - targets: ['172.17.0.1:25561']

  - job_name: 'app-metrics'
    static_configs:
      - targets: ['172.17.0.1:25562']

docker run -d -p 25564:9090 -v $(pwd)/prometheus.yml:/etc/prometheus/prometheus.yml prom/prometheus

ssh -L 25560:localhost:25560 -L 25561:localhost:25561 -L 25562:localhost:25562 -L 25563:localhost:25563 -L 25564:localhost:25564 -p 2222 group13@melnibone.wpi.edu

# To check whether the state of the site is up or down
http://localhost:25564/targets?search=

go to grafana http://localhost:25563
create a dashboard and add prometheus with the link as http://192.168.0.230:25564

# Final RUN
docker pull venkateshroshan/mlops-cs3:latest4

docker run -d -p 25560:7860 -p 25561:9100 -p 25562:8000 --name group13_app venkateshroshan/mlops-cs3:latest4

docker pull docker pull grafana/grafana:latest

docker run -d -p 25563:3000 --name group13_grafana grafana/grafana:latest

docker pull prom/prometheus

docker run -d -p 25564:9090 -v $(pwd)/prometheus.yml:/etc/prometheus/prometheus.yml --name group13_prometheus prom/prometheus

ssh -L 25560:localhost:25560 -L 25561:localhost:25561 -L 25562:localhost:25562 -L 25563:localhost:25563 -L 25564:localhost:25564 -p 2222 group13@melnibone.wpi.edu

# visualize these 4 metrics
transcription_requests_total - Total transcription requests
transcription_request_duration_seconds - Duration of transcription requests in seconds
transcription_memory_usage_bytes - Memory used by the transcription function
ram_usage_percentage - Percentage of total RAM used by the transcription function

### Case study 4
docker build --build-arg HF_TOKEN=$HF_TOKEN -t mlops-cs4 .
docker tag mlops-cs4 venkateshroshan/mlops-cs4:latest
sudo docker push venkateshroshan/mlops-cs4:latest
docker pull venkateshroshan/mlops-cs4:latest
docker run -d -p 7860:7860 -p 8001:8000 -p 9100:9100 venkateshroshan/mlops-cs4:latest