# bai-monitoring-api



## 1. Install requirements.txt

Install "requirements.txt" on your enviroment before running these codes.


## 2. Config config.py

"config.py" is for "app.py".

Configure the factors like ADMIN_ACCESS_KEY, ADMIN_SECRET_KEY,,,etc
You can find them in [ Backend.ai Control Pannel ].


## 3. Explanations of app.py

This is a api app which returns Backend.ai agent's current utilizations through json format.


Graphql api is used for fetching data from [ Backend.ai manager hub server ]. 

"generate_signature()" function generates all the signature needed for the connections with Backend.ai manager hub server.
"query_agent_list()" function queries some fields like {'id','occupied_slots','available_slots','live_stat'} from Agent Class.
"extract_utilization()" function makes json results we want
"myfunction_api()" function includes the Flask framework. Flask could make this code to an api application.


- host= ‘0.0.0.0’  => This settings make possible to access api from outside.
- port=‘31000’ => Port number accords to kubernetes's general port range. Except the numbers you already using.
- debug=True


## 4. Run api server


[ python3 agent_list_api_flask.py ]


It is running as a debug mode, so that you could see the logs in the console whenever api server gets requests.

![8ACCFB37-5919-4695-B853-2D9AA1A94802.jpeg](https://s3-us-west-2.amazonaws.com/secure.notion-static.com/0b46333b-aad0-4ba7-a886-4cffa5f0c2b3/8ACCFB37-5919-4695-B853-2D9AA1A94802.jpeg)



## 5. api url : [http://10.231.238.231:31000/api/getMonitoring]

response example

{
  "results": {
    "item1": {
      "cpu": 3.94,
      "cuda": 99.12,
      "disk": "25.02",
      "id": "i-ai-1",
      "mem": "4.85"
    },
    "item2": {
      "cpu": 4.69,
      "cuda": 14.12,
      "disk": "8.44",
      "id": "i-ai-2",
      "mem": "5.29"
    },
    "item3": {
      "cpu": 2.08,
      "cuda": 12.5,
      "disk": "8.28",
      "id": "i-ai-3",
      "mem": "3.63"
    },
    "item4": {
      "cpu": 4.25,
      "cuda": 16.5,
      "disk": "8.35",
      "id": "i-ai-4",
      "mem": "5.49"
    },
    "item5": {
      "cpu": 0.84,
      "cuda": 24.38,
      "disk": "41.31",
      "id": "i-ai-5",
      "mem": "3.48"
    },
    "item6": {
      "cpu": 2.14,
      "cuda": 0.0,
      "disk": "79.48",
      "id": "i-ai-6",
      "mem": "3.55"
    },
    "time": "2023-03-14T11:20:44.488009+09:00"
  }
}


## 6. Further usage (running on k8s)

- This git repo is used for the Dockerfile.

- Build a custom docker image by using Dockerfile.

- The custom docker image is for making kubernetes pods.

- Deploy the pods and service in k8s Cluster.

- Finally api server is running on k8s pods. You access api url with [ http://<host ip>:<service nodePort> ]