add gpu monitoring

Author: Mubangizi

app/controllers/app.py

@@ -11,13 +11,15 @@
 class AppUsageView(Resource):
     @jwt_required
     def post(self, resource):
-        if resource not in ['cpu', 'memory', 'network']:
-            return dict(status='fail', message='Invalid resource name, pass cpu, memory, network'), 400
+        if resource not in ['cpu', 'memory', 'network', 'gpu']:
+            return dict(status='fail', message='Invalid resource name, pass cpu, memory, network, gpu'), 400
 
         app = get_app_data(request)
-
-        if app.status_code != 200:
-            return dict(status='fail', message=app.message), app.status_code
+        try:
+            if app.status_code != 200:
+                return dict(status='fail', message=app.message), app.status_code
+        except:
+            return app
 
         start = app.start
         end = app.end
@@ -53,6 +55,14 @@ def post(self, resource):
                     metric='sum(rate(container_network_receive_bytes_total{namespace="' +
                     namespace + '", pod=~"' + app_alias + '.*"}[5m]))'
                 )
+            elif resource == 'gpu':
+                prom_data = prometheus.query_rang(
+                    start=start,
+                    end=end,
+                    step=step,
+                    metric='sum(rate(container_gpu_usage_seconds_total{namespace="' +
+                    namespace+'"}[5m]))'
+                )
             else:
                 return dict(status='fail', message='Invalid resource name, pass cpu, memory, network'), 400
         except Exception as error:

app/controllers/project.py

@@ -1,7 +1,7 @@
 import json
 from prometheus_http_client import Prometheus
 from flask_restful import Resource, request
-from app.helpers.utils import get_project_data,is_valid_prometheus_query
+from app.helpers.utils import get_project_data, is_valid_prometheus_query
 
 from app.helpers.authenticate import (
     jwt_required
@@ -11,8 +11,8 @@
 class ProjectUsageView(Resource):
     @jwt_required
     def post(self, resource):
-        if resource not in ['cpu', 'memory', 'network']:
-            return dict(status='fail', message='Invalid resource name, pass cpu, memory, network'), 400
+        if resource not in ['cpu', 'memory', 'network', 'gpu']:
+            return dict(status='fail', message='Invalid resource name, pass cpu, memory, network, gpu'), 400
 
         project = get_project_data(request)
 
@@ -53,6 +53,14 @@ def post(self, resource):
                     metric='sum(rate(container_network_receive_bytes_total{namespace="' +
                     namespace+'"}[5m]))'
                 )
+            elif resource == 'gpu':
+                prom_data = prometheus.query_rang(
+                    start=start,
+                    end=end,
+                    step=step,
+                    metric='sum(rate(container_gpu_usage_seconds_total{namespace="' +
+                    namespace+'"}[5m]))'
+                )
             else:
                 return dict(status='fail', message='Invalid resource name, pass cpu, memory, network'), 400
         except Exception as error:

app/helpers/utils.py

@@ -109,7 +109,6 @@ def get_app_data(request):
     )
 
 
-
 # default mx data points for prometheus
 MAX_DATA_POINTS = 11000
 STEP_UNITS_IN_SECONDS = {
@@ -120,16 +119,19 @@ def get_app_data(request):
     "w": 604800,
 }
 
+
 def parse_step_to_seconds(step: str) -> int:
     """
     Parses a Prometheus step string like '1m', '2h', '4d' into seconds.
     """
     match = re.match(r"^(\d+)([smhdw])$", step)
     if not match:
-        raise ValueError("Invalid step format. Use formats like 30s, 5m, 2h, 1d.")
+        raise ValueError(
+            "Invalid step format. Use formats like 30s, 5m, 2h, 1d.")
     value, unit = match.groups()
     return int(value) * STEP_UNITS_IN_SECONDS[unit]
 
+
 def is_valid_prometheus_query(step: str, start_ts: int, end_ts: int) -> (bool, str):
     """
     Validates if the number of points in a Prometheus query is within the allowed range.
@@ -138,15 +140,14 @@ def is_valid_prometheus_query(step: str, start_ts: int, end_ts: int) -> (bool, s
         step_seconds = parse_step_to_seconds(step)
     except ValueError as e:
         return False, str(e)
-    
+
     if start_ts >= end_ts:
         return False, "Start timestamp must be less than end timestamp."
-    
+
     total_duration = end_ts - start_ts
     num_points = total_duration // step_seconds
 
     if num_points > MAX_DATA_POINTS:
         return False, f"Query returns {num_points} points, which exceeds the limit of {MAX_DATA_POINTS}. Increase the step or reduce the time range."
-    
-    return True, f"Query valid: {num_points} data points."
 
+    return True, f"Query valid: {num_points} data points."

docker-compose.yml

@@ -8,7 +8,7 @@ services:
       POSTGRES_USER: postgres
       POSTGRES_DB: cc-monitoring
     ports:
-      - "4200:5432"
+      - "4240:5432"
     volumes:
       - db-data:/var/lib/postgresql/data
 
@@ -18,17 +18,21 @@ services:
       context: .
       dockerfile: Dockerfile
     container_name: monitoring-api
+    networks:
+      - cranecloud
+      - default
     environment:
       FLASK_APP_SECRET:
       JWT_SALT:
       PYTHONDONTWRITEBYTECODE: 1
       PYTHONUNBUFFERED: 1
       FLASK_ENV: development
+      FLASK_DEBUG: 1
       DATABASE_USER: postgres
       DATABASE_URI: ${DATABASE_URI:-postgresql://postgres:postgres@database:5432/monitor}
       TEST_DATABASE_URI: ${TEST_DATABASE_URI:-postgresql://postgres:postgres@database:5432/monitor_test}
       FLASK_APP: server.py
-      PRODUCT_BASE_URL: ${PRODUCT_BASE_URL:-http://127.0.0.1:5000}
+      PRODUCT_BASE_URL: ${PRODUCT_BASE_URL:-http://flask-api:5000}
     ports:
       - "${FLASK_PORT:-4000}:5000"
     volumes:
@@ -38,6 +42,10 @@ services:
     links:
       - database
 
-  
 volumes:
-  db-data:
+  db-data:
+
+networks:
+  cranecloud:
+    external: true
+    name: cranecloud_default

scripts/start.sh

@@ -6,4 +6,4 @@
 flask db upgrade
 
 # start server
-flask run --host=0.0.0.0 --port=5000
+flask run --host=0.0.0.0 --port=5000 --debug --reload