Prometheus Client

node exporter와 같은 서비스를 직접 만들 수 있는 SDK이고,

python, go 둘 다 지원을 하며 문서도 잘 되어있고 매우 간단하게 사용할 수 있다

 

https://github.com/prometheus/client_python

GitHub - prometheus/client_python: Prometheus instrumentation library for Python applications

https://prometheus.github.io/client_python/

client_python

 

 

특정 큐 컨슈머 숫자를 dashboard에서 보기 위해서 아래 처럼 응용을 할 수 있는데
Gauge를 추가할 때마다 /metrics에 데이터가 하나씩 추가되는 구조이고, 첫번째 파라미터로 쿼리를 할 수 있음

 

import os
import time
import requests
from prometheus_client import start_http_server, Gauge

queue_name = os.getenv('QUEUE_NAME', '***')
sleep_time = int(os.getenv('SLEEP_TIME', 10))

api_url = f"http://rabbitmq.***.com:15672/api/queues/%2F/{queue_name}"
credentials = ('***', '***')

# Prometheus gauge 설정
g = Gauge('rabbitmq_consumer_count', 'Number of consumers for the RabbitMQ queue')

def fetch_consumer_count():
    response = requests.get(api_url, auth=credentials)
    if response.status_code == 200:
        queue_info = response.json()
        consumer_count = queue_info['consumers']
        g.set(consumer_count)
        print(f"Consumer count: {consumer_count}")
    else:
        print(f"Failed to fetch queue information. Status code: {response.status_code}, Response: {response.text}")

if __name__ == '__main__':
    start_http_server(8000)

    while True:
        fetch_consumer_count()
        time.sleep(sleep_time)

 

 

prometheus 차트에 exporter 추가

prometheus:
  prometheusSpec:
    additionalScrapeConfigs:
      - job_name: 'rabbitmq'
        static_configs:
          - targets: ['consumer-count-service.default.svc.cluster.local:8000']

 

 

완료