The following page talks about a solution in python designed to monitor internet urls and provide Prometheus metrics. The solution is supposed to run on kubernetes cluster. Metrics from the python app should be exposed to prometheus and dashboard for the same can be built using grafana.
Prometheus and grafana can be installed using prometheus operator helm charts.
You will need the following software to run this program
- Python 3.X
- Install requirements.txt file.(This should be handled from within dockerfile)
Once the dependencies are installed, we can build a new docker image containing the python logic which probes URL's https://httpstat.us/200 and https://httpstat.us/503 and gets the status code and response time. The code and unit test for this python application are available under pythonprober directory in this github repo.
To execute this, we need to build a docker image using the following command. Make sure that following command is being ran from the directory containing the Dockerfile.
docker build -t pythonprober .
This should build the image with latest tag and store it in your docker repository. Once this image is build, we can use the same container image and deploy that from our kubernetes deployment.
The kubernetes cluster containing multiple components can be deployed using following command. The kubernetes deployment yaml file is also present in the same folder with the name python-prober-deployment.yml. We can run the following command to deploy different components running container image generated in above step.
kubectl apply -f python-prober-deployment.yml
If all goes well, you should see output of above command like this
serviceaccount/ish-pythonprober unchanged service/ish-pythonprober unchanged deployment.apps/ish-pythonprober unchanged ingress.networking.k8s.io/ish-pythonprober created servicemonitor.monitoring.coreos.com/ish-pythonprober unchanged pod/ish-pythonprober-test-connection unchanged
Additionally, we can check our pods using below command. Note that all the pods are deployed in default namespace.
kubectl get pods
NAME READY STATUS RESTARTS AGE ish-pythonprober-796c97db44-n2f2m 1/1 Running 13 77m
The metrics should be visible on prometheus and grafana dashboard for those metrics should look like below. Note that we have used guages in prometheus metrics to track these metrics. The 2 metrics display the response time in milliseconds and the status codes(200 or 503).
