I needed a quick way of checking the status of the different applications running in my cluster. I wanted to have a very easy and simple way of adding/removing health checks and I didn't want to reinvent the wheel, so I've used the Kubernetes CronJobs to schedule and launch the different application checks.
The frontend consumes a REST api that queries the Job objects in the API Server and returns the last 20 checks.
By adding/removing labels in the jobs we add/remove applications into the Status Page application.
For example, the following CronJob will check every 5 minutes if ElasticSearch is responding
apiVersion: batch/v1beta1
kind: CronJob
metadata:
name: elastic-web-check
labels:
run: "website-responding"
app: elastic-check-web
group: ElasticSearch
spec:
concurrencyPolicy: Forbid
schedule: "*/5 * * * *"
successfulJobsHistoryLimit: 20
failedJobsHistoryLimit: 20
jobTemplate:
spec:
backoffLimit: 0
activeDeadlineSeconds: 300 # 5min
template:
metadata:
labels:
run: "website-responding"
app: elastic-check-web
group: ElasticSearch
spec:
containers:
- name: curl
image: appropriate/curl
command: ["/bin/sh"]
args: ["-c", "curl -s -o /dev/null -I -w \"%{http_code}\" http://mon-es-elasticsearch-client.monitor.svc.cluster.local:9200"]
restartPolicy: Never
by runnning these jobs, the Status application will pick it up and display it. Here there's a screenshot illustrating the result:
