3. Observability

Date: 2022-10-28

Status

Accepted

Context

We need to be able to monitor the health of our services and be able to collect data on the metrics we are most interested in a standardised way

Decision

We will add three standard HTTP-endpoints to each service on port 9090.

• /metrics will return a prometheus formatted metrics page
• /healthz will return a json object with a status field that is either ok or error
• /ready will return a json object with a status field that is either ok or error

Dockerfile

We will add command HEALTHCHECK to the dockerfile for each service that will call the /healthz endpoint and check the status.

Helm

Also, we add k8s *-probes to each service that will call the /healthz endpoint and check the status.

For Helm-chart we use complete template:

Consequences

We will be able to monitor the health of our services and collect metrics in a standardised way.