Nodejs Healthcheck

A library for adding standard (k8s) healthchecks to nodejs applications. based on nodejs-healthcheck

It exposes 2 endpoints:

1. /health/livez - Returns 200 by default , can optionally include result evaluating all readyz passed in config.
2. /health/readyz - Returns 200 by default , can optionally include result evaluating all readyz passed in config.

Usage

Configure an express.js handler with checks.

const healthcheck = require('@hyperbyte-io/nodejs-livez-readyz');

const config = {
    livez: {
        myComplexWebCheck: healthcheck.web('https://example.com/other', {
            callback: (err, res) => {
                return res.body.status == 'good'
                    ? healthcheck.up()
                    : healthcheck.down();
            },
            timeout: 5000,
            deadline: 10000
        }),
        myRawCheck: healthcheck.raw(() => {
            return myInternalCheck() ? healthcheck.up() : healthcheck.down();
        })
    },
    readyz: {
        mySimpleWebCheck: healthcheck.web('https://example.com/status'),
        myComplexWebCheck: healthcheck.web('https://example.com/other', {
            callback: (err, res) => {
                return res.body.status == 'good'
                    ? healthcheck.up()
                    : healthcheck.down();
            },
            timeout: 5000,
            deadline: 10000
        }),
        myRawCheck: healthcheck.raw(() => {
            return myInternalCheck() ? healthcheck.up() : healthcheck.down();
        })
    }
};
healthcheck.addTo(app, config);

what to include in readiness checks

• On Kubernetes, readiness probes will be called periodically throughout the lifetime of the container. Container will be made temporarily unavailable from serving traffic when the readiness check fails.
• The requests won't even reach your application to handle errors. So, it is very important to consider what checks should be included into readiness probe.
• While adding all dependant services to readiness check can help in identifying any misconfiguration during startup, it could cause unwanted downtime for the application.
• K8s introduced startUp Probes (Alpha in 1.16 ) to handle startup cases separately.

Based on above, you should include a dependency into readiness checks only if they are exclusive/hard dependencies for your service. Unavailability of soft dependencies needs to be handled in code to give appropriate customer experience.

Good example for check to be included in readiness:

• A private cache / database like Redis or Elastic Search which are exclusive to the application (not shared).

Bad example for check to be included in readiness:

• Any shared components like IDAM, S2S or CCD.