Kubernetes-ingress: Health status

Created on 19 Sep 2017 · 8Comments · Source: nginxinc/kubernetes-ingress

90 and #100 add support for health check but I cannot see any instructions on how to use this. After some digging around the code, it looks like I would have to build an image on my own and set some flag to enable that. Is there a way that this could be enabled by annotations?

Without this health endpoint, it's hard to setup zero-downtime deployments since we cannot setup readiness and liveness probes. Is there any other workaround?

documentation

Source

maticko

Most helpful comment

@mkozjak for now, please use the example above. although, note that a few additional command-line arguments are required for running the IC. Please see https://github.com/nginxinc/kubernetes-ingress/blob/master/deployments/deployment/nginx-ingress.yaml#L35 and https://github.com/nginxinc/kubernetes-ingress/blob/master/docs/cli-arguments.md

for health checking by Kubernetes, we'll be adding an additional endpoint handled by the IC to use in readiness and liveliness probes, and update the manifests and helm chart accordingly.

pleshakov on 3 Dec 2018

👍2

All 8 comments

@maticko
Please add the argument -health-status true in the Ingress controller yaml. When enabled, you can check NGINX health via <pod-ip>/nginx-health URL.

pleshakov on 19 Sep 2017

@pleshakov Thanks for this.
Pasting my configuration if someone will have the same trouble.

    spec:
      containers:
      - image: nginxdemos/nginx-ingress:0.9.0-alpine
        imagePullPolicy: Always
        name: nginx-ingress
        ports:
        - containerPort: 80
          hostPort: 80
        - containerPort: 443
          hostPort: 443
        readinessProbe:
          httpGet:
            path: /nginx-health
            port: 80
          initialDelaySeconds: 10
          timeoutSeconds: 1
          periodSeconds: 10
        livenessProbe:
          httpGet:
            path: /nginx-health
            port: 80
          initialDelaySeconds: 10
          timeoutSeconds: 1
          periodSeconds: 10
        args:
          - -health-status=true
        resources:
          requests:
            cpu: 100m
            memory: 128Mi
          limits:
            cpu: 200m
            memory: 256Mi

If you tell me where I can update examples documentation...

maticko on 21 Sep 2017

🎉2

@maticko
thanks for sharing the example!
We'll add a doc page about command-line arguments and include a health-checking example.

pleshakov on 22 Sep 2017

@pleshakov Is this resolved, maybe? Interested in some docs about this..

mkozjak on 30 Nov 2018

for health checking by Kubernetes, we'll be adding an additional endpoint handled by the IC to use in readiness and liveliness probes, and update the manifests and helm chart accordingly.

pleshakov on 3 Dec 2018

👍2

Hi @pleshakov We also turned on the Nginx white-list to only whitelist the IP address initiating from our API gateway. However, the health probe is initiating from traffic manager (DNS).

Will this health check URL always be allowed to pass through?