NC-AIO on k8s using ingress-nginx

[Claiyc]

So... Hi there. Either I'm just hopelessly confused or the healthchecks for the aio-apache deployment don't make much sense.

I want to achieve something quite simple;

1. user sends request
2. first nginx proxy accepts, handles tls termination, routes to k8s cluster
3. ingress-nginx-controller on k8s accepts, routes to aio-apache-service according to ingress configuration
4. aio-apache-service routes to aio-apache-pod

So far so good...
HOWEVER; The https://github.com/nextcloud/all-in-one/blob/main/Containers/apache/healthcheck.sh checks whether exactly that route described above is available and working before marking the container/pod as ready. Therefore step 4 can never succeed because in order for the aio-apache-service to be able to route to the aio-apache-pod, the pod must be ready.
Am I missing something?
Why is this check done at all? Why does the aio-apache container need to connect to itself via the public domain? (for better context; my k8s cluster is running in a separate network segment with outbound controlled traffic. There should never be the scenario where the cluster sends requests to itself via the same interface that users from the outside world use, all of such communication should be handled internally)

Edit:
Clarification, I'm using the provided helm chart.
By now I have discussed this with another devops and we both agree that this healthcheck is not compatible with k8s best practices and creates a chicken-egg problem

[Claiyc]

And yes, after disabling the probes of the aio-apache deployment, everything works as expected. This should be refactored. I will create an issue from this.