refactor the way healtcheck is added

[tjamet]

Goal

Ease tests and increase code readability

Committer details

Local-Branch: HEAD

Related changes

Future changes

refactor ingress backend pod detection (#11)

Move the detection at the ingress level rather than per rule

Goals

• Prepare the case to support multiple ingress controllers through CRDs
• Increase code readability and testeability

Prevent changing dnsEndpoint name and namespace attributes in createOrUpdate (#12)

This is not permitted by controllerRuntime and should not be performed

Refactor DNS name extraction (#13)

Goal ---

Ease the implementation of CRD based multi-ingress controller traffic routing

Add fine-grain ingress DNS control through CRD (#14)

Context ---

Handling usual cluster operations, we often come to operate higher risk changes like bumping ingress controller
versions, changing the underneath ingress service type (for example moving from an AWS ELB to an AWS NLB).

Doing so, the safest way would be to be able to provision a new ingress controller and progressively migrating traffic
to the new instance.

Problems

Currently, traffic controller reads the host and load balancer reference using
the ingress status.

This prevents from being able to handle and control weighted records across the different ingress controllers.

Goal

Enable fine-grained routing between various ingress controllers in the same cluster.

Unblocked use-cases

• Progressively change and test the ingress infrastructure (load balancer, ...) and versions
• Allow sharding ingress controllers at the DNS level