Non leader instances enable readiness probe to align with other operators

[randmonkey]

Problem Statement

Non-leader instances returns false on readiness probes. If a deployment has multiple instances, it can never be seen as "ready" in some CI/CD systems (like ArgoCD) . Since the validation webhook is migrated to CELs in KGO 1.5, there are no exposed services that requires the traffic to be only sent to the leader, we can enable the readiness probed of non-leader instances.

Proposed Solution

Start the readiness probe service on both leader and non-leader instances.
...

Additional Information

...

Acceptance Criteria

• Leader and non-leader instances are all in ready state after started

[jaskeerat789]

Could I help you with this issue? I didn't see any contributing section in readme so was not sure if kong accepts contributions from outside

[pmalek]

Hi @jaskeerat789 👋

This is a consequence of the operator setting up the webhook resources and blocking the readiness based on the webhook server being ready (we don't want to mark ourselves ready before the webhook can serve requests as that would cause validation requests to fail).

It's a broader topic that we're discussing now with the team.

I didn't see any contributing section in readme so was not sure if kong accepts contributions from outside

We do accept contributions but not all repos are ready to accept them (given the way our CI is set up). KGO currently relies on some secrets in its CI that prevents the outside contributions from passing the pipeline. We're working on making that work but no ETA yet.

[jaskeerat789]

Hey @pmalek, thanks for the clarity.

[pmalek]

Will be solved by #1100

[pmalek]

Resolved by removing webhook in #1100