FluxCD - resources with references are always configured

[gecube]

Describe the bug

If we create a resource that utilises ref from another resource like

apiVersion: ec2.services.k8s.aws/v1alpha1
kind: RouteTable
metadata:
  name: production-private-route-table-eu-west-2c
  namespace: infra-production
spec:
  vpcRef:
    from:
      name: production
  routes:
    - destinationCIDRBlock: 0.0.0.0/0
      natGatewayRef:
        from:
          name: natgateway-eu-west-2c
    - destinationCIDRBlock: 10.0.0.0/16
      vpcPeeringConnectionID: pcx-0a7197b4f5ced6f01

then the target resource looks like:

spec:
  routes:
    - destinationCIDRBlock: 10.0.0.0/16
      vpcPeeringConnectionID: pcx-0a7197b4f5ced6f01
    - destinationCIDRBlock: 0.0.0.0/0
      natGatewayID: nat-07a301987eaa97785
  tags:
    - key: services.k8s.aws/namespace
      value: infra-production
    - key: services.k8s.aws/controller-version
      value: ec2-1.0.3
  vpcRef:
    from:
      name: production

So we can clearly see that natGatewayID was substituted. So it means that on every reconciliation by flux resource is changed twice. Also it is curious that vpcRef is not substituted.

Steps to reproduce

Just apply the first manifest

Expected outcome

No idea. Needs to be discussed. Probably - make substitutions in status field. Or use admission controller. No idea. The only thing I can propose - ask not to change the original description, otherwise I will need to find a way to remove a fields from reconciliation.

[gecube]

linked to #1898

[a-hilaly]

This could be very likely a bug with ACK references.. we'll have to investigate this further. Thank you for reporting this @gecube !

[ack-bot]

Issues go stale after 180d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 60d of inactivity and eventually close.
If this issue is safe to close now please do so with /close.
Provide feedback via https://github.com/aws-controllers-k8s/community.
/lifecycle stale

[gecube]

/remove-lifecycle stale

[michaelhtm]

Hello @gecube, I have made a fix to the EC2 controller.
The issue was happening due to an sdkFind in the customUpdate
function, that overwrites the spec, which is not expected. I will take
a look at the EKS issues as well, in case it's a similar issue.
Thanks for bringing this to our attention and feel free to reach
out for any other questions or concerns

[gecube]

@michaelhtm Hi! Thanks! Sound great. When could we check it on production (we need new release of ACK)?

[michaelhtm]

@gecube We shipped a release for EC2, the EKS release will be coming soon