TLS Passthrough Issue: POD IP Not Updating in VPC Lattice Registered Targets

[infinitydon]

Hi,

Currently having an issue while trying out the TLS passthrough feature, the POD IP is not getting updated in the VPC Lattice Registered targets when there is a change in the POD IP.

An example of when this happens is if the deployment is scaled down and up.

• Status before scaling the deployment to zero

kubectl -n core5g get po -o wide
NAME                                       READY   STATUS    RESTARTS   AGE    IP           NODE                                        NOMINATED NODE   READINESS GATES
telco-5g-nrf-deployment-5dbbdb97df-r9c7w   1/1     Running   0          114s   10.0.31.102   ip-10-0-28-105.us-west-2.compute.internal   <none>           <none>

• Status after scaling the deployment to one:

kubectl -n core5g get po -o wide
NAME                                       READY   STATUS    RESTARTS   AGE   IP            NODE                                        NOMINATED NODE   READINESS GATES
telco-5g-nrf-deployment-5dbbdb97df-x7rgf   1/1     Running   0          5s    10.0.23.231   ip-10-0-28-105.us-west-2.compute.internal   <none>           <none>

But when I check the Registered targets, the IP remains 10.0.31.102 instead of reflecting the new POD IP 10.0.23.231.

Expectation is that Registered target IP should always reflect the active POD IP in the EKS cluster.

[mihaelabalas84]

We faced the exact same issue in our tests.

Thank you for bringing this to our attention. We are currently investigating the root cause.

[infinitydon]

@rlymbur -- Thanks for the update, is this resolved yet? I can see some PRs that are related to this that have been merged.

@infinitydon I have validated this is corrected in the latest release and I have added a new integration test to ensure it does not break again in the future.

[infinitydon]

Tested with release v1.1.2, POD IPs are now updating.