Just wanted to to write down the problem I was having and the solution I found to solve it.
I’m running edge-24.7.2 in high-availability into aws eks v1.30.2-eks-db838b0.
After installing linkerd, a simple kubectl expose pod nginx --port=80 was taking around 11 seconds to be created.
After debuging I found that I needed to create an inbound securitygroup allowing tcp/8080 from the eks control plane (managed by aws) to the nodes. After that, the time for creating the k8s service went back to around 1 second.
The closest documentation I found about this: Cluster Configuration | Linkerd, however the port is different.
I don’t have much more time to write the right reason this was happening. It would be great if someone could enrich this topic for further consulting.