I am trying to understand the usage of the ip addresses from my vpc in the eks with fargate environment. I can see that each pod has its own private ip address, which seems to be the same ip address for fargate node as well. It seems that the ENI allocates a single primary address for a EC2 node and many ip address as secondary based on the size, but I cannot find the same information on fargate. Does that mean it does not have any secondary ip addresses allocated ?

Extending on the question, it seems that a network load balancer requires a minimum of 8 free ip addressses to be created, Does that mean it blocks all the 8 ?

CodePudding user response：

...Does that mean it does not have any secondary ip addresses allocated ?

Correct, since you can only run 1 pod on each Fargate instance, while on EC2 node you can run many pods.

...Does that mean it blocks all the 8 ?

The LB controller won't block subnet IP. When you make request to create NLB in a subnet that has insufficient IP, you will see error message like: ..."error":"InvalidSubnet: Not enough IP space available in subnet-.... ELB requires at least 8 free IP addresses in each subnet. Note this is a requirement of ELB and not EKS.