Expose Kubernetes Services Running on Amazon EKS (9 Ways)

[@AntonPutra]

🔴 - To support my channel, I’d like to offer Mentorship/On-the-Job Support/Consulting - [e-mail]

[@bernardoteisceira975]

Awesome explanation. I'm setting up a EKS Cluster in my personal AWS account to learn k8s via aws. I cannot tell you how much this video helped me understand the different ways of exposing k8s services. Also, i'm watching your k8s setup guides/videos.
Thank you very much for your work!! It helps a lot
Regards from Argentina!!

[@AntonPutra]

thanks for the feedback! glad that it helped

[@pi3ni0]

I am really impressed! Your videos are super helpful! Keep going Anton!

[@yassinekhouaja1257]

Once again, your videos proved to be incredibly valuable and beneficial! 👌

[@AntonPutra]

Thanks!

[@pronoe]

I've used kubernetes for a few years in the past and I'm just getting back into it using EKS. I was confused by the AWS LB Controller requiring one LB per ingress which was not how I remember ingresses worked. Your video was super helpful to jog my memory.

[@AntonPutra]

thanks! i still prefer to use the aws lb controller to create load balancers and not ingresses

[@meme22319]

This video is to awsome to be true. Really simple explanation with examples, 30 minutes video could take at least few days to create samples and show it. Many thanks

[@csmithDevCove]

There I was using EKSCTL get spin up a eks cluster and attempting to get it connected to the outside world. I came across your video to do this in Terraform (lesson 133 very useful). Then i come across this video which it looks like you have improved the terraform side of things.
soo much help. Mr Putra, you rock.

[@AntonPutra]

Thanks Chris, if you have anything other besides EKS, eksctl will only complicate your setup.

[@csmithDevCove]

@@AntonPutra I agree. It was nice to setup a cluster and "simply get started" using EKSCTL. However, to your point, realized I needed Terraform to handle other infrastructure needs. Sooo.... why not just use terraform for everything. Cheers.

[@pocketdimension_]

Awesome video, as always 👌.

[@AntonPutra]

Thank you so much 😀

[@George-mk7lp]

great stuff as always. thanks for your hard work!

[@AntonPutra]

Thanks!

[@sujithsurendran7686]

Awesome. Thank you so much for taking time. Could you please share a video on opentelemetry, opentelemetry collector and observability.

[@AntonPutra]

Thanks Sujith! Will do in the future. I used it in one of the previous tutorials to collect metrics. Maybe it can be helpful for you - github.com/antonputra/tutorials/blob/main/lessons/151/go-app/cmd/grpc-server/main.go#L12-L16

[@madhank93]

Another awesome content. Thanks you so much.

[@AntonPutra]

Thanks Madhan!

[@agun21st]

very useful video about EKS expose outside world. 😇

[@AntonPutra]

Thanks :)

[@papubhat]

you did a great job

[@AntonPutra]

Thanks!

[@AndresGorostidi]

Thks for your video, it was very useful. But the way, I have a specific case where I would like your suggestion. I have a monolithic traditional server (HCL Domino) that is using a specific UDP port, UDP 1352. All the clients deployed externally will try to connect using that port, and I can not change the port to be used by the clients. Since NodePort will only expose ports starting at 3000, I can not use NodePort. I can not either use traditional Ingress controllers, since the traffic is not HTTP specific (as said, is using UDP). I am using EKS. Any suggestion on that approach to follow to handle this kind of application?

[@AntonPutra]

Sounds like you have at least two options:
1. Simply use the LoadBalancer service type.
2. Use Nginx ingress controller and the "UDP service". This allows Nginx to proxy custom TCP/UDP protocols using the same load balancer. - kubernetes.github.io/ingress-nginx/user-guide/exposing-tcp-udp-services/

[@rehantayyab82]

for production eks cluster setup worker nodes should be in private subnets ..... plz reply

[@AntonPutra]

yes, there is no need to place worker noods in public subnets only for nodeports and other edge cases

[@TahaYassineJguirim]

@ Host network : what if we make a function lambda to update IPs ?
data "aws_instance" "node1" {
filter {
name = "tag:Name"
values = ["name_node1"]
}
}
data "aws_instance" "node2" {
filter {
name = "tag:Name"
values = ["name_node2"]
}
}
resource "aws_route53_record" "web" {
// ...
records = [data.aws_instance.node1.public_ip, data.aws_instance.node2.public_ip]
}

[@kyles5559]

Hi Anton, how are you managing CRD updates using helm_release?

[@AntonPutra]

I only deploy helm charts with terraform for ephemeral environments or for demo/poc. Terraform does not not support drift detection in general. For production use cases, I only use kustomize with gitops tools such as argocd or helm.