Dieser Inhalt ist in der ausgewählten Sprache nicht verfügbar

Wir arbeiten ständig daran, Inhalte in der ausgewählten Sprache bereitzustellen. Vielen Dank für deine Geduld.

How do I troubleshoot issues when creating a Kubernetes service that's backed by a load balancer on Amazon EKS?

Lesedauer: 6 Minute

I can't create a Kubernetes service that's backed by a load balancer on Amazon Elastic Kubernetes Service (Amazon EKS).

Short description

To troubleshoot your service load balancer, verify that you have the following configurations:

The correct tags for your Amazon Virtual Private Cloud (Amazon VPC) subnets
The required AWS Identity and Access Management (IAM) permissions for your cluster's IAM role
A valid Kubernetes service definition
Load balancers that stay within your account limit
Enough free IP addresses on your subnets
A correctly configured load balancer to avoid connection timeout issues
Healthy load balancer targets

If you still have an issue after verifying all the preceding items, then follow the steps in the Try additional troubleshooting steps section.

Resolution

Note:

The following steps apply to the Classic Load Balancer and the Network Load Balancer. For the Application Load Balancer, see Application load balancing on Amazon EKS.
If you receive errors when running AWS Command Line Interface (AWS CLI) commands, make sure that you’re using the most recent AWS CLI version.

Use the correct tags for your Amazon VPC subnets

1. Open the Amazon VPC console.

2. On the navigation pane, choose Subnets.

3. Choose the Tags tab for each subnet, and then confirm that a tag exists. For example:

Key: kubernetes.io/cluster/yourEKSClusterName
Value: shared

4. For your public subnets, confirm that the following tag exists:

Key: kubernetes.io/role/elb
Value: 1

Note: To see if a subnet is public, check the route table that's associated with the subnet. A public subnet has a route to an internet gateway (igw-xxxxxxxxxxx). A private subnet has a route to the internet through a NAT gateway or NAT instance, or no route to the internet at all.

Important: You must have the tag in step 4 to create an internet-facing load balancer service. An internet-facing load balancer routes requests from clients to targets over the internet.

5. For your private subnets, confirm that the following tag exists:

Key: kubernetes.io/role/internal-elb
Value: 1

Important: You must have the tag in step 5 to create an internal load balancer service. An internal load balancer routes requests to targets using private IP addresses.

Set the required IAM permissions for your cluster's IAM role

1. Open the Amazon EKS console.

2. On the navigation pane, choose Clusters.

3. Select your cluster, and then note your Cluster IAM Role ARN.

4. Open the IAM console.

5. On the navigation pane, choose Roles.

6. Select the role that matches the Cluster IAM Role ARN that you identified in step 3.

7. Confirm that the AWS managed policy AmazonEKSClusterPolicy is attached to your role.

Note: The Amazon EKS control plane assumes the preceding IAM role to create a load balancer for your service.

Use a valid Kubernetes service definition

1. In the YAML file for your Kubernetes service, verify that spec.type is set to LoadBalancer.

Here's an example of a Kubernetes service that's backed by a load balancer:

apiVersion: v1
kind: Service
metadata:
  annotations:
    # This annotation is only required if you are creating an internal facing ELB. Remove this annotation to create public facing ELB.
    service.beta.kubernetes.io/aws-load-balancer-internal: "true"
  name: nginx-elb
  labels:
    app: nginx
spec:
  type: LoadBalancer
  ports:
  - name: "http"
    port: 80
    targetPort: 80
  selector:
    app: nginx

Note: To customize your service with a different annotation, see Internal load balancer and TLS support on AWS on the Kubernetes website. If you use the AWS Load Balancer Controller to manage a Network Load Balancer, then your annotations must meet the definition for the Kubernetes service. For more information, see Network load balancing on Amazon EKS.

Verify that your load balancers are within your account limit

An AWS account has a maximum of 20 load balancers per AWS Region by default.

To check how many load balancers that you have, open the Amazon Elastic Compute Cloud (Amazon EC2) console. Then, choose Load Balancers from the navigation pane.

If you reach the maximum number of load balancers, then you can apply for an increase with Service Quotas.

Verify that there are enough free IP addresses on your subnets

To create a load balancer, each subnet of that load balancer must have a minimum of eight free IP addresses. This is required for both the Classic Load Balancer and Network Load Balancer.

Connection timeout issues

If your load balancer is experiencing intermittent or constant timeout issues, then there might be an issue with the load balancer configuration. This can happen when the backend service can’t handle the incoming traffic, or too many requests are sent to the load balancer. For more information, see How do I troubleshoot Elastic Load Balancing "Connection timed out" errors?

Verify load balancer health checks

If you don't correctly configure the endpoint, or the backend service isn't responding to health check requests, then Load balancer health check issues occur.

To verify that the load balancer is pointing to the correct endpoint, run the AWS CLI command describe-target-health similar to the following:

aws elbv2 describe-target-health --target-group-arn arn:aws:elasticloadbalancing:us-west-2:1234567890:targetgroup/my-targets/6d0ecf831eec9f09

To check the health check configuration of the target group, run the AWS CLI command describe-target-groups:

aws elbv2 describe-target-groups --target-group-arns arn:aws:elasticloadbalancing:us-west-2:1234567890:targetgroup/my-targets/6d0ecf831eec9f09

For more information, see Health checks for your target groups.

Try additional troubleshooting steps

To check the Kubernetes service for an error message that can help you troubleshoot the issue, run the following describe-service command:

$ kubectl describe service my-elb-service

If the service is created successfully, then the output is similar to the following:

...
...
Events:
  Type    Reason                Age   From                Message
  ----    ------                ----  ----                -------
  Normal  EnsuringLoadBalancer  47s   service-controller  Ensuring load balancer
  Normal  EnsuredLoadBalancer   44s   service-controller  Ensured load balancer

If the service isn't created successfully, then you receive an error message.

To get more information about error messages, check the following resources: