kOps addons

kOps supports two types of addons:

  • Managed addons, which are configurable through the cluster spec
  • Static addons, which are manifest files that are applied as-is

Managed addons

The following addons are managed by kOps and will be upgraded following the kOps and kubernetes lifecycle, and configured based on your cluster spec. kOps will consider both the configuration of the addon itself as well as what other settings you may have configured where applicable.

Available addons

AWS Load Balancer Controller

Introduced
kOps 1.20

AWS Load Balancer Controller offers additional functionality for provisioning ELBs.

  1. spec:
  2. awsLoadBalancerController:
  3. enabled: true

Though the AWS Load Balancer Controller can integrate the AWS WAF and Shield services with your Application Load Balancers (ALBs), kOps disables those capabilities by default.

Introduced
kOps 1.24

You can enable use of either or both of the WAF and WAF Classic services by including the following fields in the cluster spec:

  1. spec:
  2. awsLoadBalancerController:
  3. enabled: true
  4. enableWAF: true
  5. enableWAFv2: true

Note that the controller will only succeed in associating one WAF with a given ALB at a time, despite it accepting both the “alb.ingress.kubernetes.io/waf-acl-id” and “alb.ingress.kubernetes.io/wafv2-acl-arn” annotations on the same Ingress object.

You can enable use of Shield Advanced by including the following fields in the cluster spec:

  1. spec:
  2. awsLoadBalancerController:
  3. enabled: true
  4. enableShield: true

Support for the WAF and Shield services in kOps is currently beta, meaning that the accepted configuration and the AWS resources involved may change.

Read more in the official documentation.

Cluster autoscaler

Introduced
kOps 1.19

Cluster autoscaler can be enabled to automatically adjust the size of the kubernetes cluster.

  1. spec:
  2. clusterAutoscaler:
  3. enabled: true
  4. expander: least-waste
  5. balanceSimilarNodeGroups: false
  6. awsUseStaticInstanceList: false
  7. scaleDownUtilizationThreshold: 0.5
  8. skipNodesWithLocalStorage: true
  9. skipNodesWithSystemPods: true
  10. newPodScaleUpDelay: 0s
  11. scaleDownDelayAfterAdd: 10m0s
  12. image: <the latest supported image for the specified kubernetes version>
  13. cpuRequest: "100m"
  14. memoryRequest: "300Mi"

Read more about cluster autoscaler in the official documentation.

Expander strategies

Cluster autoscaler supports several different expander strategies.

Note that the priority expander requires additional configuration through a ConfigMap as described in its documentation - you will need to create this ConfigMap in your cluster before selecting this expander.

Disabling cluster autoscaler for a given instance group
Introduced
kOps 1.20

You can disable the autoscaler for a given instance group by adding the following to the instance group spec.

  1. spec:
  2. autoscale: false

Cert-manager

IntroducedMinimum K8s Version
kOps 1.20k8s 1.16

Cert-manager handles x509 certificates for your cluster.

  1. spec:
  2. certManager:
  3. enabled: true
  4. defaultIssuer: yourDefaultIssuer

Warning: cert-manager only supports one installation per cluster. If you are already running cert-manager, you need to either remove this installation prior to enabling this addon, or mark cert-manger as not being managed by kOps (see below). As long as you are using v1 versions of the cert-manager resources, it is safe to remove existing installs and replace it with this addon

Self-provisioned cert-manager
IntroducedMinimum K8s Version
kOps 1.20.2k8s 1.16

The following cert-manager configuration allows provisioning cert-manager externally and allows all dependent plugins to be deployed. Please note that addons might run into errors until cert-manager is deployed.

  1. spec:
  2. certManager:
  3. enabled: true
  4. managed: false
DNS nameserver configuration for cert-manager pod
IntroducedMinimum K8s Version
kOps 1.23.3k8s 1.16

Optional list of DNS nameserver IP addresses for the cert-manager pod to use. This is useful if you have a public and private DNS zone for the same domain to ensure that cert-manager can access ingress, or DNS01 challenge TXT records at all times.

You can set pod DNS nameserver configuration for cert-manager like so:

  1. spec:
  2. certManager:
  3. enabled: true
  4. nameservers:
  5. - 1.1.1.1
  6. - 8.8.8.8

Read more about cert-manager in the official documentation

Karpenter

Introduced
kOps 1.24

The Karpenter addon enables Karpenter-managed InstanceGroups.

  1. spec:
  2. karpenter:
  3. enabled: true

See more details on how to configure Karpenter in the kOps Karpenter docs and the official documentation

Metrics server

Introduced
kOps 1.19

Metrics Server is a scalable, efficient source of container resource metrics for Kubernetes built-in autoscaling pipelines.

  1. spec:
  2. metricsServer:
  3. enabled: true

Read more about Metrics Server in the official documentation.

Secure TLS
Introduced
kOps 1.20

By default, API server will not verify the metrics server TLS certificate. To enable TLS verification, set the following in the cluster spec:

  1. spec:
  2. certManager:
  3. enabled: true
  4. metricsServer:
  5. enabled: true
  6. insecure: false

This requires that cert-manager is installed in the cluster.

Node local DNS cache

IntroducedMinimum K8s Version
kOps 1.18k8s 1.15

NodeLocal DNSCache can be enabled if you are using CoreDNS. It is used to improve the Cluster DNS performance by running a dns caching agent on cluster nodes as a DaemonSet.

memoryRequest and cpuRequest for the node-local-dns pods can also be configured. If not set, they will be configured by default to 5Mi and 25m respectively.

If forwardToKubeDNS is enabled, kubedns will be used as a default upstream

  1. spec:
  2. kubeDNS:
  3. provider: CoreDNS
  4. nodeLocalDNS:
  5. enabled: true
  6. memoryRequest: 5Mi
  7. cpuRequest: 25m

Node termination handler

Introduced
kOps 1.19

Node Termination Handler ensures that the Kubernetes control plane responds appropriately to events that can cause your EC2 instance to become unavailable, such as EC2 maintenance events, EC2 Spot interruptions, and EC2 instance rebalance recommendations. If not handled, your application code may not stop gracefully, take longer to recover full availability, or accidentally schedule work to nodes that are going down.

  1. spec:
  2. nodeTerminationHandler:
  3. cpuRequest: 200m
  4. enabled: true
  5. enableSQSTerminationDraining: true
  6. managedASGTag: "aws-node-termination-handler/managed"
Queue Processor Mode
Introduced
kOps 1.21

If enableSQSTerminationDraining is true Node Termination Handler will operate in Queue Processor mode. In addition to the events mentioned above, Queue Processor mode allows Node Termination Handler to take care of ASG Scale-In, AZ-Rebalance, Unhealthy Instances, EC2 Instance Termination via the API or Console, and more. kOps will provision the necessary infrastructure: an SQS queue, EventBridge rules, and ASG Lifecycle hooks. managedASGTag can be configured with Queue Processor mode to distinguish resource ownership between multiple clusters.

The kOps CLI requires additional IAM permissions to manage the requisite EventBridge rules and SQS queue:

  1. {
  2. "Version": "2012-10-17",
  3. "Statement": [
  4. {
  5. "Effect": "Allow",
  6. "Action": [
  7. "events:DeleteRule",
  8. "events:ListRules",
  9. "events:ListTargetsByRule",
  10. "events:ListTagsForResource",
  11. "events:PutEvents",
  12. "events:PutRule",
  13. "events:PutTargets",
  14. "events:RemoveTargets",
  15. "events:TagResource",
  16. "sqs:CreateQueue",
  17. "sqs:DeleteQueue",
  18. "sqs:GetQueueAttributes",
  19. "sqs:ListQueues",
  20. "sqs:ListQueueTags"
  21. ],
  22. "Resource": "*"
  23. }
  24. ]
  25. }

Warning: If you switch between the two operating modes on an existing cluster, the old resources have to be manually deleted. For IMDS to Queue Processor, this means deleting the k8s nth daemonset. For Queue Processor to IMDS, this means deleting the Kubernetes NTH deployment and the AWS resources: the SQS queue, EventBridge rules, and ASG Lifecycle hooks.

Node Problem Detector

Introduced
kOps 1.22

Node Problem Detector aims to make various node problems visible to the upstream layers in the cluster management stack. It is a daemon that runs on each node, detects node problems and reports them to apiserver.

  1. spec:
  2. nodeProblemDetector:
  3. enabled: true
  4. memoryRequest: 32Mi
  5. cpuRequest: 10m

Pod Identity Webhook

Introduced
kOps 1.23

When using IAM roles for Service Accounts (IRSA), Pods require an additinal token to authenticate with the AWS API. In addition, the SDK requires specific environment variables set to make use of these tokens. This addon will mutate Pods configured to use IRSA so that users do not need to do this themselves.

All ServiceAccounts configured with AWS privileges in the Cluster spec will automatically be mutated to assume the configured role.

  1. spec:
  2. certManager:
  3. enabled: true
  4. podIdentityWebhook:
  5. enabled: true

The EKS annotations on ServiceAccounts are typically not necessary as kOps will configure the webhook with all ServiceAccount to role mapping configured in the Cluster spec. But if you need specific configuration, you may annotate the ServiceAccount, overriding the kOps configuration.

Read more about Pod Identity Webhook in the official documentation.

Snapshot controller

IntroducedMinimum K8s Version
kOps 1.21k8s 1.20

Snapshot controller implements the volume snapshot features of the Container Storage Interface (CSI).

You can enable the snapshot controller by adding the following to the cluster spec:

  1. spec:
  2. snapshotController:
  3. enabled: true

Note that the in-tree volume drivers do not support this feature. If you are running a cluster on AWS, you can enable the EBS CSI driver by adding the following:

  1. spec:
  2. cloudConfig:
  3. awsEBSCSIDriver:
  4. enabled: true
Self-managed aws-ebs-csi-driver
Introduced
kOps 1.25

The following configuration allows for a self-managed aws-ebs-csi-driver. Please note that if you’re using Amazon EBS volumes, you must install the Amazon EBS CSI driver. If the Amazon EBS CSI plugin is not installed, then volume operations will fail.

If IRSA is not enabled, the control plane will have the permissions to provision nodes, and the self-managed controllers should run on the control plane. If IRSA is enabled, kOps will create the respective AWS IAM Role, assign the policy, and establish a trust relationship allowing the ServiceAccount to assume the IAM Role. To configure Pods to assume the given IAM roles, enable the Pod Identity Webhook. Without this webhook, you need to modify your Pod specs yourself for your Pod to assume the defined roles.

  1. spec:
  2. cloudConfig:
  3. awsEBSCSIDriver:
  4. enabled: true
  5. managed: false

Custom addons

The command kops create cluster does not support specifying addons to be added to the cluster when it is created. Instead they can be added after cluster creation using kubectl. Alternatively when creating a cluster from a yaml manifest, addons can be specified using spec.addons.

  1. spec:
  2. addons:
  3. - manifest: s3://my-kops-addons/addon.yaml

The docs about the addon management describe in more detail how to define a addon resource with regards to versioning. Here is a minimal example of an addon manifest that would install two different addons.

  1. kind: Addons
  2. metadata:
  3. name: example
  4. spec:
  5. addons:
  6. - name: foo.addons.org.io
  7. version: 0.0.1
  8. selector:
  9. k8s-addon: foo.addons.org.io
  10. manifest: foo.addons.org.io/v0.0.1.yaml
  11. - name: bar.addons.org.io
  12. version: 0.0.1
  13. selector:
  14. k8s-addon: bar.addons.org.io
  15. manifest: bar.addons.org.io/v0.0.1.yaml

In this example the folder structure should look like this;

  1. addon.yaml
  2. foo.addons.org.io
  3. v0.0.1.yaml
  4. bar.addons.org.io
  5. v0.0.1.yaml

The yaml files in the foo/bar folders can be any kubernetes resource. Typically this file structure would be pushed to S3 or another of the supported backends and then referenced as above in spec.addons. In order for master nodes to be able to access the S3 bucket containing the addon manifests, one might have to add additional iam policies to the master nodes using spec.additionalPolicies, like so:

  1. spec:
  2. additionalPolicies:
  3. master: |
  4. [
  5. {
  6. "Effect": "Allow",
  7. "Action": [
  8. "s3:GetObject"
  9. ],
  10. "Resource": ["arn:aws:s3:::my-kops-addons/*"]
  11. },
  12. {
  13. "Effect": "Allow",
  14. "Action": [
  15. "s3:GetBucketLocation",
  16. "s3:ListBucket"
  17. ],
  18. "Resource": ["arn:aws:s3:::my-kops-addons"]
  19. }
  20. ]

The masters will poll for changes in the bucket and keep the addons up to date.