eks

func ConfigureKubectlForEks ¶

func ConfigureKubectlForEks(
	eksCluster *eks.Cluster,
	kubectlOptions *kubectl.KubectlOptions,
) error

ConfigureKubectlForEks adds a new context to the kubeconfig located at the given path that can authenticate with the EKS cluster referenced by the given ARN.

func GetAsgByName ¶

func GetAsgByName(svc *autoscaling.AutoScaling, asgName string) (*autoscaling.Group, error)

GetAsgByName will lookup an AutoScalingGroup that matches the given name. This will return an error if it can not find any ASG that matches the given name.

func GetClusterByArn ¶

func GetClusterByArn(eksClusterArn string) (*eks.Cluster, error)

GetClusterByArn returns the EKS Cluster object that corresponds to the given ARN.

func GetClusterNameFromArn ¶

func GetClusterNameFromArn(eksClusterArnString string) (string, error)

GetClusterNameFromArn extracts the EKS cluster name given the ARN for the cluster.

func GetRegionFromArn ¶

func GetRegionFromArn(eksClusterArnString string) (string, error)

GetRegionFromArn extracts the AWS region that the EKS cluster is in from the ARN of the EKS cluster.

func NewAuthenticatedSession ¶

func NewAuthenticatedSession(region string) (*session.Session, error)

NewAuthenticatedSession gets an AWS Session, checking that the user has credentials properly configured in their environment.

func NewEksClient ¶

func NewEksClient(region string) (*eks.EKS, error)

NewEksClient creates an EKS client.

func RollOutDeployment ¶

func RollOutDeployment(
	region string,
	eksAsgName string,
	kubectlOptions *kubectl.KubectlOptions,
	drainTimeout time.Duration,
	maxRetries int,
	sleepBetweenRetries time.Duration,
) error

RollOutDeployment will perform a zero downtime roll out of the current launch configuration associated with the provided ASG in the provided EKS cluster. This is accomplished by:

1. Double the desired capacity of the Auto Scaling Group that powers the EKS Cluster. This will launch new EKS workers with the new launch configuration.
2. Wait for the new nodes to be ready for Pod scheduling in Kubernetes.
3. Drain the pods scheduled on the old EKS workers (using the equivalent of "kubectl drain"), so that they will be rescheduled on the new EKS workers.
4. Wait for all the pods to migrate off of the old EKS workers.
5. Set the desired capacity down to the original value and remove the old EKS workers from the ASG.

TODO feature request: Break up into stages/checkpoints, and store state along the way so that command can pick up from a stage if something bad happens.

func VerifyCluster ¶

func VerifyCluster(
	eksClusterArn string,
	waitForCluster bool,
	waitMaxRetries int,
	waitSleepBetweenRetries time.Duration,
) error

VerifyCluster verifies that the cluster exists, and that the Kubernetes api server is up and accepting traffic. If waitForCluster is true, this command will wait for each stage to reach the true state.