testutils

func GetCurrentAndNextVersionsForUpgrade ¶ added in v0.161.0

func GetCurrentAndNextVersionsForUpgrade(cvm eks.ClusterVersionsManagerInterface, testVersion string) (currentVersion, nextVersion string)

func InProgressUpdates ¶ added in v0.231.0

func InProgressUpdates(ctx context.Context, eksAPI awsapi.EKS, clusterName string) ([]string, error)

InProgressUpdates returns the IDs of any EKS updates on clusterName that are still in progress. An empty result means the cluster is safe to mutate.

func ParseClusterConfig ¶ added in v0.99.0

func ParseClusterConfig(clusterName, region, filename string) *api.ClusterConfig

func Reader ¶

func Reader(clusterConfig *api.ClusterConfig) io.Reader

func ReaderFromFile ¶

func ReaderFromFile(clusterName, region, filename string) io.Reader

func WaitForClusterToSettle ¶ added in v0.231.0

func WaitForClusterToSettle(ctx context.Context, eksAPI awsapi.EKS, clusterName string)

WaitForClusterToSettle blocks until clusterName has no EKS update in flight.

EKS reports a cluster as ACTIVE while asynchronous post-create work -- default addon installs, access-config changes -- is still running. Any mutating call issued during that window is rejected with HTTP 409:

ResourceInUseException: Cannot VersionUpdate because cluster <name> currently has an
update in progress

So a DescribeCluster status of ACTIVE is not a sufficient readiness signal, and neither are eksctl's own CanUpdate/CanOperate guards, which only consult that status. ListUpdates is the signal that actually reflects the condition named in the error.

Call this after creating a cluster and before the first mutating operation a suite performs on it. It is read-only and a no-op on a settled cluster, so it is safe to call speculatively.