v0.9.2-rc.4

Fixes

• Fix the issue that the second and the subsequent nodes in each spot fleet couldn't be registered thus unable to run any pods (#141)

Features

• Experimental feature to taint worker nodes with user-provided key=value:effect (#132)
  • This feature is useful for anyone:
    • who wants to implement dedicated nodes
    • who wants to utilize spot fleets but save innocent users from scheduling their pods unintentionally to spot instances
• Add workerSecurityGroupIds to cluster.yaml so that you can attach existing security groups to worker nodes (#127)
• More node-pools sub-commands
  • kube-aws node-pools destroy to destroy a node pool (#139)
  • kube-aws node-pools update to update a node pool (#130, #140)
    • CAUTION: Spot Fleets are not updated one-by-one but replaced thus some downtime may occur
• The following experimental features are now supported for node pools (#128)
  • awsEnvironment
  • waitSignal
  • nodeLabel
  • loadBalancer

Improvements

• Enable ASG metrics (#137, thanks to @c-knowles)
• Update dashboard to v1.5.0 (#133, thanks to @danielfm)
• No more unnecessary full node replacement when kube-aws update is run
  • Persist encrypted credentials under the credentials/ directory (#125, thanks to @pieterlange for reporting the issue)
• Name the subnets after the node pool (#129, thanks to @c-knowles)