v0.9.1-rc.4

Improvements

• Kubernetes 1.4.6

Fixes

• EFS systemd unit start ordering (#69, thanks to @pieterlange)
• Flanneld startup slowness (#73)

v0.9.1-rc.3

Improvements

• Easily manage etcd cluster with env vars(#58, thanks to @gianrubio)
• Use non-deprecated parameter for pod manifests. (#54, thanks to @pieterlange)

Fixes

• Persist /var/lib/cni between kubelet restarts (#64, thanks to @groner)

Known issues

• flanneld startup possibly slow (#65)

v0.9.1-rc.2

Features:

• Add the --s3-uri s3://<bucket>/<directory flag to kube-aws up, kube-aws update and kube-aws validate to automatically avoid the 51200 bytes limitation errors of CloudFormation (#45)

Improvements

• Allow ECR pull from controller IAM role (#35, thanks to @c-knowles)
• Add name tags where applicable (#36, thanks to @c-knowles)

Fixes

• Remove "Restart=on-failure" from kube-node-drainer.service (which was breaking whole startup process via oem-cloud-init.service, #41, thanks to @camilb)
• Replace docker with rkt in kube-node-drainer.service (#48, thanks to @camilb)

v0.9.1-rc.1

Fixes

• Stabilize kubelet startup (#34) ---> this introduced flanneld startup slowness and fixed in v-0.9.1-rc.4 via #73
• Fix calico to work with secure etcd (3f90573, thanks to @camilb)

Improvements

• Add option to mount existing Amazon EFS volume (#17, thanks to @pieterlange)
• Don't overwrite the CA by default on deprecated 'render' command (#28, thanks to @pieterlange)

Known issues

• oem-cloud-init fails to start, resulting all the worker nodes to not come up when kube-node-drainer.service is enabled(requires manual fix to cloud-init-worker) ---> fixed in rc.2 via #41
• a lot of error logs are generated on kubelet/flanneld startup(the service itself isn't affected)

v0.9.0

Features

• Kubernetes v1.4.5
• Discrete (and HA) etcd cluster (thanks to @colhom)
• HA control plane (thanks to @colhom)
• Cluster upgrades (thanks to @colhom)
• Node draining/cordoning

Improvements

• Allow to override the AMI ID of CoreOS (thanks to @qqshfox)
• Add mapPublicIPs: Simple toggle to stop the automatic assignment of public IP's for all
  the nodes. (thanks to @pieterlange)

Fixes

• Use hosts' DNS resolvers instead of Google (thanks to @pieterlange)
• Add a new var IsChinaRegion to fix for China Region (thanks to @qqshfox)

v0.9.0-rc.4

• Kubernetes v1.4.5

v0.9.0-rc.3

Features

• Add a new var IsChinaRegion to fix for China Region
• Allow to override the AMI ID of CoreOS

v0.9.0-rc.2

Bug fixes

• fix MapPublicIPs in stack template which resulted in rendering errors

v0.9.0-rc.1

• Kubernetes v1.4.3
• Discrete (and HA) etcd cluster
• HA control plane
• Cluster upgrades
• Node draining/cordoning