kubernetes-retired/kube-aws v0.9.4 on GitHub

Features

Kubernetes 1.5.3 (#324, thanks to @gianrubio)
Calico 2.0.2 (#282, thanks to @heschlie)
Bypass 16KB userdata limit (#268, thanks to @redbaron)
Easier to configure, easier to version-control, more tightly integrated node pools
(#315, thanks to @redbaron @sasso @c-knowles for reviewing)
Flexible and customizable network topology (#284) based on all the awesome works:
- #227 "Use existing subnets when creating/updating cluster" (which is followed by #278) for @sasso
  - Deployment to existing subnets for etcd is backed up by "Don't precalculate etcd static IP addresses" (#226 #263, thanks to @redbaron)
- #169 "Create etcd and workers in private subnets, controllers in public subnet" for @neoandroid
kube-aws calculator for estimating cfn stack costs (#314, thanks to @gianrubio)
Add support for ALBs Target Groups (#353, thanks to @gbtechhub)
Add DOCKER_OPTS environment variable to limit docker log files (#323, thanks to @jollinshead)
Encryption support for etcd data volume (#307, thanks to @checkr)
Add experimental support for pod security policies (#339, @soellman)
Experimental webhook token authentication support (#321, thanks to @olesku)
Experimental kube2iam support (#297, thanks to @Fsero)
Experimental ClusterAutoscalerSupport option to setup IAM ASG permissions (#294, thanks to @pwillie)
- The worker nodes require extra privileges for the cluster-autoscaler
  to run successfully within the cluster. Can be optionally
  enabled via boolean in cluster.yaml.

Minimal RBAC policies (#316, thanks to @gianrubio)
credentials: Add /O=systems:master for kube-admin (#345, thanks to @ankon)
Explicitly create EBS volumes for Etcd nodes (#317, thanks to @redbaron)
- Having EBS volumes which survive node termination makes them more manageable
Wait for signals by default (#304, thanks to @gianrubio)
- Enable feature waitSignal as default
- Setup initial value for RollingUpdateMinInstancesInService as 1
- Reuquire s3-flag on validate command
Remove checks for dedicated spot instances (#280, thanks to @iameli)
More robust etcd search and replace step (#292, thanks to @redbaron)
Do not require keyName if sshAuthorizedKeys are set (#242, thanks to @redbaron)
Remove obsolete Calico files (#274, thanks to @redbaron)

Fix missing Name and ControllerHost in status and up commands (#356, thanks to @Thermi, @ankon for reporting)
Fix the inconsistent S3 object prefix issue (#357, thanks to @c-knowles for reporting)
Fix customSettings unavailability for node pools (#358, thanks to @ankon for reporting)
Fix various typos (#329, thanks to @ankon)
Fix a typo in a image reference (#328, thanks to @ankon)
Remove a reference to waitsignal being experimental (#327, thanks to @ankon)
Update readme (#337, thanks to @philips)
Fix typo in comment (#350, thanks to @ankon)
Fix typo in provisioner.go (#354, thanks to @msanterre)
Fix error message when failed to init cluster (#313, thanks to @cheungpat)
kubelet SecurityGroupIngress (#310, thanks to @soellman)
- Add a SecurityGroupIngress for controllers to access the kubelet
  port (10250) on other controllers. This is needed for when there
  are multiple masters.
Fix the bug which was causing a private subnet to have no route to an existing nat gateway (#301)
Fix lint (#250, thanks to @jmcarp and @icereval)
Fix connection drop after 60 sec when executing commands inside containers (#275, thanks to @camilb)