As the TODO in plugin/pkg/scheduler/scheduler.go described:

move everything from pkg/scheduler into this package. Remove
references from registry.

Update cAdvisor godep.

Update CHANGELOG for 0.17.0

Added label kubernetes.io/name to cluster services

Make MIG deletion during GCE kube down blocking, so that subsequent template deletion doesn't fail.

Make the Elasticsearch logging pods discover each other

Handle out of disk situation on kubelet.

- Fixes working set reporting.
- Adds disk Available size.
- Adds support for ps inside containers via cAdvisor.

kubelet/container: Remove ipcMode and netMode from  RunContainerOptions.

Improve kubelet test for env vars

Convert node level logging to Cloud Logging to use token-system-logging

Convert nodel level logging to ES to use token-system-logging

Rename default DNS domain to cluster.local

Since createPodInfraContainer() is moved into dockertools. There is no
need to pass the ipcMode or netMode via RunContainerOptions.

Refactor master vs node kube-env and salt auth

Run Docker Daemon in a Resource-only Container

Switch exec to look at exit code not output status.

Kubelet will stop accepting new pods if it detects low disk space on root fs or fs holding docker images.
Running pods are not affected. low-diskspace-threshold-mb is used to configure the low diskspace threshold.

ServiceIP allocation test is flaky

Inconsistency in api terminology of pod conditions

Enabled load e2e test for gce-scalability

Fix boilerplate check crash if .go file has no comments

Adds Karma to TravisCI

ServiceAccounts

Implementation maintains a thread which ensures that the Docker daemon
is in a container.

This component will manage containers on the machine not managed by the
runtime.

Add zone to jenkins version-getting call.

Duplicate service port name can't be validated

Change default CgroupRoot to /.

Make SyncPodSync as the default SyncPodType.