Change "time/Time" to "util.Time", which is what we use in our API for time

Re-enable PATCH integration test for v1beta3

Add a docker image for hyperkube, and instructions on how to use.

Prevent docker load.

Ubuntu-cluster updates: configure.sh fix and guide update

Remove "-f Dockerfile.blah" from make release

etcd should be started after the trap func set

Fix the kubelet node watch

Rework the parallel docker build path to create separate docker build
directories for each binary, rather than just separate files,
eliminating the use of "-f Dockerfile.foo". (I think this also shaves
a little more time off, because it was previously sending the whole
dir each time to the docker daemon.)

Also some minor cleanup.

Fixes #6463

Oops, let's be careful about licensing.

First hack at upgrade script for GCE

Do not log "event already exists" errors

Make gofmt message precise  (-s option)

Fix flakey integration test due to slow endpoints controller

fix broken tables in docs/resouce_quota_admin.md

Updating cadvisor deps. 

PersistentVolume & PersistentVolumeClaim Client

Using 10.0.0.0/16 for flannel network and 10.10.103.x for example cluster network is a bit misleading as people may think they should use the same network for both flannel and eth0 interfaces. This commit tries to clarify that flannel shall use different network.

Without this fix kube-proxy on minions cannot connect to master as `MASTER_IP` is not reachable.

Address #6075: Shoot the master VM while saving the master-pd. This
takes a couple of minor changes to configure-vm.sh, some of which also
would be necessary for reboot. In particular, I changed it so that the
kube-token instance metadata is no longer required after inception;
instead, we mount the master-pd and see if we've already created the
known tokens file before blocking on the instance metadata.

Also partially addresses #6099 in bash by refactoring the kube-push
path.

Remove use of events from NodeController, to temporarily fix #6199

When the server rejects an event because it has already been created, log it
at a very high level (debug) instead of the default level. Duplicate events
typically only occur due to programmer error or failure conditions, so they
can safely be ignored in production environments.

Return a typed error for config validation, and make errors simple