As kube-dns transitioned to fat manifests, it's no longer required to use arch
suffixed images. This change makes use of fat manifests for kube-dns and
removes the last few calls to the GetGenericArchImage function, thus removing
GetGenericArchImage too.
Signed-off-by: Rostislav M. Georgiev <rostislavg@vmware.com>

Restore *filter table for ipvs

Updating CSI e2e test to create CSI CRDs

Update kube-dns container images.

Fix drain for evicting terminal DS pods and pods with local storage

Add coverage profile retrieval to log dumps

This fixes an issue where SRV records were incorrectly being compressed.
Also updated kubedns version for kubeadm
Upgrade to 1.14.12 with manifest support. Runs dnsmasq version 2.78

Avoid setting Masked/ReadOnly paths when pod is privileged

Disable DNS federation test

bazel: update debian-iptables and debian-hyperkube-base

In the recent PR on adding ProcMount, we introduced a regression when
pods are privileged. This shows up in 18.06 docker with kubeadm in the
kube-proxy container.

The kube-proxy container is privilged, but we end up setting the
`/proc/sys` to Read-Only which causes failures when running kube-proxy
as a pod. This shows up as a failure when using sysctl to set various
network things.

Change-Id: Ic61c4c9c961843a4e064e783fab0b54350762a8d

UnmountDevice (GetMountRefs) should not fail if path does not exist

Renew certificates as part of upgrade rather than recreating them

Bump metrics-server to v0.3.1

Upgrade kubeadm's version of docker support

Resolve: #68194

Update to use debian-iptables v10.2 and debian-hyperkube-base 0.10.2

Install netbase in debian-iptables and debian-hyperkube-base as it is needed by ipvs

master/staging/README.md update

GetMountRefs should not fail if the path supplied does not exist anymore. It has no mount references

Update kubeadm v1alpha3 example

Rewrite finalURLTemplate used only for metrics because of dynamic client change

kubeadm: remove unit test TestNewCmdReset

kubeadm: update MinimumControlPlaneVersion to v1.11.0

This fixes an issue with overly aggressive discardings of node data sets
when a single pod is missing data.

Signed-off-by: Chuck Ha <ha.chuck@gmail.com>

Check network condition ready when waiting for ready nodes in e2e tests