When job.spec.completions is nil, only
one task needs to succeed for the job to succeed,
and parallelism can be scaled freely during runtime.

Added tests.

Release Note:

This causes two minor changes to the API.

First, unset parallelism previously was defaulted to be
equal to completions.  Now it always defaults to 1 if unset.

Second, having parallelism=N and completions unset would previously
be defaulted to 1 completion and N parallelism.
(this is not something we expect people to do, though)
Now, no defaulting occurs in that case, and the job's
behavior is different (any completion causes success).

build: Use Go version in cross image tag

Add ECR credential provider

Adding an example for an independent apiserver

The new tag format is cross-<go version>-<cross version>, starting
with cross-1.5.3-1.

Also bump the image we pre-pull / warn-on to 1.5.3

Fixes #19990

Remove monitor_resources test

test: make ValidateCount throw an error instead of a log

AWS: Use a strongly typed mountDevice

Indexed job

kubectl: Make scaling smarter

Provide a RetryOnConflict helper for client libraries

Mention some new api groups.

Promote ServiceAccount e2e out of [Flaky] status

Up to golang 1.5.3

Proposal for disk accounting

e2e kubectl proxy test: Make future flake diagnosis possible

Fixed syntax issue in Heapster config

Bump to container-vm-v20151215

fix heapster rc file jinja syntax error

Fix for panic when instance not found

Merge registered and latest and move to apimachinery

It's currently unclear if the request failed because the proxy wasn't
quite up yet, or if it failed because traffic couldn't get from the
proxy to apiserver. Getting the proxy logs unconditionally should help
narrow this down.

This removes a panic I mistakenly introduced when an instance is not
found, and also restores the exact prior behaviour for
getInstanceByName, where it returns cloudprovider.InstanceNotFound when
the instance is not found.

Was marked as flaky prior to #11291, 6 months ago.
Code has changed many times.
No flakes in last 28 runs, and no Jenkins
history before that.
Follow up in #19024 if you see a flake.

Use Allocatable to replace Capacity

Add WrappedRoundTripper methods to round trippers

Fix job status conditions bloat

Support rolling update to 0 desired replicas

Trigger syncClaim after PV provisioning to reduce wait

AWS: Create a cloudwatch alarm to reboot the master on failure

Install protoc into kube-cross Docker image

Export transport constructors