Getting an error when running it directly:

Starting etcd
!!! Timed out waiting for http://localhost:4001/v2/keys/
!!! Error in /home/vagrant/go/src/github.com/GoogleCloudPlatform/kubernetes/hack/lib/util.sh:37
  'return 1' exited with status 1
Call stack:
  1: /home/vagrant/go/src/github.com/GoogleCloudPlatform/kubernetes/hack/lib/util.sh:37 kube::util::start_etcd(...)
  2: hack/local-up-cluster.sh:37 main(...)
Exiting with status 1

Probably this - etcd running on 127.0.0.1, not localhost:

etcd -name test -data-dir /tmp/test-etcd.D8OC00 -bind-addr 127.0.0.1:4001

Fixed the last by making wait_for_url use ${host}:${port}.

Seeing:

!!! Timed out waiting for http://127.0.0.1:8080/api/v1beta1/pods

I'm not sure, but that could be --public_address_override being unset.

It wasn't, it's $GO_OUT not being correct anymore. GO_OUT should be

GO_OUT=$(dirname $0)/../_output/local/bin/linux/amd64

Interesting... now i'm wondering why it works on my machine, the binaries aren't there anymore

@smarterclayton Let me know if it works... thx

With those two fixes in place everything works for me. Try a make clean, you might have caught the end of Joe's refactor?

Yeah, that's possible; if the new build process doesn't wipe out the old directory. I just rm -rf _output, so i'm unable to reproduce...

The PR was updated. PTAL

The local setup guide says to install the newest etcd which now is pointed to the etcd v0.5.0-alpha.1 .But this version isn't compatible with the newest kubernetes. So the local cluster will fail.
The solution is to use etcd v0.4.6 now.

The local setup document is probabaly outdated about this, one should use hack/install_etcd.sh to install etcd.
https://github.com/GoogleCloudPlatform/kubernetes/blob/master/docs/devel/development.md#integration-tests

yeah, the install_etcd.sh is now in hack/travis/ directory