Kubernetes executor in cluster auto discovery fails on GKE with kubernetes 1.3.6

I've built the stable-1.6 branch and deployed the runner like described in https://gitlab.com/gitlab-org/gitlab-ci-multi-runner/blob/master/docs/install/kubernetes.md

When I trigger a build, the preparation fails with the following message (with --debug)

WARNING: Preparation failed: error connecting to Kubernetes: invalid configuration: no configuration has been provided
..
ERROR: Build failed (system failure): error connecting to Kubernetes: invalid configuration: no configuration has been provided

I did not provide any Kubernetes API connection options so that should be managed by auto-discovery. The "default-token" volume is correctly mounted into the pod:

➔ kubectl exec -it gitlab-runner-3250438794-gbvtb -- find /var/run/secrets/kubernetes.io/serviceaccount
/var/run/secrets/kubernetes.io/serviceaccount
/var/run/secrets/kubernetes.io/serviceaccount/token
/var/run/secrets/kubernetes.io/serviceaccount/namespace
/var/run/secrets/kubernetes.io/serviceaccount/ca.crt

I'm happy to provide more details or do some tests.