Also related, there is a case in e2e.sh for kubernetes-e2e-aws-parallel, but no corresponding Jenkins job.

Ping on this. Cluster up is succeeding now, but the e2e test driver is timing out waiting for kube-system pods to start running.

any update on this?

It looks like both aws and aws-1.1 are running tests, with one failing consistently on aws and several failing on aws-1.1.

There is one test failing now, same place 3x in a row. @brendandburns you might be interested.

[BeforeEach] Services
  /go/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/test/e2e/service.go:70

[It] should be able to change the type and ports of a service
  /go/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/test/e2e/service.go:699
...
STEP: changing the TCP service to type=LoadBalancer
STEP: changing the UDP service to type=LoadBalancer
STEP: waiting for the TCP service to have a load balancer
Feb 11 18:53:08.217: INFO: Waiting up to 20m0s for service "mutability-test" to have a LoadBalancer
Feb 11 18:53:10.291: INFO: TCP load balancer: a999de63bd13311e59c500a99e494a48-1780227540.us-east-1.elb.amazonaws.com
STEP: waiting for the UDP service mutability-test to have a load balancer
STEP: waiting for the UDP service to have a load balancer
Feb 11 18:53:10.291: INFO: Waiting up to 20m0s for service "mutability-test" to have a LoadBalancer
Feb 11 19:13:10.433: FAIL: Timeout waiting for service "mutability-test" to have a load balancer
...

• Failure [1281.610 seconds]
Services
/go/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/test/e2e/service.go:885
  should be able to change the type and ports of a service [It]
  /go/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/test/e2e/service.go:699

  Feb 11 19:13:10.433: Timeout waiting for service "mutability-test" to have a load balancer

  /go/src/k8s.io/kubernetes/_output/dockerized/go/src/k8s.io/kubernetes/test/e2e/service.go:1602

Assigning to @justinsb since I believe he is fixing this stuff. Ping back if not true.

Thank you. I'm pretty sure this is fixed (I'm running the e2e tests locally using a simulate-Jenkins hack), and they are currently all green. But once we get the pending PRs merged I will verify / investigate whether there is something wrong with Jenkins!

Jenkins ran AWS tests on 3/1 and 2/29

The cluster successfully comes up, but the two tests: Services should be able to up and down services and SSH should SSH to all nodes and run commands are failing. We can either close this and open a new issue or just track those here.

On release-1.1 branch there are lots of tests failing, but I don't think that's a priority.

@spxtr those are failing on 1.2? The default SSH username changed to "admin" if you're using jessie, which is now the default if you don't set KUBE_OS_DISTRIBUTION. So KUBE_SSH_USER needs to be changed from ubuntu -> jessie. I can file a PR for that.

I think service up and down is a flake, but I've seen it come and go also.

I would hope 1.1 should pass tests, but that shouldn't be a priority over 1.2.

I propose we close this and open 2 issues: default SSH username change, and 1.1 e2e being not happy. And then if we see the service up and down again we open it too...

SGTM

Opened those two issues; closing this one.