test flake: exit status 1 leaked firewall rules

[ahmetb]

I had one of the PRs (#714) randomly failed with an error like the following.

https://gubernator.knative.dev/build/knative-prow/pr-logs/pull/knative_docs/714/pull-knative-docs-integration-tests/1082724531375706114/

W0108 20:08:52.863] 2019/01/08 20:08:52 process.go:155: Step 'gcloud compute firewall-rules delete -q kdocs-e2e-net1082724531375706114-2wnao3jebww7ldrn463stwke kdocs-e2e-net1082724531375706114-7mzjmae3tlidh4yoidvnpe53 kdocs-e2e-net1082724531375706114-ba5h4uquy4cktbsldj6ba2g3 kdocs-e2e-net1082724531375706114-knsku4qwwbtr3bhcf3y6vcmu kdocs-e2e-net1082724531375706114-pjf5geul3rfwfyd5mynbasya kdocs-e2e-net1082724531375706114-rpeifunzrsxxfwyacvtmctpr kdocs-e2e-net1082724531375706114-uuaca3r2wn7yqlbdyzue5lrn kdocs-e2e-net1082724531375706114-x233wi3sx7onq4enxk4rf6mk kdocs-e2e-net1082724531375706114-yieefqzg6fev2ygounhzgoee --project=knative-boskos-10' finished in 23.452468516s
W0108 20:08:52.863] 2019/01/08 20:08:52 process.go:153: Running: gcloud compute networks delete -q kdocs-e2e-net1082724531375706114 --project=knative-boskos-10
W0108 20:08:52.863] sh: 0: getcwd() failed: No such file or directory
W0108 20:08:53.143] sh: 0: getcwd() failed: No such file or directory
W0108 20:10:06.951] Deleted [https://www.googleapis.com/compute/v1/projects/knative-boskos-10/global/networks/kdocs-e2e-net1082724531375706114].
W0108 20:10:07.400] 2019/01/08 20:10:07 process.go:155: Step 'gcloud compute networks delete -q kdocs-e2e-net1082724531375706114 --project=knative-boskos-10' finished in 1m14.537213121s
W0108 20:10:11.837] 2019/01/08 20:10:11 main.go:312: Something went wrong: encountered 2 errors: [error during /go/src/github.com/knative/docs/test/e2e-tests.sh --run-tests: exit status 1 leaked firewall rules]
I0108 20:10:11.938] Test subprocess exited with code 1
I0108 20:10:16.554] Test result code is 1
W0108 20:10:16.587] Run: ('/workspace/./test-infra/jenkins/../scenarios/../hack/coalesce.py',)
E0108 20:10:16.587] Command failed

I'm opening this to track the flake as it might impact other PRs.

/kind test

[adrcunha]

This is a red herring. The real error above is:

I0108 20:03:49.947] ERROR: timeout waiting for pods to come up

The resource leaks are kubetest's fault, that's why e2e-tests.sh does extra cleanup work. Also because of the resource leak, kubetest will always return 1, and e2e-tests.sh also handles that correctly. This was originally discovered in knative/serving#959 thus the workarounds; hopefully once we get to knative/test-infra#424 we won't have any of these issues or workarounds.