UPSTREAM: <carry>: Add ocp-e2e-tests script

When running the e2e tests on OCP, we need to do some things
slightly different. This adds a script and the necessary files to
do so.

hack/ocp-e2e-tests.sh

@@ -0,0 +1,32 @@
+#!/bin/bash
+
+# Make sure to set the IMAGE_REPO env variable to your quay.io username
+# before running this script.
+
+# Additionally, the e2e tests rely on extra nics being configured on the
+# node. If running from dev-scripts, it will be necessary to configure it to
+# deploy the extra nics.
+# See https://github.com/openshift-metal3/dev-scripts/pull/1286 for an example.
+
+set -ex
+
+export KUBEVIRT_PROVIDER=external
+export IMAGE_BUILDER=podman
+export DEV_IMAGE_REGISTRY=quay.io
+export KUBEVIRTCI_RUNTIME=podman
+export SSH=./hack/ssh.sh
+export PRIMARY_NIC=enp2s0
+export FIRST_SECONDARY_NIC=enp3s0
+export SECOND_SECONDARY_NIC=enp4s0
+
+make cluster-sync-operator
+# Will fail on subsequent runs, this is fine.
+oc create -f build/_output/manifests/scc.yaml || :
+oc create -f test/e2e/nmstate.yaml
+# On first deployment, it can take a while for all of the pods to come up
+# First wait for the handler pods to be created
+while ! oc get pods -n nmstate | grep handler; do sleep 1; done
+# Then wait for them to be ready
+while oc get pods -n nmstate | grep "0/1"; do sleep 1; done
+# NOTE(bnemec): The test being filtered with "bridged" was re-enabled in 4.8, but seems to be consistently failing on OCP.
+make test-e2e-handler E2E_TEST_ARGS="--skip=user-guide --skip=bridged" E2E_TEST_TIMEOUT=120m

hack/ssh.sh

@@ -0,0 +1,4 @@
+#!/bin/bash
+node_name=${1}
+node_ip=$(oc get node ${node_name} --no-headers -o wide | awk '{print $6}')
+ssh -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null core@${node_ip} -- ${@:3}

test/e2e/nmstate.yaml

@@ -0,0 +1,4 @@
+apiVersion: nmstate.io/v1beta1
+kind: NMState
+metadata:
+  name: nmstate