Add CI e2e script

hack/.ci/run-e2e-gke.sh

@@ -0,0 +1,113 @@
+#!/usr/bin/env bash
+#
+# Copyright (C) 2023 ScyllaDB
+#
+
+set -euExo pipefail
+shopt -s inherit_errexit
+
+if [ -z ${E2E_SUITE+x} ]; then
+  echo "E2E_SUITE can't be empty" > /dev/stderr
+  exit 2
+fi
+
+if [ -z ${E2E_IMAGE+x} ]; then
+  echo "E2E_IMAGE can't be empty" > /dev/stderr
+  exit 2
+fi
+
+if [ -z ${E2E_TESTS_IMAGE+x} ]; then
+  echo "E2E_TESTS_IMAGE can't be empty" > /dev/stderr
+  exit 2
+fi
+
+if [ -z ${ARTIFACTS+x} ]; then
+  echo "ARTIFACTS can't be empty" > /dev/stderr
+  exit 2
+fi
+
+REENTRANT=${REENTRANT:-false}
+export REENTRANT
+
+field_manager=run-e2e-script
+
+function kubectl_create {
+    if [[ "${REENTRANT}" != "true" ]]; then
+        # In an actual CI run we have to enforce that no two objects have the same name.
+        kubectl create --field-manager="${field_manager}" "$@"
+    else
+        # For development iterations we want to update the objects.
+        kubectl apply --server-side=true --field-manager="${field_manager}" --force-conflicts "$@"
+    fi
+}
+
+function gather-artifacts {
+  return # FIXME: remove
+  kubectl -n e2e run --restart=Never --image="quay.io/scylladb/scylla-operator:latest" --labels='app=must-gather' --command=true must-gather -- bash -euExo pipefail -O inherit_errexit -c "function wait-for-artifacts { touch /tmp/done && until [[ -f '/tmp/exit' ]]; do sleep 1; done } && trap wait-for-artifacts EXIT && mkdir /tmp/artifacts && scylla-operator must-gather --all-resources --loglevel=2 --dest-dir=/tmp/artifacts"
+  kubectl -n e2e wait --for=condition=Ready pod/must-gather
+
+  # Setup artifacts transfer when finished and unblock the must-gather pod when done.
+  (
+    function unblock-must-gather-pod {
+      kubectl -n e2e exec pod/must-gather -- bash -euEo pipefail -O inherit_errexit -c "touch /tmp/exit"
+    }
+    trap unblock-must-gather-pod EXIT
+
+    kubectl -n e2e exec pod/must-gather -- bash -euEo pipefail -O inherit_errexit -c "until [[ -f /tmp/done ]]; do sleep 1; done; ls -l /tmp/artifacts"
+    kubectl -n e2e cp --retries=42 must-gather:/tmp/artifacts "${ARTIFACTS}/must-gather"
+    ls -l "${ARTIFACTS}"
+  ) &
+  must_gather_bg_pid=$!
+
+  kubectl -n e2e logs -f pod/must-gather
+  exit_code=$( kubectl -n e2e get pods/must-gather --output='jsonpath={.status.containerStatuses[0].state.terminated.exitCode}' )
+  kubectl -n e2e delete pod/must-gather --wait=false
+
+  if [[ "${exit_code}" != "0" ]]; then
+    echo "Collecting artifacts using must-gather failed"
+    exit "${exit_code}"
+  fi
+
+  wait "${must_gather_bg_pid}"
+}
+
+function handle-exit {
+  gather-artifacts || "Error gathering artifacts" > /dev/stderr
+}
+
+trap handle-exit EXIT
+
+timeout -v 10m ./hack/ci-deploy.sh "${E2E_IMAGE}"
+
+kubectl create namespace e2e --dry-run=client -o yaml | kubectl_create -f -
+kubectl create clusterrolebinding e2e --clusterrole=cluster-admin --serviceaccount=e2e:default --dry-run=client -o yaml | kubectl_create -f -
+kubectl create -n e2e pdb my-pdb --selector='app=e2e' --min-available=1 --dry-run=client -o yaml | kubectl_create -f -
+
+kubectl -n e2e run --restart=Never --image="${E2E_TESTS_IMAGE}" --labels='app=e2e' --command=true e2e -- bash -euExo pipefail -O inherit_errexit -c "function wait-for-artifacts { touch /tmp/done && until [[ -f '/tmp/exit' ]]; do sleep 1; done } && trap wait-for-artifacts EXIT && mkdir /tmp/artifacts && local-csi-driver-tests run '${E2E_SUITE}' --loglevel=2 --color=false --artifacts-dir=/tmp/artifacts"
+kubectl -n e2e wait --for=condition=Ready pod/e2e
+
+# Setup artifacts transfer when finished and unblock the e2e pod when done.
+(
+  function unblock-e2e-pod {
+    kubectl -n e2e exec pod/e2e -- bash -euEo pipefail -O inherit_errexit -c "touch /tmp/exit"
+  }
+  trap unblock-e2e-pod EXIT
+
+  kubectl -n e2e exec pod/e2e -- bash -euEo pipefail -O inherit_errexit -c "until [[ -f /tmp/done ]]; do sleep 1; done; ls -l /tmp/artifacts"
+  kubectl -n e2e cp --retries=42 e2e:/tmp/artifacts "${ARTIFACTS}"
+  ls -l "${ARTIFACTS}"
+) &
+e2e_bg_pid=$!
+
+kubectl -n e2e logs -f pod/e2e
+exit_code=$( kubectl -n e2e get pods/e2e --output='jsonpath={.status.containerStatuses[0].state.terminated.exitCode}' )
+kubectl -n e2e delete pod/e2e --wait=false
+
+wait "${e2e_bg_pid}" || ( echo "Collecting e2e artifacts failed" && exit 2 )
+
+if [[ "${exit_code}" != "0" ]]; then
+  echo "E2E tests failed"
+  exit "${exit_code}"
+fi
+
+echo "E2E tests finished successfully"