fix: check in the olm test that the spod resource exists before waiti…

…ng to be ready

This is to avoid that the olm test is randomly failing with 'resource not found', due to this
unsolved issue in kubernetes https://github.com/kubernetes/kubernetes/issues/83242.

Signed-off-by: Cosmin Cojocar <gcojocar@adobe.com>

hack/ci/e2e-olm.sh

@@ -90,6 +90,17 @@ function check_spo_is_running() {
     # wait for the operator to be ready
     kubectl -nsecurity-profiles-operator wait --for=condition=ready pod -lname=security-profiles-operator
     kubectl -nsecurity-profiles-operator wait --for=condition=ready pod -lname=security-profiles-operator-webhook
+
+    # wait for spod pod to be created, kubectl wait for un-existed resource seems to exit with error
+    # which is causing random test failure
+    # see https://github.com/kubernetes/kubernetes/issues/83242
+    for i in $(seq 1 10); do 
+        found=$(kubectl get -nsecurity-profiles-operator pods -lname=spod 2> /dev/null)
+        if [[ $found ]]; then
+            break
+        fi
+        sleep 5
+    done
     kubectl -nsecurity-profiles-operator wait --for=condition=ready pod -lname=spod
 }