test: run OLM tests first, III

Top-level nodes were still getting
randomized (onsi/ginkgo#966 (comment)).

test/e2e/deploy/deploy.go

@@ -32,6 +32,7 @@ import (
 	metav1 "k8s.io/apimachinery/pkg/apis/meta/v1"
 	"k8s.io/apimachinery/pkg/labels"
 	"k8s.io/apimachinery/pkg/util/wait"
+	"k8s.io/klog/v2"
 	"k8s.io/klog/v2/klogr"
 	"k8s.io/kubernetes/test/e2e/framework"
 	"k8s.io/kubernetes/test/e2e/framework/skipper"
@@ -1511,7 +1512,19 @@ func Describe(deployment, describe, what string, f func(d *Deployment)) bool {
 }
 
 // DefineTests must be called to register all tests defined so far via Describe.
+//
+// They are placed inside a "Deploy" node because:
+// - The order of specs inside a node is not randomized by default,
+//   which is what we want, in contrast to the top-level nodes,
+//   which always get randomized (https://github.com/onsi/ginkgo/issues/966#issuecomment-1110046593).
+// - It is easier to focus on specific tests with, for example, "Deploy operator".
+//   "^operator" does not work because the internal test name string has some
+//   builtin, hidden string at the beginning (something like "top level").
 func DefineTests() {
+	ginkgo.Context("Deploy", defineTests)
+}
+
+func defineTests() {
 	all := allDeployments[:]
 	for deploymentName := range tests {
 		if !haveDeployment(all, deploymentName) {
@@ -1527,6 +1540,7 @@ func DefineTests() {
 		deploymentName := deploymentName
 		for describe, funcs := range group {
 			funcs := funcs
+			klog.InfoS("defining tests", "deployment", deploymentName, "test", describe)
 			ginkgo.Describe(describe, func() {
 				var deployment *Deployment
 				if deploymentName != "" {