fix: logic for error on duplicate job

grafana · Feb 2, 2024 · 0561eb1 · 0561eb1
1 parent b1a5e24
commit 0561eb1
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 6 deletions.
diff --git a/api/v1alpha1/k6conditions.go b/api/v1alpha1/k6conditions.go
@@ -88,7 +88,6 @@ func Initialize(k6 TestRunI) {
 
 	// PLZ test run case
 	if len(k6.GetSpec().TestRunID) > 0 {
-		UpdateCondition(k6, CloudTestRun, metav1.ConditionTrue)
 		UpdateCondition(k6, CloudPLZTestRun, metav1.ConditionTrue)
 		UpdateCondition(k6, CloudTestRunCreated, metav1.ConditionTrue)
 		UpdateCondition(k6, CloudTestRunFinalized, metav1.ConditionFalse)

diff --git a/controllers/k6_create.go b/controllers/k6_create.go
@@ -88,9 +88,11 @@ func createJobSpecs(ctx context.Context, log logr.Logger, k6 v1alpha1.TestRunI,
 		log.Info(err.Error())
 
 		// is it possible to implement this delay with resourceVersion of the job?
+
 		t, condUpdated := v1alpha1.LastUpdate(k6, v1alpha1.CloudTestRun)
-		// if condition has not been updated yet or has been updated very recently
-		if !condUpdated || time.Since(t).Seconds() <= 30 {
+		// If condition is unknown then resource hasn't been updated with `k6 inspect` results.
+		// If it has been updated but very recently, wait a bit before throwing an error.
+		if v1alpha1.IsUnknown(k6, v1alpha1.CloudTestRun) || !condUpdated || time.Since(t).Seconds() <= 30 {
 			// try again before returning an error
 			return ctrl.Result{RequeueAfter: time.Second * 10}, true, nil
 		}

diff --git a/controllers/testrun_controller.go b/controllers/testrun_controller.go
@@ -80,9 +80,13 @@ func (r *TestRunReconciler) Reconcile(ctx context.Context, req ctrl.Request) (ct
 	return r.reconcile(ctx, req, log, k6)
 }
 
+func isCloudTestRun(k6 v1alpha1.TestRunI) bool {
+	return v1alpha1.IsTrue(k6, v1alpha1.CloudTestRun) || v1alpha1.IsTrue(k6, v1alpha1.CloudPLZTestRun)
+}
+
 func (r *TestRunReconciler) reconcile(ctx context.Context, req ctrl.Request, log logr.Logger, k6 v1alpha1.TestRunI) (ctrl.Result, error) {
 	var err error
-	if v1alpha1.IsTrue(k6, v1alpha1.CloudTestRun) {
+	if isCloudTestRun(k6) {
 		// bootstrap the client
 		found, err := r.createClient(ctx, k6, log)
 		if err != nil {
@@ -100,7 +104,7 @@ func (r *TestRunReconciler) reconcile(ctx context.Context, req ctrl.Request, log
 	// Decision making here is now a mix between stages and conditions.
 	// TODO: refactor further.
 
-	if v1alpha1.IsTrue(k6, v1alpha1.CloudTestRun) && v1alpha1.IsFalse(k6, v1alpha1.CloudTestRunAborted) {
+	if isCloudTestRun(k6) && v1alpha1.IsFalse(k6, v1alpha1.CloudTestRunAborted) {
 		// check in with the BE for status
 		if r.ShouldAbort(ctx, k6, log) {
 			log.Info("Received an abort signal from the k6 Cloud: stopping the test.")
@@ -141,7 +145,7 @@ func (r *TestRunReconciler) reconcile(ctx context.Context, req ctrl.Request, log
 					msg := fmt.Sprintf(errMessageTooLong, "initializer pod", "initializer job and pod")
 					log.Info(msg)
 
-					if v1alpha1.IsTrue(k6, v1alpha1.CloudTestRun) {
+					if isCloudTestRun(k6) {
 						events := cloud.ErrorEvent(cloud.K6OperatorStartError).
 							WithDetail(msg).
 							WithAbort()