update documentation and improve deletion logs

docs/reference/operator_parameters.md

@@ -339,10 +339,11 @@ configuration they are grouped under the `kubernetes` key.
   cannot fully sync it, there can be leftovers. By enabling finalizers the
   operator will ensure all managed resources are deleted prior to the
   Postgresql resource. There is a trade-off though: The deletion is only
-  performed at the next cluster SYNC cycle when finding a `deletionTimestamp`
-  in the metadata and not immediately after issueing a delete command. The
-  final removal of the custom resource will add a DELETE event to the worker
-  queue but the child resources are already gone at this point.
+  performed after the next two SYNC cycles with the first one updating the
+  internal spec and the latter reacting on the `deletionTimestamp` while
+  processing the SYNC event. The final removal of the custom resource will
+  add a DELETE event to the worker queue but the child resources are already
+  gone at this point.
   The default is `false`.
 
 * **enable_pod_disruption_budget**

pkg/controller/postgresql.go

@@ -331,7 +331,6 @@ func (c *Controller) processEvent(event ClusterEvent) {
 		c.curWorkerCluster.Store(event.WorkerID, cl)
 
 		// has this cluster been marked as deleted already, then we shall start cleaning up
-		lg.Infof("cluster DeletionTimestamp is %s", cl.ObjectMeta.DeletionTimestamp)
 		if !cl.ObjectMeta.DeletionTimestamp.IsZero() {
 			lg.Infof("cluster has a DeletionTimestamp of %s, starting deletion now.", cl.ObjectMeta.DeletionTimestamp.Format(time.RFC3339))
 			if err = cl.Delete(); err != nil {
@@ -340,7 +339,6 @@ func (c *Controller) processEvent(event ClusterEvent) {
 				lg.Error(cl.Error)
 				return
 			}
-			lg.Infof("cluster has been deleted")
 		} else {
 			if err = cl.Sync(event.NewSpec); err != nil {
 				cl.Error = fmt.Sprintf("could not sync cluster: %v", err)