diff --git a/cmd/background-newdisks-heal-ops.go b/cmd/background-newdisks-heal-ops.go
index 7f8407c44..2dc11d9d4 100644
--- a/cmd/background-newdisks-heal-ops.go
+++ b/cmd/background-newdisks-heal-ops.go
@@ -383,6 +383,10 @@ func healFreshDisk(ctx context.Context, z *erasureServerPools, endpoint Endpoint
 		logger.Info("Healing drive '%s' complete (healed: %d, failed: %d).", disk, tracker.ItemsHealed, tracker.ItemsFailed)
 	}
 
+	if len(tracker.QueuedBuckets) > 0 {
+		return fmt.Errorf("not all buckets were healed: %v", tracker.QueuedBuckets)
+	}
+
 	if serverDebugLog {
 		tracker.printTo(os.Stdout)
 		logger.Info("\n")
diff --git a/cmd/global-heal.go b/cmd/global-heal.go
index a95c6a59e..7b2160be0 100644
--- a/cmd/global-heal.go
+++ b/cmd/global-heal.go
@@ -214,11 +214,7 @@ func (er *erasureObjects) healErasureSet(ctx context.Context, buckets []string,
 
 		disks, _ := er.getOnlineDisksWithHealing()
 		if len(disks) == 0 {
-			// all disks are healing in this set, this is allowed
-			// so we simply proceed to next bucket, marking the bucket
-			// as done as there are no objects to heal.
-			tracker.bucketDone(bucket)
-			logger.LogIf(ctx, tracker.update(ctx))
+			logger.LogIf(ctx, fmt.Errorf("no online disks found to heal the bucket `%s`", bucket))
 			continue
 		}