ReplicatedRepository installs user triggers on the replica again, but it now disables all triggers during resync to prevent errors. When triggers were on master, downstream triggers would not see changes made by earlier triggers.

1 files changed, 21 insertions, 15 deletions

diff --git a/src/test/java/com/amazon/carbonado/repo/replicated/TestRepair.java b/src/test/java/com/amazon/carbonado/repo/replicated/TestRepair.java
index c993e4b..2594810 100644
--- a/src/test/java/com/amazon/carbonado/repo/replicated/TestRepair.java
+++ b/src/test/java/com/amazon/carbonado/repo/replicated/TestRepair.java
@@ -490,28 +490,34 @@ public class TestRepair extends TestCase {
 

         // Resync to repair.

 

-        Trigger trigger = null;

+        class Prevent extends Trigger {

+            boolean didRun;

+

+            @Override

+            public Object beforeDelete(Object s) throws PersistException {

+                didRun = true;

+                throw new PersistException("Cannot delete me!");

+            }

+        }

+

+        Prevent prevent = null;

         if (preventDelete) {

-            // The resync will try to delete the corrupt replica entries, but

-            // this trigger will cause the delete to fail. Instead, the resync

-            // will skip over these entries instead of crashing outright.

-            trigger = new Trigger() {

-                @Override

-                public Object beforeDelete(Object s) throws PersistException {

-                    throw new PersistException("Cannot delete me!");

-                }

-            };

-            storage0.addTrigger(trigger);

+            // This is a partially vestigial test. It used to be the case that

+            // triggers would run during a resync or replication repair. This

+            // is no longer the case. Instead, use this as an opportunity to

+            // ensure the trigger does not run.

+            prevent = new Prevent();

+            storage0.addTrigger(prevent);

         }

 

         ResyncCapability cap = mReplicated.getCapability(ResyncCapability.class);

         cap.resync(type0, 1.0, null);

 

         if (preventDelete) {

-            // If this point is reached, the resync didn't crash, but it hasn't

-            // actually repaired the broken records. Remove the trigger and do

-            // the resync again.

-            storage0.removeTrigger(trigger);

+            // Again, this is a partially vestigial test. The trigger should

+            // not have run at all during the resync.

+            assertFalse(prevent.didRun);

+            storage0.removeTrigger(prevent);

             cap.resync(type0, 1.0, null);

         }