kgusakov commented on code in PR #4905:
URL: https://github.com/apache/ignite-3/pull/4905#discussion_r1895747419


##########
modules/table/src/integrationTest/java/org/apache/ignite/internal/table/distributed/disaster/ItHighAvailablePartitionsRecoveryTest.java:
##########
@@ -230,4 +230,64 @@ void testNoHaRecoveryForScZone() throws 
InterruptedException {
 
         assertRecoveryKeyIsEmpty(node);
     }
+
+    @Test
+    void testScaleUpAfterHaRecoveryWhenMajorityLoss() throws Exception {
+        startNode(3);
+
+        startNode(4);
+
+        createHaZoneWithTable();
+
+        IgniteImpl node = igniteImpl(0);
+
+        assertRecoveryKeyIsEmpty(node);
+
+        stopNodes(1, 2, 3, 4);
+
+        waitAndAssertRecoveryKeyIsNotEmpty(node);
+
+        assertRecoveryRequestForHaZoneTable(node);
+
+        waitAndAssertStableAssignmentsOfPartitionEqualTo(node, HA_TABLE_NAME, 
Set.of(0, 1), Set.of(node.name()));

Review Comment:
   You can use 
org.apache.ignite.internal.table.distributed.disaster.AbstractHighAvailablePartitionsRecoveryTest#PARTITION_IDS
 here and there, it will scale with the number of partitions



##########
modules/distribution-zones/src/main/java/org/apache/ignite/internal/distributionzones/rebalance/ZoneRebalanceUtil.java:
##########
@@ -155,7 +158,33 @@ public static CompletableFuture<Void> 
updatePendingAssignmentsKeys(
 
         ByteArray partAssignmentsStableKey = 
stablePartAssignmentsKey(zonePartitionId);
 
-        Set<Assignment> partAssignments = 
calculateAssignmentForPartition(dataNodes, partNum, replicas);
+        Set<Assignment> calculatedAssignments = 
calculateAssignmentForPartition(dataNodes, partNum, replicas);
+
+        Set<Assignment> partAssignments;
+
+        if (consistencyMode == ConsistencyMode.HIGH_AVAILABILITY) {

Review Comment:
   If you will provide the example with tricky case in comments, it will be 
great and simplify the understanding process



##########
modules/table/src/integrationTest/java/org/apache/ignite/internal/table/distributed/disaster/ItHighAvailablePartitionsRecoveryTest.java:
##########
@@ -230,4 +230,64 @@ void testNoHaRecoveryForScZone() throws 
InterruptedException {
 
         assertRecoveryKeyIsEmpty(node);
     }
+
+    @Test
+    void testScaleUpAfterHaRecoveryWhenMajorityLoss() throws Exception {
+        startNode(3);
+
+        startNode(4);
+
+        createHaZoneWithTable();
+
+        IgniteImpl node = igniteImpl(0);
+
+        assertRecoveryKeyIsEmpty(node);
+
+        stopNodes(1, 2, 3, 4);
+
+        waitAndAssertRecoveryKeyIsNotEmpty(node);
+
+        assertRecoveryRequestForHaZoneTable(node);
+
+        waitAndAssertStableAssignmentsOfPartitionEqualTo(node, HA_TABLE_NAME, 
Set.of(0, 1), Set.of(node.name()));
+
+        var node1 = startNode(1);
+
+        waitAndAssertStableAssignmentsOfPartitionEqualTo(
+                node,
+                HA_TABLE_NAME,
+                Set.of(0, 1),
+                Set.of(node.name(), node1.name())
+        );
+
+        var node2 = startNode(2);
+
+        waitAndAssertStableAssignmentsOfPartitionEqualTo(
+                node,
+                HA_TABLE_NAME,
+                Set.of(0, 1),
+                Set.of(node.name(), node1.name(), node2.name())
+        );
+    }
+
+    @Test
+    void testHaZoneScaleDownNodesDoNotRemovedFromStable() throws 
InterruptedException {
+        startNode(3);
+
+        startNode(4);
+
+        createHaZoneWithTable();
+
+        IgniteImpl node = igniteImpl(0);
+
+        Set<String> allNodes = 
runningNodes().map(Ignite::name).collect(Collectors.toUnmodifiableSet());
+
+        stopNodes(3);

Review Comment:
   stopNodes(3, 4) or stopNode(3)



##########
modules/table/src/main/java/org/apache/ignite/internal/table/distributed/disaster/GroupUpdateRequest.java:
##########
@@ -337,7 +336,43 @@ private static CompletableFuture<Integer> partitionUpdate(
                         : Assignments.toBytes(partAssignments, 
assignmentsTimestamp)
         );
 
-        return 
metaStorageMgr.invoke(invokeClosure).thenApply(StatementResult::getAsInt);
+        return metaStorageMgr.invoke(invokeClosure).thenApply(sr -> {
+            switch (UpdateStatus.valueOf(sr.getAsInt())) {
+                case PENDING_KEY_UPDATED:
+                    LOG.info(
+                            "Force update metastore pending partitions key 
[key={}, partition={}, table={}, newVal={}]",
+                            pendingChangeTriggerKey(partId).toString(),
+                            partId.partitionId(),
+                            partId.tableId(),
+                            nextAssignment
+                    );
+
+                    break;
+                case ASSIGNMENT_NOT_UPDATED:

Review Comment:
   It looks like this status can't be thrown here? Am I missing smth?



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: notifications-unsubscr...@ignite.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org

Reply via email to