cshannon commented on code in PR #5353:
URL: https://github.com/apache/accumulo/pull/5353#discussion_r1976087870


##########
test/src/main/java/org/apache/accumulo/test/functional/TabletMergeabilityIT.java:
##########
@@ -0,0 +1,219 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   https://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+package org.apache.accumulo.test.functional;
+
+import static 
org.apache.accumulo.test.compaction.ExternalCompactionTestUtils.countTablets;
+import static org.junit.jupiter.api.Assertions.assertTrue;
+
+import java.time.Duration;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.Map;
+import java.util.Set;
+import java.util.SortedMap;
+import java.util.TreeMap;
+import java.util.TreeSet;
+
+import org.apache.accumulo.core.client.Accumulo;
+import org.apache.accumulo.core.client.AccumuloClient;
+import org.apache.accumulo.core.client.admin.NewTableConfiguration;
+import org.apache.accumulo.core.client.admin.TabletMergeability;
+import org.apache.accumulo.core.clientImpl.TabletMergeabilityUtil;
+import org.apache.accumulo.core.conf.Property;
+import org.apache.accumulo.core.data.TableId;
+import org.apache.accumulo.core.dataImpl.KeyExtent;
+import org.apache.accumulo.core.metadata.schema.TabletMetadata;
+import org.apache.accumulo.harness.MiniClusterConfigurationCallback;
+import org.apache.accumulo.harness.SharedMiniClusterBase;
+import org.apache.accumulo.miniclusterImpl.MiniAccumuloConfigImpl;
+import org.apache.accumulo.server.ServerContext;
+import org.apache.accumulo.test.TestIngest;
+import org.apache.accumulo.test.VerifyIngest;
+import org.apache.accumulo.test.VerifyIngest.VerifyParams;
+import org.apache.accumulo.test.util.Wait;
+import org.apache.hadoop.conf.Configuration;
+import org.apache.hadoop.io.Text;
+import org.junit.jupiter.api.AfterAll;
+import org.junit.jupiter.api.BeforeAll;
+import org.junit.jupiter.api.Test;
+
+public class TabletMergeabilityIT extends SharedMiniClusterBase {
+
+  @Override
+  protected Duration defaultTimeout() {
+    return Duration.ofMinutes(5);
+  }
+
+  @BeforeAll
+  public static void setup() throws Exception {
+    SharedMiniClusterBase.startMiniClusterWithConfig(new Callback());
+  }
+
+  @AfterAll
+  public static void teardown() {
+    SharedMiniClusterBase.stopMiniCluster();
+  }
+
+  private static class Callback implements MiniClusterConfigurationCallback {
+    @Override
+    public void configureMiniCluster(MiniAccumuloConfigImpl cfg, Configuration 
coreSite) {
+      // Configure a short period of time to run the auto merge thread for 
testing
+      cfg.setProperty(Property.MANAGER_TABLET_MERGEABILITY_INTERVAL, "3s");
+    }
+  }
+
+  @Test
+  public void testMergeabilityAll() throws Exception {
+    String tableName = getUniqueNames(1)[0];
+    try (AccumuloClient c = 
Accumulo.newClient().from(getClientProps()).build()) {
+      c.tableOperations().create(tableName);
+      var tableId = 
TableId.of(c.tableOperations().tableIdMap().get(tableName));
+
+      TreeSet<Text> splits = new TreeSet<>();
+      splits.add(new Text(String.format("%09d", 333)));
+      splits.add(new Text(String.format("%09d", 666)));
+      splits.add(new Text(String.format("%09d", 999)));
+
+      // create splits with mergeabilty disabled so the task does not merge 
them away
+      // The default tablet is always mergeable, but it is currently the only 
one that is mergeable,
+      // so nothing will merge
+      c.tableOperations().putSplits(tableName, 
TabletMergeabilityUtil.userDefaultSplits(splits));
+      Wait.waitFor(() -> countTablets(getCluster().getServerContext(), 
tableName, tm -> true) == 4,
+          5000, 200);
+
+      // update to always mergeable so the task can now merge tablets
+      c.tableOperations().putSplits(tableName, 
TabletMergeabilityUtil.systemDefaultSplits(splits));
+
+      // Wait for merge, we should have one tablet
+      Wait.waitFor(() -> hasExactTablets(getCluster().getServerContext(), 
tableId,
+          Set.of(new KeyExtent(tableId, null, null))), 10000, 200);
+
+    }
+  }
+
+  @Test
+  public void testMergeabilityMultipleRanges() throws Exception {
+    String tableName = getUniqueNames(1)[0];
+    try (AccumuloClient c = 
Accumulo.newClient().from(getClientProps()).build()) {
+      c.tableOperations().create(tableName);
+      var tableId = 
TableId.of(c.tableOperations().tableIdMap().get(tableName));
+
+      SortedMap<Text,TabletMergeability> splits = new TreeMap<>();
+      splits.put(new Text(String.format("%09d", 333)), 
TabletMergeability.never());
+      splits.put(new Text(String.format("%09d", 555)), 
TabletMergeability.never());
+      splits.put(new Text(String.format("%09d", 666)), 
TabletMergeability.never());
+      splits.put(new Text(String.format("%09d", 999)), 
TabletMergeability.never());
+
+      c.tableOperations().putSplits(tableName, splits);
+      Wait.waitFor(() -> countTablets(getCluster().getServerContext(), 
tableName, tm -> true) == 5,
+          5000, 500);
+
+      splits.put(new Text(String.format("%09d", 333)), 
TabletMergeability.always());
+      splits.put(new Text(String.format("%09d", 555)), 
TabletMergeability.always());
+      // Keep tablet 666 as never, this should cause two fate jobs for merging
+      splits.put(new Text(String.format("%09d", 999)), 
TabletMergeability.always());
+      c.tableOperations().putSplits(tableName, splits);
+
+      // Wait for merge, we should have 3 tablets
+      // 333 and 555 should be merged into 555
+      // 666
+      // 999 and default merged into default
+      Wait.waitFor(() -> hasExactTablets(getCluster().getServerContext(), 
tableId,
+          Set.of(new KeyExtent(tableId, new Text(String.format("%09d", 555)), 
null),
+              new KeyExtent(tableId, new Text(String.format("%09d", 666)),
+                  new Text(String.format("%09d", 555))),
+              new KeyExtent(tableId, null, new Text(String.format("%09d", 
666))))),
+          10000, 200);
+
+    }
+  }
+
+  @Test
+  public void testSplitAndMergeAll() throws Exception {
+    String tableName = getUniqueNames(1)[0];
+    try (AccumuloClient c = 
Accumulo.newClient().from(getClientProps()).build()) {
+      Map<String,String> props = new HashMap<>();
+      props.put(Property.TABLE_SPLIT_THRESHOLD.getKey(), "32K");
+      props.put(Property.TABLE_FILE_COMPRESSED_BLOCK_SIZE.getKey(), "1K");
+      c.tableOperations().create(tableName, new 
NewTableConfiguration().setProperties(props));
+      var tableId = 
TableId.of(c.tableOperations().tableIdMap().get(tableName));
+
+      // Ingest data so tablet will split
+      VerifyParams params = new VerifyParams(getClientProps(), tableName, 
50_000);
+      TestIngest.ingest(c, params);
+      VerifyIngest.verifyIngest(c, params);
+
+      // Wait for table to split, should be more than 10 tablets
+      Wait.waitFor(() -> c.tableOperations().listSplits(tableName).size() > 
10, 10000, 200);
+
+      // Delete all the data
+      c.tableOperations().deleteRows(tableName, null, null);

Review Comment:
   If I recall, I believe the reason why we use the merge code to merge away 
empty tablets is to preserve the behavior of previous versions of accumulo that 
deleted tablets. 
   
   In older versions, like 2.1, when we ran delete rows, the code would run a 
chop compaction on the tablets at the boundaries and then it could just delete 
the all the tablets in between the delete range. In 3.1, with no chop merges we 
fence the boundary tablets but still run deletion on the in between tablets.
   
   In 4.0 this was changed further and the code will fence the tablets on the 
boundaries (just like 3.1), but instead of deleting the middle tablets in the 
range it will instead delete the data so the tablets are empty and then merge 
away those empty tablets.
   
   The end result in all 3 cases is those tablets are gone, so if we do decide 
to keep the tablets going forward that would be a different behavior than 
before.



-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to