zhtaoxiang commented on code in PR #13636:
URL: https://github.com/apache/pinot/pull/13636#discussion_r1718826384


##########
pinot-segment-local/src/main/java/org/apache/pinot/segment/local/dedup/ConcurrentMapPartitionDedupMetadataManager.java:
##########
@@ -19,106 +19,99 @@
 package org.apache.pinot.segment.local.dedup;
 
 import com.google.common.annotations.VisibleForTesting;
-import java.util.HashMap;
+import com.google.common.util.concurrent.AtomicDouble;
+import java.io.IOException;
 import java.util.Iterator;
-import java.util.List;
-import java.util.Map;
 import java.util.concurrent.ConcurrentHashMap;
+import org.apache.commons.lang3.tuple.Pair;
 import org.apache.pinot.common.metrics.ServerGauge;
-import org.apache.pinot.common.metrics.ServerMetrics;
-import org.apache.pinot.segment.local.segment.readers.PinotSegmentColumnReader;
 import org.apache.pinot.segment.local.utils.HashUtils;
 import org.apache.pinot.segment.spi.IndexSegment;
-import org.apache.pinot.spi.config.table.HashFunction;
-import org.apache.pinot.spi.data.readers.PrimaryKey;
-import org.apache.pinot.spi.utils.ByteArray;
 
-class ConcurrentMapPartitionDedupMetadataManager implements 
PartitionDedupMetadataManager {
-  private final String _tableNameWithType;
-  private final List<String> _primaryKeyColumns;
-  private final int _partitionId;
-  private final ServerMetrics _serverMetrics;
-  private final HashFunction _hashFunction;
 
+class ConcurrentMapPartitionDedupMetadataManager extends 
BasePartitionDedupMetadataManager {
   @VisibleForTesting
-  final ConcurrentHashMap<Object, IndexSegment> _primaryKeyToSegmentMap = new 
ConcurrentHashMap<>();
+  final AtomicDouble _largestSeenTime = new AtomicDouble(0);
+  @VisibleForTesting
+  final ConcurrentHashMap<Object, Pair<IndexSegment, Double>> 
_primaryKeyToSegmentAndTimeMap =
+      new ConcurrentHashMap<>();
 
-  public ConcurrentMapPartitionDedupMetadataManager(String tableNameWithType, 
List<String> primaryKeyColumns,
-      int partitionId, ServerMetrics serverMetrics, HashFunction hashFunction) 
{
-    _tableNameWithType = tableNameWithType;
-    _primaryKeyColumns = primaryKeyColumns;
-    _partitionId = partitionId;
-    _serverMetrics = serverMetrics;
-    _hashFunction = hashFunction;
+  protected ConcurrentMapPartitionDedupMetadataManager(String 
tableNameWithType, int partitionId,
+      DedupContext dedupContext) {
+    super(tableNameWithType, partitionId, dedupContext);
   }
 
-  public void addSegment(IndexSegment segment) {
-    // Add all PKs to _primaryKeyToSegmentMap
-    Iterator<PrimaryKey> primaryKeyIterator = getPrimaryKeyIterator(segment);
-    while (primaryKeyIterator.hasNext()) {
-      PrimaryKey pk = primaryKeyIterator.next();
-      _primaryKeyToSegmentMap.put(HashUtils.hashPrimaryKey(pk, _hashFunction), 
segment);
+  @Override
+  protected void doAddSegment(IndexSegment segment, Iterator<DedupRecordInfo> 
dedupRecordInfoIterator) {
+    while (dedupRecordInfoIterator.hasNext()) {
+      DedupRecordInfo dedupRecordInfo = dedupRecordInfoIterator.next();
+      double dedupTime = dedupRecordInfo.getDedupTime();
+      _largestSeenTime.getAndUpdate(time -> Math.max(time, dedupTime));
+      
_primaryKeyToSegmentAndTimeMap.compute(HashUtils.hashPrimaryKey(dedupRecordInfo.getPrimaryKey(),
 _hashFunction),

Review Comment:
   The default implementation of the replaceSegment method in the previous 
change removes old segment first, then adds the new segment. 
   
   This implementation has a problem: this replacement logic is not atomic. 
Specifically, just after the old segment is removed and before the replaced 
segment is added, a record with a primary key that is contained in both and old 
and the new segments can be added successfully - this is a duplication.
   
   This is fixed in 
https://github.com/apache/pinot/pull/13636/commits/eb4065fb406302fb6b21275fcb193494fd7c27da



-- 
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: commits-unsubscr...@pinot.apache.org

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


---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org
For additional commands, e-mail: commits-h...@pinot.apache.org

Reply via email to