rohityadav1993 commented on code in PR #17789:
URL: https://github.com/apache/pinot/pull/17789#discussion_r2885708176
##########
pinot-core/src/main/java/org/apache/pinot/core/data/manager/offline/OfflineTableDataManager.java:
##########
@@ -57,8 +101,122 @@ protected void doAddOnlineSegment(String segmentName)
}
}
+ @Override
+ public void addSegment(ImmutableSegment immutableSegment, @Nullable
SegmentZKMetadata zkMetadata) {
+ String segmentName = immutableSegment.getSegmentName();
+ Preconditions.checkState(!_shutDown,
+ "Table data manager is already shut down, cannot add segment: %s to
table: %s",
+ segmentName, _tableNameWithType);
+ if (isUpsertEnabled()) {
+ handleUpsert(immutableSegment, zkMetadata);
+ return;
+ }
+ super.addSegment(immutableSegment, zkMetadata);
+ }
+
+ @Override
+ public List<SegmentContext> getSegmentContexts(List<IndexSegment>
selectedSegments,
+ Map<String, String> queryOptions) {
+ List<SegmentContext> segmentContexts = new
ArrayList<>(selectedSegments.size());
+ selectedSegments.forEach(s -> segmentContexts.add(new SegmentContext(s)));
+ if (isUpsertEnabled() && !QueryOptionsUtils.isSkipUpsert(queryOptions)) {
+ _tableUpsertMetadataManager.setSegmentContexts(segmentContexts,
queryOptions);
+ }
+ return segmentContexts;
+ }
+
@Override
public void addConsumingSegment(String segmentName) {
throw new UnsupportedOperationException("Cannot add CONSUMING segment to
OFFLINE table");
}
+
+ public boolean isUpsertEnabled() {
+ return _tableUpsertMetadataManager != null;
+ }
+
+ @VisibleForTesting
+ public TableUpsertMetadataManager getTableUpsertMetadataManager() {
+ return _tableUpsertMetadataManager;
+ }
+
+ public Map<Integer, Long> getPartitionToPrimaryKeyCount() {
+ if (isUpsertEnabled()) {
+ return _tableUpsertMetadataManager.getPartitionToPrimaryKeyCount();
+ }
+ return Collections.emptyMap();
+ }
+
+ private void handleUpsert(ImmutableSegment immutableSegment, @Nullable
SegmentZKMetadata zkMetadata) {
Review Comment:
we may also have to think of a segment assignement strategy for offline
upsert tables. Right now balanced segment assignment will not guaranteed all
segments of a partition(primary key based) will be assigned to a single server
so we may have duplicate rows for primary key.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]