Skip to content

Move partial-upsert resolveComparisonTies method from base class to implementations #18640

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
xiangfu0 merged 2 commits into from
Jun 2, 2026
+11 −16
Merged

Move partial-upsert resolveComparisonTies method from base class to implementations #18640

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
79b4e81
Move partial-upsert resolveComparisonTies from base wrapper into subc…
deepthi912 May 31, 2026
3cbd93a
Fix Checkstyle line-length violation in BasePartitionUpsertMetadataMa…
deepthi912 May 31, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
18 changes: 4 additions & 14 deletions ...c/main/java/org/apache/pinot/segment/local/upsert/BasePartitionUpsertMetadataManager.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -494,17 +494,7 @@ public void addSegment(ImmutableSegmentImpl segment, @Nullable ThreadSafeMutable
if (queryableDocIds == null && _deleteRecordColumn != null) {
queryableDocIds = new ThreadSafeMutableRoaringBitmap();
}
addOrReplaceSegment(segment, validDocIds, queryableDocIds, recordInfoIterator, null, null);
}

protected void addOrReplaceSegment(ImmutableSegmentImpl segment, ThreadSafeMutableRoaringBitmap validDocIds,
@Nullable ThreadSafeMutableRoaringBitmap queryableDocIds, Iterator<RecordInfo> recordInfoIterator,
@Nullable IndexSegment oldSegment, @Nullable MutableRoaringBitmap validDocIdsForOldSegment) {
if (_partialUpsertHandler != null) {
recordInfoIterator = resolveComparisonTies(recordInfoIterator, _hashFunction);
}
doAddOrReplaceSegment(segment, validDocIds, queryableDocIds, recordInfoIterator, oldSegment,
validDocIdsForOldSegment);
doAddOrReplaceSegment(segment, validDocIds, queryableDocIds, recordInfoIterator, null, null);
}

protected abstract void doAddOrReplaceSegment(ImmutableSegmentImpl segment,
Expand All @@ -514,7 +504,7 @@ protected abstract void doAddOrReplaceSegment(ImmutableSegmentImpl segment,

protected void addSegmentWithoutUpsert(ImmutableSegmentImpl segment, ThreadSafeMutableRoaringBitmap validDocIds,
@Nullable ThreadSafeMutableRoaringBitmap queryableDocIds, Iterator<RecordInfo> recordInfoIterator) {
addOrReplaceSegment(segment, validDocIds, queryableDocIds, recordInfoIterator, null, null);
doAddOrReplaceSegment(segment, validDocIds, queryableDocIds, recordInfoIterator, null, null);
}

/**
Expand Down Expand Up @@ -668,8 +658,8 @@ public void replaceSegment(ImmutableSegment segment, @Nullable ThreadSafeMutable
if (queryableDocIds == null && _deleteRecordColumn != null) {
queryableDocIds = new ThreadSafeMutableRoaringBitmap();
}
addOrReplaceSegment((ImmutableSegmentImpl) segment, validDocIds, queryableDocIds, recordInfoIterator, oldSegment,
validDocIdsForOldSegment);
doAddOrReplaceSegment((ImmutableSegmentImpl) segment, validDocIds, queryableDocIds, recordInfoIterator,
oldSegment, validDocIdsForOldSegment);
}
if (_upsertViewManager != null) {
// When using consistency mode, the old segment's bitmap is updated in place, so we get the validDocIds after
Expand Down
3 changes: 3 additions & 0 deletions ...va/org/apache/pinot/segment/local/upsert/ConcurrentMapPartitionUpsertMetadataManager.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,6 +71,9 @@ protected long getNumPrimaryKeys() {
protected void doAddOrReplaceSegment(ImmutableSegmentImpl segment, ThreadSafeMutableRoaringBitmap validDocIds,
@Nullable ThreadSafeMutableRoaringBitmap queryableDocIds, Iterator<RecordInfo> recordInfoIterator,
@Nullable IndexSegment oldSegment, @Nullable MutableRoaringBitmap validDocIdsForOldSegment) {
if (_partialUpsertHandler != null) {
recordInfoIterator = resolveComparisonTies(recordInfoIterator, _hashFunction);
}
Comment on lines +74 to +76
Copy link
Copy Markdown
Collaborator

@J-HowHuang J-HowHuang Jun 1, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For this method resolveComparisonTies can we try to not construct the entire Map in-memory here just to get the iterator?

We have something similar in BasePartitionDedupMetadataManager#addOrReplaceSegment, and it's creating the iterator properly via some readers without allocating the memory all at once.

Copy link
Copy Markdown
Collaborator

@J-HowHuang J-HowHuang Jun 1, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Okay it seems we need to do a deduplication here on all records from recordInfoIterator, so a huge on-heap allocation is inevitable.

String segmentName = segment.getSegmentName();
segment.enableUpsert(this, validDocIds, queryableDocIds);

Expand Down
6 changes: 4 additions & 2 deletions ...segment/local/upsert/ConcurrentMapPartitionUpsertMetadataManagerForConsistentDeletes.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -90,7 +90,9 @@ protected long getNumPrimaryKeys() {
protected void doAddOrReplaceSegment(ImmutableSegmentImpl segment, ThreadSafeMutableRoaringBitmap validDocIds,
@Nullable ThreadSafeMutableRoaringBitmap queryableDocIds, Iterator<RecordInfo> recordInfoIterator,
@Nullable IndexSegment oldSegment, @Nullable MutableRoaringBitmap validDocIdsForOldSegment) {
if (_partialUpsertHandler == null) {
if (_partialUpsertHandler != null) {
recordInfoIterator = resolveComparisonTies(recordInfoIterator, _hashFunction);
} else {
// for full upsert, we are de-duping primary key once here to make sure that we are not adding
// primary-key multiple times and subtracting just once in removeSegment.
// for partial-upsert, we call this method in base class.
Expand Down Expand Up @@ -298,7 +300,7 @@ public void replaceSegment(ImmutableSegment segment, @Nullable ThreadSafeMutable
if (queryableDocIds == null && _deleteRecordColumn != null) {
queryableDocIds = new ThreadSafeMutableRoaringBitmap();
}
addOrReplaceSegment((ImmutableSegmentImpl) segment, validDocIds, queryableDocIds, recordInfoIterator,
doAddOrReplaceSegment((ImmutableSegmentImpl) segment, validDocIds, queryableDocIds, recordInfoIterator,
oldSegment, validDocIdsForOldSegment);
}
if (validDocIdsForOldSegment != null && !validDocIdsForOldSegment.isEmpty()) {
Expand Down
Loading