Is batch.Flush() Atomic?

I have a Table in Clickhouse as a ReplicatedMergeTree.

I am inserting large parquet files and I want to confirm whether Batch.flush() offers Atomicity as a guarantee?

Is it possible flushing will only insert rows partially?

Thank you for any assistance , relevant answers will help a lot!

Start by reading the Batch.flush() implementation and checking how ClickHouse handles inserts into ReplicatedMergeTree tables. Determine whether a flush can produce partial rows for large parquet inputs, then document the atomicity guarantee and any relevant limitations or conditions.

Written by the indexing model from the issue text.