diff --git a/src/mito2/src/read/scan_util.rs b/src/mito2/src/read/scan_util.rs
index 79c2ad4478..df790d191a 100644
--- a/src/mito2/src/read/scan_util.rs
+++ b/src/mito2/src/read/scan_util.rs
@@ -172,7 +172,6 @@ pub(crate) fn scan_file_ranges(
             let build_reader_start = Instant::now();
             let reader = range.reader(stream_ctx.input.series_row_selector).await?;
             let build_cost = build_reader_start.elapsed();
-            // common_telemetry::info!("DEBUG_SCAN: scan file ranges, build reader, file_id: {}, index: {:?}, build_cost: {:?}", range.file_handle().file_id(), index, build_cost);
             part_metrics.inc_build_reader_cost(build_cost);
             let compat_batch = range.compat_batch();
             let mut source = Source::PruneReader(reader);
diff --git a/src/mito2/src/read/seq_scan.rs b/src/mito2/src/read/seq_scan.rs
index 5dd80f3d52..04baddd4b0 100644
--- a/src/mito2/src/read/seq_scan.rs
+++ b/src/mito2/src/read/seq_scan.rs
@@ -197,7 +197,7 @@ impl SeqScan {
 
         let stream_ctx = self.stream_ctx.clone();
         let semaphore = if self.properties.target_partitions() > self.properties.num_partitions() {
-            // We can use addtional tasks to read the data if we have more target partitions than acutal partitions.
+            // We can use additional tasks to read the data if we have more target partitions than actual partitions.
             // This semaphore is partition level.
             // We don't use a global semaphore to avoid a partition waiting for others. The final concurrency
             // of tasks usually won't exceed the target partitions a lot as compaction can reduce the number of
diff --git a/src/query/src/optimizer/parallelize_scan.rs b/src/query/src/optimizer/parallelize_scan.rs
index 296cd6c46f..03aadafce6 100644
--- a/src/query/src/optimizer/parallelize_scan.rs
+++ b/src/query/src/optimizer/parallelize_scan.rs
@@ -108,10 +108,10 @@ impl ParallelizeScan {
         Ok(result)
     }
 
-    // TODO(yingwen): Update comment.
     /// Distribute [`PartitionRange`]s to each partition.
     ///
-    /// Currently we use a simple round-robin strategy to assign ranges to partitions.
+    /// Currently we assign ranges to partitions according to their rows so each partition
+    /// has similar number of rows.
     /// This method may return partitions with smaller number than `expected_partition_num`
     /// if the number of ranges is smaller than `expected_partition_num`. But this will
     /// return at least one partition.