oom처리

2026-04-15 12:01:53 +09:00
parent 4fcc645f63
commit 0f7d794a38
5 changed files with 193 additions and 165 deletions
--- a/shp-exporter/CLAUDE.md
+++ b/shp-exporter/CLAUDE.md
@@ -4,14 +4,14 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co

 ## Project Overview

-Spring Boot 3.5.7 CLI application that converts PostgreSQL PostGIS spatial data to ESRI shapefiles and GeoJSON formats. The application uses **Spring Batch** for memory-efficient processing of large datasets (1M+ records) and supports automatic GeoServer layer registration via REST API.
+Spring Boot 3.5.7 / Java 21 CLI application that converts PostgreSQL PostGIS spatial data to ESRI shapefiles and GeoJSON formats. The application uses **Spring Batch** for memory-efficient processing of large datasets (1M+ records) and supports automatic GeoServer layer registration via REST API.

 **Key Features**:
 - Memory-optimized batch processing (90-95% reduction: 2-13GB → 150-200MB)
 - Chunk-based streaming with cursor pagination (fetch-size: 1000)
 - Automatic geometry validation and type conversion (MultiPolygon → Polygon)
 - Coordinate system validation (EPSG:5186 Korean 2000 / Central Belt)
- Dual execution modes: Spring Batch (recommended) and Legacy mode
+- Three execution modes: Spring Batch (recommended), Legacy, and GeoServer registration-only

 ## Build and Run Commands

@@ -25,6 +25,8 @@ Spring Boot 3.5.7 CLI application that converts PostgreSQL PostGIS spatial data

 Output: `build/libs/shp-exporter.jar` (fixed name, no version suffix)

+> **Note**: The `Dockerfile` currently references `shp-exporter-v2.jar` in its `COPY` step, which does not match the actual build output. Update the Dockerfile if building a Docker image.
+
 ### Run Application

 #### Spring Batch Mode (Recommended)
@@ -113,6 +115,7 @@ ConverterCommandLineRunner
        → JdbcCursorItemReader (fetch-size: 1000)
        → FeatureConversionProcessor (InferenceResult → SimpleFeature)
        → StreamingShapefileWriter (chunk-based append)
+    → Step 2-1: PostShapefileUpdateTasklet (post-export DB UPDATE hook)
    → Step 3: generateGeoJsonStep (chunk-oriented, same pattern)
    → Step 4: CreateZipTasklet (creates .zip for GeoServer)
    → Step 5: GeoServerRegistrationTasklet (conditional, if --geoserver.enabled=true)
@@ -379,6 +382,21 @@ public Step myNewStep(JobRepository jobRepository,
 ```
 4. **Always include `BatchExecutionHistoryListener`** to track execution metrics

+### Post-Export DB Hook (`PostShapefileUpdateTasklet`)
+
+`PostShapefileUpdateTasklet` runs immediately after `generateShapefileStep` and is designed as a placeholder for running UPDATE SQL after shapefile export (e.g., marking rows as exported). The SQL body is intentionally left as a `// TODO` — add your UPDATE statement inside `execute()`:
+
+```java
+// batch/tasklet/PostShapefileUpdateTasklet.java
+int updated = jdbcTemplate.update(
+    "UPDATE some_table SET status = 'EXPORTED' WHERE batch_id = ANY(?)",
+    ps -> {
+      ps.setArray(1, ps.getConnection().createArrayOf("bigint", batchIdList.toArray()));
+    });
+```
+
+Job parameters available: `inferenceId` (String), `batchIds` (comma-separated String → `List<Long>`).
+
 ### Modifying ItemReader Configuration

 ItemReaders are **not thread-safe**. Each step requires its own instance:
--- a/shp-exporter/Dockerfile
+++ b/shp-exporter/Dockerfile
@@ -30,7 +30,7 @@ ENV GEOSERVER_USERNAME=""
 ENV GEOSERVER_PASSWORD=""

 ENTRYPOINT ["java", \
-  "-Xmx4g", "-Xms512m", \
+  "-Xmx128g", "-Xms8g", \
  "-XX:+UseG1GC", \
  "-XX:MaxGCPauseMillis=200", \
  "-XX:G1HeapRegionSize=16m", \
--- a/shp-exporter/src/main/java/com/kamco/makesample/batch/config/MergedModeJobConfig.java
+++ b/shp-exporter/src/main/java/com/kamco/makesample/batch/config/MergedModeJobConfig.java
@@ -6,12 +6,15 @@ import com.kamco.makesample.batch.processor.FeatureConversionProcessor;
 import com.kamco.makesample.batch.tasklet.CreateZipTasklet;
 import com.kamco.makesample.batch.tasklet.GeoServerRegistrationTasklet;
 import com.kamco.makesample.batch.tasklet.GeometryTypeValidationTasklet;
+import com.kamco.makesample.batch.tasklet.PostShapefileUpdateTasklet;
 import com.kamco.makesample.batch.writer.MapIdGeoJsonWriter;
 import com.kamco.makesample.batch.writer.MapIdShapefileWriter;
 import com.kamco.makesample.batch.writer.StreamingGeoJsonWriter;
 import com.kamco.makesample.batch.writer.StreamingShapefileWriter;
 import com.kamco.makesample.model.InferenceResult;
+
 import java.util.Arrays;
+
 import org.geotools.api.feature.simple.SimpleFeature;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -52,33 +55,35 @@ public class MergedModeJobConfig {
  /**
   * MERGED 모드 Job 정의
   *
-   * @param jobRepository JobRepository
+   * @param jobRepository            JobRepository
   * @param validateGeometryTypeStep Geometry type 검증 Step
-   * @param generateShapefileStep Shapefile 생성 Step
-   * @param generateGeoJsonStep GeoJSON 생성 Step
-   * @param createZipStep ZIP 생성 Step
-   * @param registerToGeoServerStep GeoServer 등록 Step (merge 폴더의 shapefile만)
-   * @param generateMapIdFilesStep Map ID별 파일 생성 Step (병렬 처리)
+   * @param generateShapefileStep    Shapefile 생성 Step
+   * @param generateGeoJsonStep      GeoJSON 생성 Step
+   * @param createZipStep            ZIP 생성 Step
+   * @param registerToGeoServerStep  GeoServer 등록 Step (merge 폴더의 shapefile만)
+   * @param generateMapIdFilesStep   Map ID별 파일 생성 Step (병렬 처리)
   * @return Job
   */
  @Bean
  public Job mergedModeJob(
-      JobRepository jobRepository,
-      Step validateGeometryTypeStep,
-      Step generateShapefileStep,
-      Step generateGeoJsonStep,
-      Step createZipStep,
-      Step registerToGeoServerStep,
-      Step generateMapIdFilesStep) {
+    JobRepository jobRepository,
+    Step validateGeometryTypeStep,
+    Step generateShapefileStep,
+    Step postShapefileUpdateStep,
+    Step generateGeoJsonStep,
+    Step createZipStep,
+    Step registerToGeoServerStep,
+    Step generateMapIdFilesStep) {

    return new JobBuilder("mergedModeJob", jobRepository)
-        .start(validateGeometryTypeStep)
-        .next(generateShapefileStep)
-        .next(generateGeoJsonStep)
-        .next(createZipStep)
-        .next(registerToGeoServerStep) // Conditional execution
-        .next(generateMapIdFilesStep) // Map ID별 개별 파일 생성
-        .build();
+      .start(validateGeometryTypeStep)
+      .next(generateShapefileStep)
+      .next(generateGeoJsonStep)
+      .next(createZipStep)
+      .next(registerToGeoServerStep) // Conditional execution
+      .next(generateMapIdFilesStep) // Map ID별 개별 파일 생성
+      .next(postShapefileUpdateStep) // Shapefile 생성 후 UPDATE 실행
+      .build();
  }

  /**
@@ -86,23 +91,23 @@ public class MergedModeJobConfig {
   *
   * <p>Shapefile은 homogeneous geometry type을 요구하므로 사전 검증
   *
-   * @param jobRepository JobRepository
+   * @param jobRepository      JobRepository
   * @param transactionManager TransactionManager
-   * @param validationTasklet GeometryTypeValidationTasklet
-   * @param historyListener BatchExecutionHistoryListener
+   * @param validationTasklet  GeometryTypeValidationTasklet
+   * @param historyListener    BatchExecutionHistoryListener
   * @return Step
   */
  @Bean
  public Step validateGeometryTypeStep(
-      JobRepository jobRepository,
-      PlatformTransactionManager transactionManager,
-      GeometryTypeValidationTasklet validationTasklet,
-      BatchExecutionHistoryListener historyListener) {
+    JobRepository jobRepository,
+    PlatformTransactionManager transactionManager,
+    GeometryTypeValidationTasklet validationTasklet,
+    BatchExecutionHistoryListener historyListener) {

    return new StepBuilder("validateGeometryTypeStep", jobRepository)
-        .tasklet(validationTasklet, transactionManager)
-        .listener(historyListener)
-        .build();
+      .tasklet(validationTasklet, transactionManager)
+      .listener(historyListener)
+      .build();
  }

  /**
@@ -116,32 +121,54 @@ public class MergedModeJobConfig {
   *   <li>Writer: StreamingShapefileWriter (chunk 단위 쓰기)
   * </ul>
   *
-   * @param jobRepository JobRepository
-   * @param transactionManager TransactionManager
-   * @param shapefileReader ItemReader (Shapefile용)
+   * @param jobRepository              JobRepository
+   * @param transactionManager         TransactionManager
+   * @param shapefileReader            ItemReader (Shapefile용)
   * @param featureConversionProcessor ItemProcessor
-   * @param shapefileWriter ItemWriter
-   * @param chunkSize Chunk size (default: 1000)
-   * @param historyListener BatchExecutionHistoryListener
+   * @param shapefileWriter            ItemWriter
+   * @param chunkSize                  Chunk size (default: 1000)
+   * @param historyListener            BatchExecutionHistoryListener
   * @return Step
   */
  @Bean
  public Step generateShapefileStep(
-      JobRepository jobRepository,
-      PlatformTransactionManager transactionManager,
-      JdbcCursorItemReader<InferenceResult> shapefileReader,
-      FeatureConversionProcessor featureConversionProcessor,
-      StreamingShapefileWriter shapefileWriter,
-      @Value("${converter.batch.chunk-size:1000}") int chunkSize,
-      BatchExecutionHistoryListener historyListener) {
+    JobRepository jobRepository,
+    PlatformTransactionManager transactionManager,
+    JdbcCursorItemReader<InferenceResult> shapefileReader,
+    FeatureConversionProcessor featureConversionProcessor,
+    StreamingShapefileWriter shapefileWriter,
+    @Value("${converter.batch.chunk-size:1000}") int chunkSize,
+    BatchExecutionHistoryListener historyListener) {

    return new StepBuilder("generateShapefileStep", jobRepository)
-        .<InferenceResult, SimpleFeature>chunk(chunkSize, transactionManager)
-        .reader(shapefileReader)
-        .processor(featureConversionProcessor)
-        .writer(shapefileWriter)
-        .listener(historyListener)
-        .build();
+      .<InferenceResult, SimpleFeature>chunk(chunkSize, transactionManager)
+      .reader(shapefileReader)
+      .processor(featureConversionProcessor)
+      .writer(shapefileWriter)
+      .listener(historyListener)
+      .build();
+  }
+
+  /**
+   * Step 2-1: Shapefile 생성 후 UPDATE 실행
+   *
+   * @param jobRepository              JobRepository
+   * @param transactionManager         TransactionManager
+   * @param postShapefileUpdateTasklet PostShapefileUpdateTasklet
+   * @param historyListener            BatchExecutionHistoryListener
+   * @return Step
+   */
+  @Bean
+  public Step postShapefileUpdateStep(
+    JobRepository jobRepository,
+    PlatformTransactionManager transactionManager,
+    PostShapefileUpdateTasklet postShapefileUpdateTasklet,
+    BatchExecutionHistoryListener historyListener) {
+
+    return new StepBuilder("postShapefileUpdateStep", jobRepository)
+      .tasklet(postShapefileUpdateTasklet, transactionManager)
+      .listener(historyListener)
+      .build();
  }

  /**
@@ -149,54 +176,54 @@ public class MergedModeJobConfig {
   *
   * <p>Shapefile과 동일한 데이터를 GeoJSON 형식으로 출력
   *
-   * @param jobRepository JobRepository
-   * @param transactionManager TransactionManager
-   * @param geoJsonReader ItemReader (GeoJSON용 - 별도 인스턴스)
+   * @param jobRepository              JobRepository
+   * @param transactionManager         TransactionManager
+   * @param geoJsonReader              ItemReader (GeoJSON용 - 별도 인스턴스)
   * @param featureConversionProcessor ItemProcessor (재사용)
-   * @param geoJsonWriter ItemWriter
-   * @param chunkSize Chunk size
-   * @param historyListener BatchExecutionHistoryListener
+   * @param geoJsonWriter              ItemWriter
+   * @param chunkSize                  Chunk size
+   * @param historyListener            BatchExecutionHistoryListener
   * @return Step
   */
  @Bean
  public Step generateGeoJsonStep(
-      JobRepository jobRepository,
-      PlatformTransactionManager transactionManager,
-      JdbcCursorItemReader<InferenceResult> geoJsonReader,
-      FeatureConversionProcessor featureConversionProcessor,
-      StreamingGeoJsonWriter geoJsonWriter,
-      @Value("${converter.batch.chunk-size:1000}") int chunkSize,
-      BatchExecutionHistoryListener historyListener) {
+    JobRepository jobRepository,
+    PlatformTransactionManager transactionManager,
+    JdbcCursorItemReader<InferenceResult> geoJsonReader,
+    FeatureConversionProcessor featureConversionProcessor,
+    StreamingGeoJsonWriter geoJsonWriter,
+    @Value("${converter.batch.chunk-size:1000}") int chunkSize,
+    BatchExecutionHistoryListener historyListener) {

    return new StepBuilder("generateGeoJsonStep", jobRepository)
-        .<InferenceResult, SimpleFeature>chunk(chunkSize, transactionManager)
-        .reader(geoJsonReader)
-        .processor(featureConversionProcessor)
-        .writer(geoJsonWriter)
-        .listener(historyListener)
-        .build();
+      .<InferenceResult, SimpleFeature>chunk(chunkSize, transactionManager)
+      .reader(geoJsonReader)
+      .processor(featureConversionProcessor)
+      .writer(geoJsonWriter)
+      .listener(historyListener)
+      .build();
  }

  /**
   * Step 4: ZIP 파일 생성
   *
-   * @param jobRepository JobRepository
+   * @param jobRepository      JobRepository
   * @param transactionManager TransactionManager
-   * @param createZipTasklet CreateZipTasklet
-   * @param historyListener BatchExecutionHistoryListener
+   * @param createZipTasklet   CreateZipTasklet
+   * @param historyListener    BatchExecutionHistoryListener
   * @return Step
   */
  @Bean
  public Step createZipStep(
-      JobRepository jobRepository,
-      PlatformTransactionManager transactionManager,
-      CreateZipTasklet createZipTasklet,
-      BatchExecutionHistoryListener historyListener) {
+    JobRepository jobRepository,
+    PlatformTransactionManager transactionManager,
+    CreateZipTasklet createZipTasklet,
+    BatchExecutionHistoryListener historyListener) {

    return new StepBuilder("createZipStep", jobRepository)
-        .tasklet(createZipTasklet, transactionManager)
-        .listener(historyListener)
-        .build();
+      .tasklet(createZipTasklet, transactionManager)
+      .listener(historyListener)
+      .build();
  }

  /**
@@ -204,23 +231,23 @@ public class MergedModeJobConfig {
   *
   * <p>Conditional execution: geoserver.enabled=true일 때만 실행
   *
-   * @param jobRepository JobRepository
-   * @param transactionManager TransactionManager
+   * @param jobRepository       JobRepository
+   * @param transactionManager  TransactionManager
   * @param registrationTasklet GeoServerRegistrationTasklet
-   * @param historyListener BatchExecutionHistoryListener
+   * @param historyListener     BatchExecutionHistoryListener
   * @return Step
   */
  @Bean
  public Step registerToGeoServerStep(
-      JobRepository jobRepository,
-      PlatformTransactionManager transactionManager,
-      GeoServerRegistrationTasklet registrationTasklet,
-      BatchExecutionHistoryListener historyListener) {
+    JobRepository jobRepository,
+    PlatformTransactionManager transactionManager,
+    GeoServerRegistrationTasklet registrationTasklet,
+    BatchExecutionHistoryListener historyListener) {

    return new StepBuilder("registerToGeoServerStep", jobRepository)
-        .tasklet(registrationTasklet, transactionManager)
-        .listener(historyListener)
-        .build();
+      .tasklet(registrationTasklet, transactionManager)
+      .listener(historyListener)
+      .build();
  }

  /**
@@ -229,21 +256,21 @@ public class MergedModeJobConfig {
   * <p>각 map_id마다 개별 shapefile과 geojson 파일을 순차적으로 생성합니다. SyncTaskExecutor를 명시적으로 지정하여 병렬 실행을 방지하고
   * DB connection pool 고갈 방지
   *
-   * @param jobRepository JobRepository
-   * @param partitioner MapIdPartitioner
+   * @param jobRepository   JobRepository
+   * @param partitioner     MapIdPartitioner
   * @param mapIdWorkerStep Worker Step (각 파티션에서 실행)
   * @return Partitioned Step
   */
  @Bean
  public Step generateMapIdFilesStep(
-      JobRepository jobRepository, MapIdPartitioner partitioner, Step mapIdWorkerStep) {
+    JobRepository jobRepository, MapIdPartitioner partitioner, Step mapIdWorkerStep) {

    return new StepBuilder("generateMapIdFilesStep", jobRepository)
-        .partitioner("mapIdWorker", partitioner)
-        .step(mapIdWorkerStep)
-        .taskExecutor(new SyncTaskExecutor()) // 명시적으로 순차 실행 지정
-        .listener(partitioner) // Register partitioner as StepExecutionListener
-        .build();
+      .partitioner("mapIdWorker", partitioner)
+      .step(mapIdWorkerStep)
+      .taskExecutor(new SyncTaskExecutor()) // 명시적으로 순차 실행 지정
+      .listener(partitioner) // Register partitioner as StepExecutionListener
+      .build();
  }

  /**
@@ -251,39 +278,39 @@ public class MergedModeJobConfig {
   *
   * <p>각 파티션에서 실행되며, 해당 map_id의 데이터를 읽어 shapefile과 geojson을 동시에 생성합니다.
   *
-   * @param jobRepository JobRepository
-   * @param transactionManager TransactionManager
-   * @param mapIdModeReader ItemReader (map_id별)
+   * @param jobRepository              JobRepository
+   * @param transactionManager         TransactionManager
+   * @param mapIdModeReader            ItemReader (map_id별)
   * @param featureConversionProcessor ItemProcessor
-   * @param mapIdShapefileWriter Shapefile Writer
-   * @param mapIdGeoJsonWriter GeoJSON Writer
-   * @param chunkSize Chunk size
-   * @param historyListener BatchExecutionHistoryListener
+   * @param mapIdShapefileWriter       Shapefile Writer
+   * @param mapIdGeoJsonWriter         GeoJSON Writer
+   * @param chunkSize                  Chunk size
+   * @param historyListener            BatchExecutionHistoryListener
   * @return Worker Step
   */
  @Bean
  public Step mapIdWorkerStep(
-      JobRepository jobRepository,
-      PlatformTransactionManager transactionManager,
-      JdbcCursorItemReader<InferenceResult> mapIdModeReader,
-      FeatureConversionProcessor featureConversionProcessor,
-      MapIdShapefileWriter mapIdShapefileWriter,
-      MapIdGeoJsonWriter mapIdGeoJsonWriter,
-      @Value("${converter.batch.chunk-size:1000}") int chunkSize,
-      BatchExecutionHistoryListener historyListener) {
+    JobRepository jobRepository,
+    PlatformTransactionManager transactionManager,
+    JdbcCursorItemReader<InferenceResult> mapIdModeReader,
+    FeatureConversionProcessor featureConversionProcessor,
+    MapIdShapefileWriter mapIdShapefileWriter,
+    MapIdGeoJsonWriter mapIdGeoJsonWriter,
+    @Value("${converter.batch.chunk-size:1000}") int chunkSize,
+    BatchExecutionHistoryListener historyListener) {

    // CompositeItemWriter로 shapefile과 geojson 동시 생성
    CompositeItemWriter<SimpleFeature> compositeWriter = new CompositeItemWriter<>();
    compositeWriter.setDelegates(Arrays.asList(mapIdShapefileWriter, mapIdGeoJsonWriter));

    return new StepBuilder("mapIdWorkerStep", jobRepository)
-            .<InferenceResult, SimpleFeature>chunk(chunkSize, transactionManager)
-            .reader(mapIdModeReader)
-            .processor(featureConversionProcessor)
-            .writer(compositeWriter)
-            .stream(mapIdShapefileWriter)
-            .stream(mapIdGeoJsonWriter)
-            .listener(historyListener)
-            .build();
+      .<InferenceResult, SimpleFeature>chunk(chunkSize, transactionManager)
+      .reader(mapIdModeReader)
+      .processor(featureConversionProcessor)
+      .writer(compositeWriter)
+      .stream(mapIdShapefileWriter)
+      .stream(mapIdGeoJsonWriter)
+      .listener(historyListener)
+      .build();
  }
 }
--- a/shp-exporter/src/main/java/com/kamco/makesample/batch/tasklet/GeometryTypeValidationTasklet.java
+++ b/shp-exporter/src/main/java/com/kamco/makesample/batch/tasklet/GeometryTypeValidationTasklet.java
@@ -139,6 +139,8 @@ public class GeometryTypeValidationTasklet implements Tasklet {
        SELECT COUNT(*) as valid_count
        FROM inference_results_testing
        WHERE batch_id = ANY(?)
+          AND after_c IS NOT NULL
+          AND after_p IS NOT NULL
          AND geometry IS NOT NULL
          AND ST_GeometryType(geometry) IN ('ST_Polygon', 'ST_MultiPolygon')
          AND ST_SRID(geometry) = 5186
--- a/shp-exporter/src/main/java/com/kamco/makesample/batch/writer/StreamingShapefileWriter.java
+++ b/shp-exporter/src/main/java/com/kamco/makesample/batch/writer/StreamingShapefileWriter.java
@@ -10,13 +10,11 @@ import java.nio.file.Paths;
 import java.util.HashMap;
 import java.util.List;
 import java.util.Map;
-import org.geotools.api.data.SimpleFeatureStore;
+import org.geotools.api.data.FeatureWriter;
 import org.geotools.api.data.Transaction;
 import org.geotools.api.feature.simple.SimpleFeature;
 import org.geotools.api.feature.simple.SimpleFeatureType;
 import org.geotools.api.referencing.crs.CoordinateReferenceSystem;
-import org.geotools.data.DefaultTransaction;
-import org.geotools.data.collection.ListFeatureCollection;
 import org.geotools.data.shapefile.ShapefileDataStore;
 import org.geotools.data.shapefile.ShapefileDataStoreFactory;
 import org.slf4j.Logger;
@@ -69,8 +67,7 @@ public class StreamingShapefileWriter implements ItemStreamWriter<SimpleFeature>
  private String outputPath;

  private ShapefileDataStore dataStore;
-  private Transaction transaction;
-  private SimpleFeatureStore featureStore;
+  private FeatureWriter<SimpleFeatureType, SimpleFeature> featureWriter;
  private SimpleFeatureType featureType;

  private int chunkCount = 0;
@@ -145,13 +142,9 @@ public class StreamingShapefileWriter implements ItemStreamWriter<SimpleFeature>
      dataStore = (ShapefileDataStore) factory.createNewDataStore(params);
      dataStore.createSchema(featureType);

-      // Transaction 시작
-      transaction = new DefaultTransaction("create");
-
-      // FeatureStore 가져오기
+      // FeatureWriter를 append 모드로 직접 열기 (Diff 누적 없이 파일에 직접 씀)
      String typeName = dataStore.getTypeNames()[0];
-      featureStore = (SimpleFeatureStore) dataStore.getFeatureSource(typeName);
-      featureStore.setTransaction(transaction);
+      featureWriter = dataStore.getFeatureWriterAppend(typeName, Transaction.AUTO_COMMIT);

      startTimeMs = System.currentTimeMillis();
      log.info("ShapefileDataStore initialized successfully");
@@ -172,10 +165,13 @@ public class StreamingShapefileWriter implements ItemStreamWriter<SimpleFeature>
    int itemCount = items.size();
    totalRecordCount += itemCount;

-    // FeatureStore에 추가 - GeoTools ShapefileDataStore는 Diff 없이 파일에 직접 씀
-    // 트랜잭션은 afterStep()에서 단일 커밋 (per-chunk 커밋 시 setTransaction()이 .shx 재스캔 → O(n²))
-    ListFeatureCollection collection = new ListFeatureCollection(featureType, items);
-    featureStore.addFeatures(collection);
+    // FeatureWriter로 직접 append - Diff 누적 없이 O(1) per record
+    for (SimpleFeature feature : items) {
+      SimpleFeature newFeature = featureWriter.next();
+      newFeature.setAttributes(feature.getAttributes());
+      newFeature.setDefaultGeometry(feature.getDefaultGeometry());
+      featureWriter.write();
+    }

    if (chunkCount % LOG_INTERVAL_CHUNKS == 0) {
      logProgress();
@@ -191,15 +187,10 @@ public class StreamingShapefileWriter implements ItemStreamWriter<SimpleFeature>
        chunkCount);

    try {
-      if (transaction != null) {
-        transaction.commit();
-        log.info("Final transaction committed successfully");
-      }
-    } catch (IOException e) {
-      log.error("Failed to commit final transaction", e);
-      throw new ItemStreamException("Failed to commit shapefile transaction", e);
-    } finally {
      cleanup();
+    } catch (Exception e) {
+      log.error("Failed to close shapefile writer", e);
+      throw new ItemStreamException("Failed to close shapefile writer", e);
    }
  }

@@ -212,23 +203,13 @@ public class StreamingShapefileWriter implements ItemStreamWriter<SimpleFeature>
  public void onError(Exception exception, Chunk<? extends SimpleFeature> chunk) {
    log.error("Error writing chunk #{}: {}", chunkCount, exception.getMessage(), exception);

-    try {
-      if (transaction != null) {
-        transaction.rollback();
-        log.info("Transaction rolled back due to error");
-      }
+    cleanup();

-      // 부분 파일 삭제
-      File shpFile = new File(outputPath);
-      if (shpFile.exists()) {
-        shpFile.delete();
-        log.info("Deleted partial shapefile: {}", outputPath);
-      }
-
-    } catch (IOException e) {
-      log.error("Failed to rollback transaction", e);
-    } finally {
-      cleanup();
+    // 부분 파일 삭제
+    File shpFile = new File(outputPath);
+    if (shpFile.exists()) {
+      shpFile.delete();
+      log.info("Deleted partial shapefile: {}", outputPath);
    }
  }

@@ -264,13 +245,13 @@ public class StreamingShapefileWriter implements ItemStreamWriter<SimpleFeature>
  }

  private void cleanup() {
-    if (transaction != null) {
+    if (featureWriter != null) {
      try {
-        transaction.close();
+        featureWriter.close();
      } catch (IOException e) {
-        log.warn("Failed to close transaction", e);
+        log.warn("Failed to close feature writer", e);
      }
-      transaction = null;
+      featureWriter = null;
    }

    if (dataStore != null) {