하드링크 수정

src/main/java/com/kamco/cd/training/train/service/TmpDatasetService.java

@@ -20,71 +20,187 @@ public class TmpDatasetService {
   @Value("${train.docker.basePath}")
   private String trainBaseDir;
 
-  private final DataSetCountersService dataSetCountersService;
-
   /**
-   * 다른 데이터셋 파일과 이름이 겹치면 그 파일은 skip함
+   * train, val, test 폴더별로 link
    *
-   * @param uid
+   * @param uid 임시폴더 uuid
-   * @param type
+   * @param type train, val, test
-   * @param links
+   * @param links tif pull path
    * @return
    * @throws IOException
    */
-  public String buildTmpDatasetSymlink(String uid, String type, List<ModelTrainLinkDto> links)
+  public void buildTmpDatasetHardlink(String uid, String type, List<ModelTrainLinkDto> links)
       throws IOException {
 
-    if (uid == null || uid.isBlank()) throw new IOException("uid is empty");
+    if (links == null || links.isEmpty()) {
-    if (type == null || type.isBlank()) throw new IOException("type is empty");
+      throw new IOException("links is empty");
-    if (links == null || links.isEmpty()) throw new IOException("links is empty");
+    }
 
-    log.info("========== buildTmpDatasetHardlink MERGE START ==========");
+    Path tmp = Path.of(trainBaseDir, "tmp", uid);
-    log.info("uid={}, type={}, links.size={}", uid, type, links.size());
+
+    long hardlinksMade = 0;
+
+    for (ModelTrainLinkDto dto : links) {
+
+      if (type == null) {
+        log.warn("SKIP - trainType null: {}", dto);
+        continue;
+      }
+
+      // type별 디렉토리 생성
+      Files.createDirectories(tmp.resolve(type).resolve("input1"));
+      Files.createDirectories(tmp.resolve(type).resolve("input2"));
+      Files.createDirectories(tmp.resolve(type).resolve("label"));
+      Files.createDirectories(tmp.resolve(type).resolve("label-json"));
+
+      // comparePath → input1
+      hardlinksMade += link(tmp, type, "input1", dto.getComparePath());
+
+      // targetPath → input2
+      hardlinksMade += link(tmp, type, "input2", dto.getTargetPath());
+
+      // labelPath → label
+      hardlinksMade += link(tmp, type, "label", dto.getLabelPath());
+
+      // geoJsonPath -> label-json
+      hardlinksMade += link(tmp, type, "label-json", dto.getGeoJsonPath());
+    }
+
+    if (hardlinksMade == 0) {
+      throw new IOException("No hardlinks created.");
+    }
+
+    log.info("tmp dataset created: {}, hardlinksMade={}", tmp, hardlinksMade);
+  }
+
+  private long link(Path tmp, String type, String part, String fullPath) throws IOException {
+
+    if (fullPath == null || fullPath.isBlank()) return 0;
+
+    Path src = Path.of(fullPath);
+
+    if (!Files.isRegularFile(src)) {
+      log.warn("SKIP (not file): {}", src);
+      return 0;
+    }
+
+    String fileName = src.getFileName().toString();
+    Path dst = tmp.resolve(type).resolve(part).resolve(fileName);
+
+    // 충돌 시 덮어쓰기
+    if (Files.exists(dst)) {
+      Files.delete(dst);
+    }
+
+    Files.createLink(dst, src);
+
+    return 1;
+  }
+
+  private String safe(String s) {
+    return (s == null || s.isBlank()) ? null : s.trim();
+  }
+
+  /**
+   * request 전체 폴더 link
+   *
+   * @param uid
+   * @param datasetUids
+   * @return
+   * @throws IOException
+   */
+  public String buildTmpDatasetSymlink(String uid, List<String> datasetUids) throws IOException {
+
+    log.info("========== buildTmpDatasetHardlink START ==========");
+    log.info("uid={}", uid);
+    log.info("datasetUids={}", datasetUids);
+    log.info("requestDir(raw)={}", requestDir);
 
     Path BASE = toPath(requestDir);
     Path tmp = Path.of(trainBaseDir, "tmp", uid);
 
-    long hardlinksMade = 0;
+    log.info("BASE={}", BASE);
-    long skippedCollision = 0;
+    log.info("BASE exists? {}", Files.isDirectory(BASE));
-    long noDir = 0;
+    log.info("tmp={}", tmp);
 
-    // tmp/<type>/<part> 준비
+    long noDir = 0, scannedDirs = 0, regularFiles = 0, hardlinksMade = 0;
+
+    // tmp 디렉토리 준비
+    for (String type : List.of("train", "val", "test")) {
       for (String part : List.of("input1", "input2", "label", "label-json")) {
-      Files.createDirectories(tmp.resolve(type).resolve(part));
+        Path dir = tmp.resolve(type).resolve(part);
+        Files.createDirectories(dir);
+        log.info("createDirectories: {}", dir);
+      }
     }
 
-    for (ModelTrainLinkDto dto : links) {
+    // 하드링크는 "같은 파일시스템"에서만 가능하므로 BASE/tmp가 같은 FS인지 미리 확인(권장)
-      String datasetUid = safe(dto.getDatasetUid());
+    try {
-      if (datasetUid == null) {
+      var baseStore = Files.getFileStore(BASE);
-        log.warn("SKIP dto (datasetUid null): {}", dto);
+      var tmpStore = Files.getFileStore(tmp.getParent()); // BASE/tmp
-        continue;
+      if (!baseStore.name().equals(tmpStore.name()) || !baseStore.type().equals(tmpStore.type())) {
+        throw new IOException(
+            "Hardlink requires same filesystem. baseStore="
+                + baseStore.name()
+                + "("
+                + baseStore.type()
+                + "), tmpStore="
+                + tmpStore.name()
+                + "("
+                + tmpStore.type()
+                + ")");
+      }
+    } catch (Exception e) {
+      // FileStore 비교가 환경마다 애매할 수 있어서, 여기서는 경고만 주고 실제 createLink에서 최종 판단하게 둘 수도 있음.
+      log.warn("FileStore check skipped/failed (will rely on createLink): {}", e.toString());
     }
 
-      Path srcRoot = BASE.resolve(datasetUid);
+    for (String id : datasetUids) {
+      Path srcRoot = BASE.resolve(id);
+      log.info("---- dataset id={} srcRoot={} exists? {}", id, srcRoot, Files.isDirectory(srcRoot));
 
+      for (String type : List.of("train", "val", "test")) {
         for (String part : List.of("input1", "input2", "label", "label-json")) {
 
           Path srcDir = srcRoot.resolve(type).resolve(part);
           if (!Files.isDirectory(srcDir)) {
+            log.warn("SKIP (not directory): {}", srcDir);
             noDir++;
             continue;
           }
 
-        // 하위폴더까지 전부
+          scannedDirs++;
-        try (var walk = Files.walk(srcDir)) {
+          log.info("SCAN dir={}", srcDir);
-          for (Path f : walk.filter(Files::isRegularFile).toList()) {
 
-            String fileName = f.getFileName().toString();
+          try (DirectoryStream<Path> stream = Files.newDirectoryStream(srcDir)) {
-            Path dst = tmp.resolve(type).resolve(part).resolve(fileName);
+            for (Path f : stream) {
-
+              if (!Files.isRegularFile(f)) {
-            // 이름 유지 + 충돌은 skip
+                log.debug("skip non-regular file: {}", f);
-            if (Files.exists(dst)) {
-              skippedCollision++;
                 continue;
               }
 
+              regularFiles++;
+
+              String dstName = id + "__" + f.getFileName();
+              Path dst = tmp.resolve(type).resolve(part).resolve(dstName);
+
+              // dst가 남아있으면 삭제(심볼릭링크든 파일이든)
+              if (Files.exists(dst) || Files.isSymbolicLink(dst)) {
+                Files.delete(dst);
+                log.debug("deleted existing: {}", dst);
+              }
+
+              try {
+                // 하드링크 생성 (dst가 새 파일로 생기지만 inode는 f와 동일)
                 Files.createLink(dst, f);
                 hardlinksMade++;
+                log.debug("created hardlink: {} => {}", dst, f);
+              } catch (IOException e) {
+                // 여기서 바로 실패시키면 “tmp는 만들었는데 내용은 0개” 같은 상태를 방지할 수 있음
+                log.error("FAILED create hardlink: {} => {}", dst, f, e);
+                throw e;
+              }
+            }
           }
         }
       }
@@ -92,35 +208,29 @@ public class TmpDatasetService {
 
     if (hardlinksMade == 0) {
       throw new IOException(
-          "No hardlinks created. noDir=" + noDir + ", skippedCollision=" + skippedCollision);
+          "No hardlinks created. regularFiles="
+              + regularFiles
+              + ", scannedDirs="
+              + scannedDirs
+              + ", noDir="
+              + noDir);
     }
 
+    log.info("tmp dataset created: {}", tmp);
     log.info(
-        "tmp dataset merged: {} (type={}), hardlinksMade={}, skippedCollision={}, noDir={}",
+        "summary: scannedDirs={}, noDir={}, regularFiles={}, hardlinksMade={}",
-        tmp,
+        scannedDirs,
-        type,
+        noDir,
-        hardlinksMade,
+        regularFiles,
-        skippedCollision,
+        hardlinksMade);
-        noDir);
 
     return uid;
   }
 
   private static Path toPath(String p) {
-    if (p == null || p.isBlank()) {
+    if (p.startsWith("~/")) {
-      throw new IllegalArgumentException("path is null or blank");
+      return Paths.get(System.getProperty("user.home")).resolve(p.substring(2)).normalize();
     }
-    String trimmed = p.trim();
+    return Paths.get(p).toAbsolutePath().normalize();
-    if (trimmed.startsWith("~/")) {
-      return Paths.get(System.getProperty("user.home"))
-          .resolve(trimmed.substring(2))
-          .toAbsolutePath()
-          .normalize();
-    }
-    return Paths.get(trimmed).toAbsolutePath().normalize();
-  }
-
-  private static String safe(String s) {
-    return (s == null || s.isBlank()) ? null : s.trim();
   }
 }

src/main/java/com/kamco/cd/training/train/service/TrainJobService.java

@@ -266,6 +266,8 @@ public class TrainJobService {
     List<String> uids = modelTrainMngCoreService.findDatasetUid(datasetIds);
 
     try {
+      // 데이터셋 심볼링크 생성
+      //      String pathUid = tmpDatasetService.buildTmpDatasetSymlink(raw, uids);
       // train path
       List<ModelTrainLinkDto> trainList = modelTrainMngCoreService.findDatasetTrainPath(modelId);
       // validation path
@@ -274,13 +276,12 @@ public class TrainJobService {
       List<ModelTrainLinkDto> testList = modelTrainMngCoreService.findDatasetTestPath(modelId);
 
       // train 데이터셋 심볼링크 생성
-      tmpDatasetService.buildTmpDatasetSymlink(raw, "train", trainList);
+      tmpDatasetService.buildTmpDatasetHardlink(raw, "train", trainList);
       // val 데이터셋 심볼링크 생성
-      tmpDatasetService.buildTmpDatasetSymlink(raw, "val", valList);
+      tmpDatasetService.buildTmpDatasetHardlink(raw, "val", valList);
       // test 데이터셋 심볼링크 생성
-      tmpDatasetService.buildTmpDatasetSymlink(raw, "test", testList);
+      tmpDatasetService.buildTmpDatasetHardlink(raw, "test", testList);
-      // 카운트 로그
+
-      dataSetCounters.getCount(modelId);
       ModelTrainMngDto.UpdateReq updateReq = new ModelTrainMngDto.UpdateReq();
       updateReq.setRequestPath(raw);