feat: Improve slive stress tests with path scanning and better file h…

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/test/java/org/apache/hadoop/fs/slive/AppendOp.java

@@ -53,7 +53,7 @@ class AppendOp extends Operation {
    * @return Path
    */
   protected Path getAppendFile() {
-    Path fn = getFinder().getFile();
+    Path fn = getFinder().getFile("APPEND");
     return fn;
   }
 
@@ -100,7 +100,7 @@ List<OperationOutput> run(FileSystem fs) {
     } catch (FileNotFoundException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.NOT_FOUND, 1L));
-      LOG.warn("Error with appending", e);
+      LOG.warn("AppendOp failed: File not found", e);
     } catch (IOException | UnsupportedOperationException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.FAILURES, 1L));

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/test/java/org/apache/hadoop/fs/slive/ConfigExtractor.java

@@ -131,6 +131,29 @@ boolean shouldExitOnFirstError(String primary) {
     return Boolean.parseBoolean(val);
   }
 
+  /**
+   * @return true|false for whether to use existing files only
+   */
+  boolean shouldUseNewAlgorithm() {
+    return shouldUseNewAlgorithm(null);
+  }
+
+  /**
+   * @param primary
+   *          primary the initial string to be used for the value of this
+   * @return true|false for whether to use existing files only from primary,config,default (in that order)
+   */
+  boolean shouldUseNewAlgorithm(String primary) {
+    String val = primary;
+    if (val == null) {
+      val = config.get(ConfigOption.USE_NEW_ALGORITHM.getCfgOption());
+    }
+    if (val == null) {
+      val = ConfigOption.USE_NEW_ALGORITHM.getDefault().toString();
+    }
+    return Boolean.parseBoolean(val);
+  }
+
   /**
    * @return whether the mapper or reducer should wait for truncate recovery
    */
@@ -735,6 +758,7 @@ static void dumpOptions(ConfigExtractor cfg) {
     LOG.info("Operation amount = " + cfg.getOpCount());
     LOG.info("Total file limit = " + cfg.getTotalFiles());
     LOG.info("Total dir file limit = " + cfg.getDirSize());
+    LOG.info("Use new algorithm = " + cfg.shouldUseNewAlgorithm());
     {
       String read = "Read size = ";
       if (cfg.shouldReadFullFile()) {

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/test/java/org/apache/hadoop/fs/slive/ConfigOption.java

@@ -52,6 +52,10 @@ class ConfigOption<T> extends Option {
       "exitOnError", false, "Exit on first error", SLIVE_PREFIX
           + ".exit.on.error", false);
 
+  static final ConfigOption<Boolean> USE_NEW_ALGORITHM = new ConfigOption<Boolean>(
+      "useNewAlgorithm", false, "Use new algorithm for slivetest read/delete etc. operations", SLIVE_PREFIX
+          + ".use.new.algorithm", false);
+
   static final ConfigOption<Integer> FILES = new ConfigOption<Integer>(
       "files", true, "Max total number of files",
       SLIVE_PREFIX + ".total.files", 10);

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/test/java/org/apache/hadoop/fs/slive/CreateOp.java

@@ -19,6 +19,7 @@
 package org.apache.hadoop.fs.slive;
 
 import java.io.IOException;
+import org.apache.hadoop.fs.FileAlreadyExistsException;
 import java.util.List;
 import java.util.Random;
 
@@ -111,7 +112,7 @@ private int getBufferSize() {
    * @return Path
    */
   protected Path getCreateFile() {
-    Path fn = getFinder().getFile();
+    Path fn = getFinder().getFile("CREATE");
     return fn;
   }
 
@@ -164,10 +165,14 @@ List<OperationOutput> run(FileSystem fs) {
           ReportWriter.BYTES_WRITTEN, bytesWritten));
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.SUCCESSES, 1L));
+    } catch (FileAlreadyExistsException e) {
+      out.add(new OperationOutput(OutputType.LONG, getType(),
+          ReportWriter.FILE_ALREADY_EXISTS, 1L));
+      LOG.warn("CreateOp failed: File already exists", e);
     } catch (IOException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.FAILURES, 1L));
-      LOG.warn("Error with creating", e);
+      LOG.warn("CreateOp failed: IO error creating file", e);
     } finally {
       if (os != null) {
         try {

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/test/java/org/apache/hadoop/fs/slive/DeleteOp.java

@@ -49,7 +49,7 @@ class DeleteOp extends Operation {
    * Gets the file to delete
    */
   protected Path getDeleteFile() {
-    Path fn = getFinder().getFile();
+    Path fn = getFinder().getFile("DELETE");
     return fn;
   }
 
@@ -58,6 +58,7 @@ List<OperationOutput> run(FileSystem fs) {
     List<OperationOutput> out = super.run(fs);
     try {
       Path fn = getDeleteFile();
+      LOG.info("Deleting file: " + fn);
       long timeTaken = 0;
       boolean deleteStatus = false;
       {
@@ -80,7 +81,7 @@ List<OperationOutput> run(FileSystem fs) {
     } catch (FileNotFoundException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.NOT_FOUND, 1L));
-      LOG.warn("Error with deleting", e);
+      LOG.warn("DeleteOp failed: File not found", e);
     } catch (IOException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.FAILURES, 1L));

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/test/java/org/apache/hadoop/fs/slive/ListOp.java

@@ -53,7 +53,7 @@ class ListOp extends Operation {
    * @return Path
    */
   protected Path getDirectory() {
-    Path dir = getFinder().getDirectory();
+    Path dir = getFinder().getDirectory("LS");
     return dir;
   }
 
@@ -81,7 +81,7 @@ List<OperationOutput> run(FileSystem fs) {
     } catch (FileNotFoundException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.NOT_FOUND, 1L));
-      LOG.warn("Error with listing", e);
+      LOG.warn("ListOp failed: File not found", e);
     } catch (IOException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.FAILURES, 1L));

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/test/java/org/apache/hadoop/fs/slive/MkdirOp.java

@@ -83,7 +83,7 @@ List<OperationOutput> run(FileSystem fs) {
     } catch (FileNotFoundException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.NOT_FOUND, 1L));
-      LOG.warn("Error with mkdir", e);
+      LOG.warn("MkdirOp failed: File not found", e);
     } catch (IOException e) {
       out.add(new OperationOutput(OutputType.LONG, getType(),
           ReportWriter.FAILURES, 1L));

hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-jobclient/src/test/java/org/apache/hadoop/fs/slive/PathFinder.java

@@ -18,16 +18,26 @@
 
 package org.apache.hadoop.fs.slive;
 
+import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
 import java.util.Random;
+import java.util.UUID;
 
+import org.apache.hadoop.fs.FileStatus;
+import org.apache.hadoop.fs.FileSystem;
 import org.apache.hadoop.fs.Path;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
 
 /**
  * Class which generates a file or directory path using a simple random
  * generation algorithm stated in http://issues.apache.org/jira/browse/HDFS-708
  */
 class PathFinder {
 
+  private static final Logger LOG = LoggerFactory.getLogger(PathFinder.class);
+
   private enum Type {
     FILE, DIRECTORY
   }
@@ -38,11 +48,96 @@ private enum Type {
   private Path basePath;
   private ConfigExtractor config;
   private Random rnd;
+
+  // Used to store scanned existing paths
+  private List<Path> existingFiles;
+  private List<Path> existingDirs;
 
   PathFinder(ConfigExtractor cfg, Random rnd) {
     this.basePath = cfg.getDataPath();
     this.config = cfg;
     this.rnd = rnd;
+    this.existingFiles = new ArrayList<>();
+    this.existingDirs = new ArrayList<>();
+  }
+
+  /**
+   * Scan all paths under base_dir and record existing files and directories
+   */
+  private void scanBaseDirectory() {
+    try {
+      FileSystem fs = basePath.getFileSystem(config.getConfig());
+      LOG.info("Starting to scan base_dir: " + basePath);
+      // Clear existing lists
+      clearExistingPaths();
+
+      // Recursively scan directories
+      scanDirectoryRecursively(fs, basePath);
+
+      // Print summary only (avoid huge log output)
+      LOG.info("Scan complete: found " + existingFiles.size() + " files, " 
+          + existingDirs.size() + " directories");
+
+    } catch (IOException e) {
+      LOG.error("Error scanning base_dir: " + e.getMessage(), e);
+      clearExistingPaths();
+    }
+  }
+
+  private void clearExistingPaths() {
+    existingFiles.clear();
+    existingDirs.clear();
+  }
+
+  /**
+   * Recursively scan directories
+   */
+  private void scanDirectoryRecursively(FileSystem fs, Path dir) throws IOException {
+    if (!fs.exists(dir)) {
+      return;
+    }
+
+    FileStatus[] statuses = fs.listStatus(dir);
+    if (statuses == null || statuses.length == 0) {
+      return;
+    }
+
+    for (FileStatus status : statuses) {
+      Path path = status.getPath();
+      if (status.isFile()) {
+        existingFiles.add(path);
+      } else if (status.isDirectory()) {
+        existingDirs.add(path);
+        // Recursively scan subdirectories
+        scanDirectoryRecursively(fs, path);
+      }
+    }
+  }
+
+  /**
+   * Randomly select one from existing files
+   */
+  private Path getExistingFile() {
+    if (existingFiles.isEmpty()) {
+      throw new RuntimeException("No files found in base_dir, cannot perform read/delete operations");
+    }
+    int index = rnd.nextInt(existingFiles.size());
+    Path selectedFile = existingFiles.get(index);
+    LOG.info("Selected from existing files: " + selectedFile);
+    return selectedFile;
+  }
+
+  /**
+   * Randomly select one from existing directories
+   */
+  private Path getExistingDirectory() {
+    if (existingDirs.isEmpty()) {
+      throw new RuntimeException("No directories found in base_dir, cannot perform ls operations");
+    }
+    int index = rnd.nextInt(existingDirs.size());
+    Path selectedDir = existingDirs.get(index);
+    LOG.info("Selected from existing directories: " + selectedDir);
+    return selectedDir;
   }
 
   /**
@@ -62,6 +157,10 @@ private enum Type {
    * @return Path
    */
   private Path getPath(int curId, int limitPerDir, Type type) {
+    return getPath(curId, limitPerDir, type, null);
+  }
+
+  private Path getPath(int curId, int limitPerDir, Type type, String suffix) {
     if (curId <= 0) {
       return basePath;
     }
@@ -74,6 +173,9 @@ private Path getPath(int curId, int limitPerDir, Type type) {
       name = DIR_PREFIX + new Integer(curId % limitPerDir).toString();
       break;
     }
+    if (suffix != null) {
+      name += "_" + suffix;
+    }
     Path base = getPath((curId / limitPerDir), limitPerDir, Type.DIRECTORY);
     return new Path(base, name);
   }
@@ -85,6 +187,57 @@ private Path getPath(int curId, int limitPerDir, Type type) {
    * @return path
    */
   Path getFile() {
+    return getFile(null);
+  }
+
+  /**
+   * Gets a file path based on operation type and configuration
+   * 
+   * @param operationType the type of operation (can be null for backward compatibility)
+   * @return path
+   */
+  Path getFile(String operationType) {
+    boolean useNewAlgorithm = config.shouldUseNewAlgorithm();
+
+    // Handle operations that need existing files
+    if (isExistingFileOperation(operationType)) {
+      if (useNewAlgorithm) {
+        LOG.info("Use new algorithm mode: scanning base_dir for " + operationType + " operation");
+        scanBaseDirectory();
+        return getExistingFile();
+      }
+      // Fall through to original algorithm for normal mode
+    }
+
+    // Handle CREATE operation
+    if ("CREATE".equals(operationType)) {
+      if (useNewAlgorithm) {
+        LOG.info("Generating unique path for CREATE operation");
+        return generateUniquePath();
+      }
+      // Fall through to original algorithm for normal mode
+    }
+
+    // Use original algorithm for all other cases
+    LOG.info("Using original algorithm for " + (operationType != null ? operationType : "default") + " operation");
+    return generateOriginalPath();
+  }
+
+  private boolean isExistingFileOperation(String operationType) {
+    return "READ".equals(operationType) || "DELETE".equals(operationType) || 
+           "TRUNCATE".equals(operationType) || "APPEND".equals(operationType) || 
+           "RENAME_SRC".equals(operationType);
+  }
+
+  private Path generateUniquePath() {
+    int fileLimit = config.getTotalFiles();
+    int dirLimit = config.getDirSize();
+    int startPoint = 1 + rnd.nextInt(fileLimit);
+    String uniqueId = UUID.randomUUID().toString().replace("-", "").substring(0, 10);
+    return getPath(startPoint, dirLimit, Type.FILE, uniqueId);
+  }
+
+  private Path generateOriginalPath() {
     int fileLimit = config.getTotalFiles();
     int dirLimit = config.getDirSize();
     int startPoint = 1 + rnd.nextInt(fileLimit);
@@ -98,6 +251,30 @@ Path getFile() {
    * @return path
    */
   Path getDirectory() {
+    return getDirectory(null);
+  }
+
+  /**
+   * Gets a directory path based on operation type
+   * For CREATE/MKDIR operations: use original algorithm (write to base_dir)
+   * For LS operations: scan base_dir and select from existing directories
+   * 
+   * @param operationType the type of operation (can be null for backward compatibility)
+   * @return path
+   */
+  Path getDirectory(String operationType) { 
+    boolean useNewAlgorithm = config.shouldUseNewAlgorithm();   
+    // For LS operation, scan base_dir and select existing directories each time
+    if ("LS".equals(operationType)) {
+      if (useNewAlgorithm) {
+        LOG.info("Starting to scan base_dir and select existing directories for LS operation");
+        scanBaseDirectory();
+        return getExistingDirectory();
+      }
+      // Fall through to original algorithm for normal mode
+    }
+
+    // Use original algorithm by default
     int fileLimit = config.getTotalFiles();
     int dirLimit = config.getDirSize();
     int startPoint = rnd.nextInt(fileLimit);