apache/hadoop · error · IOException

Cannot truncate block to from oldlen (=${oldlen}) to newlen

Error message

Cannot truncate block to from oldlen (=${oldlen}) to newlen (=${newlen})

What it means

Error "Cannot truncate block to from oldlen (=${oldlen}) to newlen (=${newlen})" thrown in apache/hadoop.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/LocalReplica.java:488

    FsPermission permission = new FsPermission(oldPermission.getUserAction(),
        oldPermission.getGroupAction(), oldPermission.getOtherAction(), true);
    localFS.setPermission(path, permission);
  }

  public static void truncateBlock(
      FsVolumeSpi volume, File blockFile, File metaFile,
      long oldlen, long newlen, FileIoProvider fileIoProvider)
      throws IOException {
    LOG.info("truncateBlock: blockFile=" + blockFile
        + ", metaFile=" + metaFile
        + ", oldlen=" + oldlen
        + ", newlen=" + newlen);

    if (newlen == oldlen) {
      return;
    }
    if (newlen > oldlen) {
      throw new IOException("Cannot truncate block to from oldlen (=" + oldlen
          + ") to newlen (=" + newlen + ")");
    }

    // fis is closed by BlockMetadataHeader.readHeader.
    final FileInputStream fis = fileIoProvider.getFileInputStream(
        volume, metaFile);
    DataChecksum dcs = BlockMetadataHeader.readHeader(fis).getChecksum();
    int checksumsize = dcs.getChecksumSize();
    int bpc = dcs.getBytesPerChecksum();
    long n = (newlen - 1)/bpc + 1;
    long newmetalen = BlockMetadataHeader.getHeaderSize() + n*checksumsize;
    long lastchunkoffset = (n - 1)*bpc;
    int lastchunksize = (int)(newlen - lastchunkoffset);
    byte[] b = new byte[Math.max(lastchunksize, checksumsize)];

    try (RandomAccessFile blockRAF = fileIoProvider.getRandomAccessFile(
        volume, blockFile, "rw")) {
      //truncate blockFile

View on GitHub (pinned to 2add963021)

Solutions

1. Verify the truncate length is not larger than the current block length and the replica is in a state that allows truncation.
2. Retry the operation; if it repeats, check the block's recovery state and DataNode logs for conflicting writers.

When it happens

Trigger: A truncate request asks to change a block from oldlen to newlen in a way the replica does not support (e.g., newlen > oldlen or invalid state).

Common situations: A truncate was requested with a new length inconsistent with the current block length. Ensure the truncate length is smaller than the current length.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/fbfbce019e1fcaf1. Report an issue: GitHub.