apache/hadoop · error · IOException

Client requested checksum ${requestedChecksum} when appendin

Error message

Client requested checksum ${requestedChecksum} when appending to an existing block with different chunk size: ${checksum}

What it means

Error "Client requested checksum ${requestedChecksum} when appending to an existing block with different chunk size: ${checksum}" thrown in apache/hadoop.

Source

Thrown at hadoop-hdfs-project/hadoop-hdfs/src/main/java/org/apache/hadoop/hdfs/server/datanode/LocalReplicaInPipeline.java:321

    // the checksum that should actually be used -- this
    // may differ from requestedChecksum for appends.
    final DataChecksum checksum;

    final RandomAccessFile metaRAF =
        getFileIoProvider().getRandomAccessFile(getVolume(), metaFile, "rw");

    if (!isCreate) {
      // For append or recovery, we must enforce the existing checksum.
      // Also, verify that the file has correct lengths, etc.
      boolean checkedMeta = false;
      try {
        BlockMetadataHeader header =
            BlockMetadataHeader.readHeader(metaRAF);
        checksum = header.getChecksum();

        if (checksum.getBytesPerChecksum() !=
            requestedChecksum.getBytesPerChecksum()) {
          throw new IOException("Client requested checksum " +
              requestedChecksum + " when appending to an existing block " +
              "with different chunk size: " + checksum);
        }

        int bytesPerChunk = checksum.getBytesPerChecksum();
        int checksumSize = checksum.getChecksumSize();

        blockDiskSize = bytesOnDisk;
        crcDiskSize = BlockMetadataHeader.getHeaderSize() +
          (blockDiskSize+bytesPerChunk-1)/bytesPerChunk*checksumSize;
        if (blockDiskSize > 0 &&
            (blockDiskSize > blockFile.length() ||
               crcDiskSize>metaFile.length())) {
          throw new IOException("Corrupted block: " + this);
        }
        checkedMeta = true;
      } finally {
        if (!checkedMeta) {

View on GitHub (pinned to 2add963021)

Solutions

1. Configure the client with the same bytes-per-checksum as the existing block (or recreate the file with the desired checksum chunk size).
2. Ensure dfs.bytes-per-checksum / dfs.checksum.type are consistent across clients writing to the same file.

When it happens

Trigger: A client appends to an existing block requesting a checksum configuration whose chunk size differs from the block's recorded checksum.

Common situations: An append request specified a chunk/checksum size different from the existing block's. Recreate the file with the intended checksum configuration instead of appending.

AI-assisted analysis of apache/hadoop@2add963021 (2026-08-22). Data as JSON: /api/errors/c173f87b2e62847c. Report an issue: GitHub.