Package org.apache.hadoop.hbase.io.hfile

Class TestSeekBeforeWithInlineBlocks

java.lang.Object
org.apache.hadoop.hbase.io.hfile.TestSeekBeforeWithInlineBlocks
@Tag("org.apache.hadoop.hbase.testclassification.IOTests") @Tag("org.apache.hadoop.hbase.testclassification.MediumTests") public class TestSeekBeforeWithInlineBlocks extends Object

  • Field Summary

    Fields
    Modifier and Type
    Field
    Description
    private static final int
    BLOOM_BLOCK_SIZE
     
    private org.apache.hadoop.conf.Configuration
    conf
     
    private static final int
    DATA_BLOCK_SIZE
     
    private static final int[]
    EXPECTED_NUM_LEVELS
     
    private static final byte[]
    FAM
     
    private org.apache.hadoop.fs.FileSystem
    fs
     
    private static final int[]
    INDEX_CHUNK_SIZES
     
    private static final org.slf4j.Logger
    LOG
     
    private static final int
    NUM_KV
     
    private static final Random
    RAND
     
    private static final HBaseTestingUtil
    TEST_UTIL
     

  • Constructor Summary

    Constructors
    Constructor
    Description
    TestSeekBeforeWithInlineBlocks()
     

  • Method Summary

    Modifier and Type
    Method
    Description
    private void
    checkCell(org.apache.hadoop.hbase.ExtendedCell expected, org.apache.hadoop.hbase.ExtendedCell actual)
    Check a key/value pair after it was read by the reader
    private void
    checkNoSeekBefore(org.apache.hadoop.hbase.ExtendedCell[] cells, org.apache.hadoop.hbase.io.hfile.HFileScanner scanner, int i)
     
    private void
    checkSeekBefore(org.apache.hadoop.hbase.ExtendedCell[] cells, org.apache.hadoop.hbase.io.hfile.HFileScanner scanner, int i)
     
    void
    testMultiIndexLevelRandomHFileWithBlooms()
    Scanner.seekBefore() could fail because when seeking to a previous HFile data block, it needs to know the size of that data block, which it calculates using current data block offset and the previous data block offset.

    Methods inherited from class java.lang.Object

    clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

  • Field Details

  • Constructor Details

  • Method Details

    • testMultiIndexLevelRandomHFileWithBlooms

      @Test public void testMultiIndexLevelRandomHFileWithBlooms() throws IOException
      Scanner.seekBefore() could fail because when seeking to a previous HFile data block, it needs to know the size of that data block, which it calculates using current data block offset and the previous data block offset. This fails to work when there are leaf-level index blocks in the scannable section of the HFile, i.e. starting in HFileV2. This test will try seekBefore() on a flat (single-level) and multi-level (2,3) HFile and confirm this bug is now fixed. This bug also happens for inline Bloom blocks for the same reasons.
      Throws:
      IOException

    • checkSeekBefore

      private void checkSeekBefore(org.apache.hadoop.hbase.ExtendedCell[] cells, org.apache.hadoop.hbase.io.hfile.HFileScanner scanner, int i) throws IOException
      Throws:
      IOException

    • checkNoSeekBefore

      private void checkNoSeekBefore(org.apache.hadoop.hbase.ExtendedCell[] cells, org.apache.hadoop.hbase.io.hfile.HFileScanner scanner, int i) throws IOException
      Throws:
      IOException

    • checkCell

      private void checkCell(org.apache.hadoop.hbase.ExtendedCell expected, org.apache.hadoop.hbase.ExtendedCell actual)
      Check a key/value pair after it was read by the reader