Пример #1
0
  /**
   * Returns statistics on this table as a tabular result set. Used for the SHOW TABLE STATS
   * statement. The schema of the returned TResultSet is set inside this method.
   */
  public TResultSet getTableStats() {
    TResultSet result = new TResultSet();
    TResultSetMetadata resultSchema = new TResultSetMetadata();
    result.setSchema(resultSchema);
    resultSchema.addToColumns(new TColumn("Region Location", ColumnType.STRING.toThrift()));
    resultSchema.addToColumns(new TColumn("Start RowKey", ColumnType.STRING.toThrift()));
    resultSchema.addToColumns(new TColumn("Est. #Rows", ColumnType.BIGINT.toThrift()));
    resultSchema.addToColumns(new TColumn("Size", ColumnType.STRING.toThrift()));

    // TODO: Consider fancier stats maintenance techniques for speeding up this process.
    // Currently, we list all regions and perform a mini-scan of each of them to
    // estimate the number of rows, the data size, etc., which is rather expensive.
    try {
      long totalNumRows = 0;
      long totalHdfsSize = 0;
      List<HRegionLocation> regions =
          HBaseTable.getRegionsInRange(
              hTable_, HConstants.EMPTY_END_ROW, HConstants.EMPTY_START_ROW);
      for (HRegionLocation region : regions) {
        TResultRowBuilder rowBuilder = new TResultRowBuilder();
        HRegionInfo regionInfo = region.getRegionInfo();
        Pair<Long, Long> estRowStats =
            getEstimatedRowStats(regionInfo.getStartKey(), regionInfo.getEndKey());

        long numRows = estRowStats.first.longValue();
        long hdfsSize = getHdfsSize(regionInfo);
        totalNumRows += numRows;
        totalHdfsSize += hdfsSize;

        // Add the region location, start rowkey, number of rows and raw Hdfs size.
        rowBuilder
            .add(String.valueOf(region.getHostname()))
            .add(Bytes.toString(regionInfo.getStartKey()))
            .add(numRows)
            .addBytes(hdfsSize);
        result.addToRows(rowBuilder.get());
      }

      // Total num rows and raw Hdfs size.
      if (regions.size() > 1) {
        TResultRowBuilder rowBuilder = new TResultRowBuilder();
        rowBuilder.add("Total").add("").add(totalNumRows).addBytes(totalHdfsSize);
        result.addToRows(rowBuilder.get());
      }
    } catch (IOException e) {
      throw new RuntimeException(e);
    }
    return result;
  }
Пример #2
0
 private void checkHBaseTableCols(
     Db db,
     String hiveTableName,
     String hbaseTableName,
     String[] hiveColNames,
     String[] colFamilies,
     String[] colQualifiers,
     ColumnType[] colTypes)
     throws TableLoadingException {
   checkTableCols(db, hiveTableName, 1, hiveColNames, colTypes);
   HBaseTable tbl = (HBaseTable) db.getTable(hiveTableName);
   assertEquals(tbl.getHBaseTableName(), hbaseTableName);
   List<Column> cols = tbl.getColumns();
   assertEquals(colFamilies.length, colTypes.length);
   assertEquals(colQualifiers.length, colTypes.length);
   Iterator<Column> it = cols.iterator();
   int i = 0;
   while (it.hasNext()) {
     HBaseColumn col = (HBaseColumn) it.next();
     assertEquals(col.getColumnFamily(), colFamilies[i]);
     assertEquals(col.getColumnQualifier(), colQualifiers[i]);
     ++i;
   }
 }