[ https://issues.apache.org/jira/browse/SQOOP-1312?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14117409#comment-14117409
]
Jarek Jarcec Cecho commented on SQOOP-1312:
-------------------------------------------
What repository have you chosen [~Jongho]? You should use {{sqoop-trunk}} as you are working
on Sqoop 1.x.
> One of mappers does not load data from mySql if double column is used as split key
> ----------------------------------------------------------------------------------
>
> Key: SQOOP-1312
> URL: https://issues.apache.org/jira/browse/SQOOP-1312
> Project: Sqoop
> Issue Type: Bug
> Affects Versions: 1.4.4
> Reporter: Jong Ho Lee
> Attachments: splitter.patch, splitter.patch
>
>
> When we used Sqoop to load data from mySQL using one double column as split-key in Samsung
SDS,
> the last mapper did not load data from mySQL at all.
> The number of mappers was sometimes increased by 1.
> I think they were caused by some bugs in FloatSplitter.java
> For the last split, lowClausePrefix + Double.toString(curUpper), may be lowClausePrefix
+ Double.toString(curLower).
> In while (curUpper < maxVal) loop, because of round-off error,
> minVal + splitSize * numSplits can be smaller than maxVal.
> Therefore, using for-loop would be better.
> Attached is a proposed new FloatSplitter.java
> {code}
> /**
> * Licensed to the Apache Software Foundation (ASF) under one
> * or more contributor license agreements. See the NOTICE file
> * distributed with this work for additional information
> * regarding copyright ownership. The ASF licenses this file
> * to you under the Apache License, Version 2.0 (the
> * "License"); you may not use this file except in compliance
> * with the License. You may obtain a copy of the License at
> *
> * http://www.apache.org/licenses/LICENSE-2.0
> *
> * Unless required by applicable law or agreed to in writing, software
> * distributed under the License is distributed on an "AS IS" BASIS,
> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
> * See the License for the specific language governing permissions and
> * limitations under the License.
> */
> // modified by Jongho Lee at Samsung SDS.
> package org.apache.sqoop.mapreduce.db;
> import java.sql.ResultSet;
> import java.sql.SQLException;
> import java.util.ArrayList;
> import java.util.List;
> import org.apache.commons.logging.Log;
> import org.apache.commons.logging.LogFactory;
> import org.apache.hadoop.conf.Configuration;
> import org.apache.hadoop.mapreduce.InputSplit;
> import com.cloudera.sqoop.config.ConfigurationHelper;
> import com.cloudera.sqoop.mapreduce.db.DBSplitter;
> import com.cloudera.sqoop.mapreduce.db.DataDrivenDBInputFormat;
> /**
> * Implement DBSplitter over floating-point values.
> */
> public class FloatSplitter implements DBSplitter {
> private static final Log LOG = LogFactory.getLog(FloatSplitter.class);
> private static final double MIN_INCREMENT = 10000 * Double.MIN_VALUE;
> public List<InputSplit> split(Configuration conf, ResultSet results,
> String colName) throws SQLException {
> LOG.warn("Generating splits for a floating-point index column. Due to the");
> LOG.warn("imprecise representation of floating-point values in Java, this");
> LOG.warn("may result in an incomplete import.");
> LOG.warn("You are strongly encouraged to choose an integral split column.");
> List<InputSplit> splits = new ArrayList<InputSplit>();
> if (results.getString(1) == null && results.getString(2) == null) {
> // Range is null to null. Return a null split accordingly.
> splits.add(new DataDrivenDBInputFormat.DataDrivenDBInputSplit(
> colName + " IS NULL", colName + " IS NULL"));
> return splits;
> }
> double minVal = results.getDouble(1);
> double maxVal = results.getDouble(2);
> // Use this as a hint. May need an extra task if the size doesn't
> // divide cleanly.
> int numSplits = ConfigurationHelper.getConfNumMaps(conf);
> double splitSize = (maxVal - minVal) / (double) numSplits;
> if (splitSize < MIN_INCREMENT) {
> splitSize = MIN_INCREMENT;
> }
> String lowClausePrefix = colName + " >= ";
> String highClausePrefix = colName + " < ";
> double curLower = minVal;
> double curUpper = curLower + splitSize;
> for (int i = 0; i < numSplits - 1; i++) {
> // while (curUpper < maxVal) { // changed to for loop
> splits.add(new DataDrivenDBInputFormat.DataDrivenDBInputSplit(
> lowClausePrefix + Double.toString(curLower),
> highClausePrefix + Double.toString(curUpper)));
> curLower = curUpper;
> curUpper += splitSize;
> }
> // Catch any overage and create the closed interval for the last split.
> if (curLower <= maxVal || splits.size() == 1) {
> splits.add(new DataDrivenDBInputFormat.DataDrivenDBInputSplit(
> lowClausePrefix + Double.toString(curLower),
> colName + " <= " + Double.toString(maxVal)));
> // curUpper -> curLower // changed
> }
> if (results.getString(1) == null || results.getString(2) == null) {
> // At least one extrema is null; add a null split.
> splits.add(new DataDrivenDBInputFormat.DataDrivenDBInputSplit(
> colName + " IS NULL", colName + " IS NULL"));
> }
> return splits;
> }
> }
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
|