sqoop-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Boglarka Egyed (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SQOOP-1312) One of mappers does not load data from mySql if double column is used as split key
Date Fri, 05 Oct 2018 10:33:00 GMT

    [ https://issues.apache.org/jira/browse/SQOOP-1312?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16639637#comment-16639637
] 

Boglarka Egyed commented on SQOOP-1312:
---------------------------------------

I have updated the Affects Version to include 1.4.6 and 1.4.7 too. Patch provided in [https://reviews.apache.org/r/25621/] seems
to be abandoned, last update was 4 years ago, I think the fix is up for grabs now.

> 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, 1.4.6, 1.4.7
>            Reporter: Jong Ho Lee
>            Assignee: Jong Ho Lee
>            Priority: Major
>         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
(v7.6.3#76005)

Mime
View raw message