我的源表中有10条记录,项目数为3.我有2个分区来处理这10条记录(即前5条记录将在第一个分区中处理,剩余记录将在第二个分区中处理,而处理第二个分区中的记录我正在抛出异常,因此作业将在第二个分区的第二个块中失败.当...
我的源表中有10条记录,项目数为3.
我有2个分区来处理这10条记录(即前5条记录将在第一个分区中处理,剩余记录将在第二个分区中处理,而处理第二个分区中的记录我正在抛出异常,因此作业将在第二个分区的第二个块中失败.当我重新启动作业时,失败的分区再次处理所有记录(即第一个块和第二个块).
重新启动作业应仅处理上次失败的块记录,但不能处理该分区中的所有记录.您能否指导我如何实现此目的?
我的JSL如下:
<?xml version="1.0" encoding="UTF-8"?>
<job xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns="http://xmlns.jcp.org/xml/ns/javaee"
xsi:schemaLocation="http://xmlns.jcp.org/xml/ns/javaee http://xmlns.jcp.org/xml/ns/javaee/jobXML_1_0.xsd"
id="readingfrom-db" restartable="true" version="1.0" >
<properties >
<property name="numRec" value="#{jobParameters['numRec']}?:5;"/>
<property name="chunkSize" value="#{jobParameters['chunkSize']}?:3;"/>
<property name="whereclauseFrom" value="#{jobParameters['whereclauseFrom']}?:5;"/>
<property name="whereclauseTo" value="#{jobParameters['whereclauseTo']}?:6;"/>
<property name="dsJNDI" value="#{jobParameters['dsJNDI']}?:jdbc/db2;"/>
<property name="dsJNDI1" value="#{jobParameters['dsJNDI1']}?:jdbc/db2;"/>
<property name="tableName" value="#{jobParameters['tableName']}?:CISDW.AIF1_CH;"/>
<property name="ProcesstableName" value="#{jobParameters['ProcesstableName']}?:CISDW.PROC_AIF1_CH;"/>
</properties>
<step id="runcache" next="readFromDB">
<batchlet ref="com.cdc.runcache.CacheRunnerBatchlet" />
</step>
<step id="readFromDB">
<listeners>
<listener ref="com.cdc.dbreader.LogExceptionListener"/>
</listeners>
<chunk item-count="3" checkpoint-policy="item">
<reader ref="com.cdc.dbreader.DBItemReader">
<properties >
<property name="dsJNDI" value="#{jobProperties['dsJNDI']}"/>
<property name="tableName" value="#{jobProperties['tableName']}"/>
<property name="whereclauseFrom" value="#{partitionPlan['modrec']}"/>
</properties>
</reader>
<processor ref="com.cdc.dbreader.DBItemProcessor" />
<writer ref="com.cdc.dbreader.DBItemWriter">
<properties >
<property name="dsJNDI" value="#{jobProperties['dsJNDI1']}"/>
<property name="tableName" value="#{jobProperties['ProcesstableName']}"/>
</properties>
</writer>
</chunk>
<partition>
<plan partitions="2" threads="2">
<properties partition="0">
<property name="modrec" value="#{jobProperties['whereclauseFrom']}"/>
</properties>
<properties partition="1">
<property name="modrec" value="#{jobProperties['whereclauseTo']}"/>
</properties>
</plan>
</partition>
</step>
</job>
我的项目阅读器如下:
public class DBItemReader implements ItemReader {
@Inject
@BatchProperty
private String dsJNDI;
@Inject
@BatchProperty
private String whereclauseFrom;
@Inject
@BatchProperty
private String tableName;
private Connection conn =null;
private int totalRecords=0;
private DataSource ds = null;
List<RecObj> listRecObj=new ArrayList<RecObj>();
@Override
public Object readItem() throws SQLException {
if (listRecObj.size() == 0) {
return null;
} else {
RecObj rec =null;
Iterator<RecObj> iter =listRecObj.iterator();
while (iter.hasNext()) {
rec = iter.next();
if (Integer.parseInt(rec.getRec()) == 7) {
throw new IllegalStateException("Thrown Error");
}
iter.remove();
return rec;
}
return rec;
}
@Override
public void open(Serializable arg0) throws NamingException, SQLException {
ds = DataSource.class.cast(new InitialContext().lookup(dsJNDI));
// System.out.println("whereclauseFrom: " + whereclauseFrom);
conn = ds.getConnection();
String sql ="";
if(Integer.parseInt(whereclauseFrom) == 5){
sql = "SELECT * FROM " + tableName + " WHERE CAST(REC AS INTEGER) <= "+ whereclauseFrom;
}else if(Integer.parseInt(whereclauseFrom) == 6){
sql = "SELECT * FROM " + tableName + " WHERE CAST(REC AS INTEGER) >= "+ whereclauseFrom;
}
PreparedStatement ps = conn.prepareStatement(sql);
ResultSet rs=ps.executeQuery();
while(rs.next()){
totalRecords++;
String rec=rs.getString("REC");
if(rec != null)
listRecObj.add(new RecObj(rec));
}
rs.close();
}
@Override
public void close() throws SQLException {
conn.close();
}
@Override
public Serializable checkpointInfo() {
return null;
}
}
}
我的作家课如下:
public class DBItemWriter extends AbstractItemWriter implements ItemWriter {
@Inject
@BatchProperty
private String dsJNDI;
@Inject
@BatchProperty
private String tableName;
private DataSource ds = null;
@Override
public void open(Serializable arg0) throws NamingException {
ds = DataSource.class.cast(new InitialContext().lookup(dsJNDI));
}
@Override
public void writeItems(List<java.lang.Object> items) throws BatchUpdateException,SQLException{
Connection conn = ds.getConnection();
String sql = "INSERT INTO "+tableName+ "(MOD_REC) VALUES(?) ";
PreparedStatement ps = conn.prepareStatement(sql);
for (Object obj : items) {
RecObj v = (RecObj)obj;
System.out.println("=======Writer values===="+v.getRec());
ps.setString(1, v.getRec());
ps.addBatch();
}
ps.executeBatch();
ps.clearBatch();
ps.close();
conn.close();
}
}
以下是我的处理器:
public class DBItemProcessor implements ItemProcessor {
Integer count=0;
@Override
public Object processItem(Object arg0) {
count++;
RecObj v=(RecObj)arg0;
String vname=v.getRec();
System.out.println("=========Processer Values==="+vname);
return new RecObj(vname+count);
}
}
下面是我的Bean类
public class RecObj {
private String rec;
public RecObj(String rec) {
this.rec=rec;
}
解决方法:
您需要在读者的checkpointInfo()中返回一个检查点值,该值将在重启时传递给读者的open()方法.这是读者和批处理容器协调以在重启时提供检查点的方式.
所以你可以有类似的东西(查找CHECKPOINT注释):
public class DBItemReader implements ItemReader {
// ...
// CHECKPOINT field defined
private String checkpoint = null;
@Override
public void open(Serializable checkpoint) throws NamingException, SQLException {
// CHECKPOINT-based positioning through query value.
// Initial position = whereclauseFrom, on restart set to checkpoint
String queryVal = (String)(checkpoint == null ? whereclauseFrom : checkpoint);
if(Integer.parseInt(whereclauseFrom) == 5){
sql = "SELECT * FROM " + tableName + " WHERE CAST(REC AS INTEGER) <= "+ queryVal;
}else if(Integer.parseInt(whereclauseFrom) == 6){
sql = "SELECT * FROM " + tableName + " WHERE CAST(REC AS INTEGER) >= "+ queryVal;
}
// ..
}
@Override
public Object readItem() throws SQLException {
if (listRecObj.size() == 0) {
return null;
} else {
RecObj rec =null;
Iterator<RecObj> iter =listRecObj.iterator();
while (iter.hasNext()) {
rec = iter.next();
// CHECKPOINT updated
checkpoint = rec.getRec();
if (Integer.parseInt(rec.getRec()) == 7) {
throw new IllegalStateException("Thrown Error");
}
}
}
// ...
}
@Override
public Serializable checkpointInfo() {
// CHECKPOINT returned at end of chunk
return checkpoint;
}
}
本文标题为:java – 带有Liberty Profile的JSR 352 – 如何在ItemReader执行数据库查询时实现检查点
基础教程推荐
- Java 策略模式 if-else用法实例详解 2023-02-19
- Java zookeeper服务的使用详解 2023-03-22
- 详解Spring的@Value作用与使用场景 2023-07-14
- 全面理解Java内存模型 2023-09-01
- Java迭代器与Collection接口超详细讲解 2023-03-11
- SpringBoot详解如何进行整合Druid数据源 2023-01-29
- java获取本月日历表的方法 2023-01-02
- MQ的消息模型及在工作上应用场景 2022-11-20
- java实现简单学生成绩档案管理系统 2022-11-25
- Spring零基础到进阶之鸿蒙开篇 2023-02-27