使Spring Batch ItemReader可重启

Question

我正在开发的spring批处理程序正在从表中读取数据。它使用的是“组织”。springframework。一批项目数据库JdbcCursorItemReader的itemReader。早些时候，计划是修改表，添加一个已处理的_指示符标志，并将其预填充为状态“待定”。一旦记录被处理，writer将把已处理的_指示器标志的状态更新为“已处理”。这是为了支持重新启动。例如，如果批次拾取了100万条记录，并在½万条记录中死亡，那么当我重新启动批次时；应该从我结束的地方开始。但不幸的是，管理层没有批准这个解决方案。我正在想方设法让itemreader重新启动。根据Spring文档，“大多数ItemReader都有更复杂的重启逻辑。例如，JdbcCursorItemReader将最后处理的行的行id存储在游标中。”

有人有这样的自定义阅读器的示例，它实现了JdbcCursorItemReader并将最后处理的行存储在光标中。https://docs.spring.io/spring-batch/trunk/reference/html/readersAndWriters.html

==完整的XML配置==

<import resource="classpath:/batch/utility/skip/batch_skip.xml" />
<import resource="classpath:/batch/config/context-postgres.xml" />
<import resource="classpath:/batch/config/oracle-database.xml" />

<context:property-placeholder
    location="classpath:/batch/jobs/TPF-1001-DD-01/TPF-1001-DD-01.properties" />
<bean id="gridSizePartitioner"
    class="com.tpf.partitioner.GridSizePartitioner" />

      <task:executor id="taskExecutor" pool-size="${pool.size}" />
<batch:job id="XYZJob" job-repository="jobRepository"
    restartable="true">

    <batch:step id="XYZSTEP">
        <batch:description>Convert TIF files to PDF</batch:description>
        <batch:partition partitioner="gridSizePartitioner">

            <batch:handler task-executor="taskExecutor"
                grid-size="${pool.size}" />
            <batch:step>
                <batch:tasklet allow-start-if-complete="true">
                    <batch:chunk commit-interval="${commit.interval}"
                        skip-limit="${job.skip.limit}">

                        <batch:reader>
                            <bean id="timeReader"
                                class="org.springframework.batch.item.database.JdbcCursorItemReader"
                                scope="step">
                                <property name="dataSource" ref="oracledataSource" />
                                <property name="sql">
                                    <value>                                     
                                    select TIME_ID as timesheetId,count(*),max(CREATION_DATETIME) as creationDateTime , ILN_NUMBER as ilnNumber
                                    from TS_FAKE_NAME
                                    where creation_datetime  >= '#{jobParameters['creation_start_date1']} 12.00.00.000000000 AM' 
                                    and creation_datetime &lt;  '#{jobParameters['creation_start_date2']} 11.59.59.999999999 PM' 
                                    and mod(time_id,${pool.size})=#{stepExecutionContext['partition.id']} 
                                    group by  time_id ,ILN_NUMBER                                   

                                    </value>
                                </property>
                                <property name="rowMapper">
                                    <bean
                                        class="org.springframework.jdbc.core.BeanPropertyRowMapper">
                                        <property name="mappedClass"
                                            value="com.tpf.model.Time" />
                                    </bean>
                                </property>
                            </bean>
                        </batch:reader>
                        <batch:processor>
                            <bean id="compositeItemProcessor"
                                class="org.springframework.batch.item.support.CompositeItemProcessor">
                                <property name="delegates">
                                    <list>
                                        <ref bean="timeProcessor" />
                                    </list>
                                </property>

                            </bean>
                        </batch:processor>


                        <batch:writer>
                            <bean id="compositeItemWriter"
                                class="org.springframework.batch.item.support.CompositeItemWriter">
                                <property name="delegates">
                                    <list>
                                        <ref bean="timeWriter" />
                                    </list>
                                </property>
                            </bean>
                        </batch:writer>
                        <batch:skippable-exception-classes>
                            <batch:include
                                class="com.utility.skip.BatchSkipException" />
                        </batch:skippable-exception-classes>
                        <batch:listeners>
                            <batch:listener ref="batchSkipListener" />
                        </batch:listeners>
                    </batch:chunk>
                </batch:tasklet>
            </batch:step>
        </batch:partition>
    </batch:step>
    <batch:validator>
        <bean
            class="org.springframework.batch.core.job.DefaultJobParametersValidator">
            <property name="requiredKeys">
                <list>
                    <value>batchRunNumber</value>
                    <value>creation_start_date1</value>
                    <value>creation_start_date2</value>
                </list>
            </property>
        </bean>
    </batch:validator>
</batch:job>



<bean id="timesheetWriter" class="com.tpf.writer.TimeWriter"
    scope="step">
    <property name="dataSource" ref="dataSource" />
</bean>

<bean id="timeProcessor"
    class="com.tpf.processor.TimeProcessor" scope="step">
    <property name="dataSource" ref="oracledataSource" />
</bean> 

Answer 1

有没有人有这样的定制阅读器的示例，它实现了JdbcCursorItemReader，并在游标中存储最后处理的行

JdbcCursorItemReader可以做到这一点，参见Javadoc，这里是一个摘录：

ExecutionContext: The current row is returned as restart data,
and when restored from that same data, the cursor is opened and the current row
set to the value within the restart data.

所以你不需要定制阅读器。