多步骤并行的Spring batch Partitioning？

Question

我已经为单个步骤实现了spring批处理分区，其中主步骤将其工作委托给几个从线程，然后这些从线程被并行执行。如下图所示。(参考Spring docs)

现在，如果我有多个并行执行的步骤怎么办？如何批量配置？我当前的配置是

 <batch:job id="myJob" restartable="true" job-repository="jobRepository" >
        <batch:listeners>
            <batch:listener ref="myJoblistener"></batch:listener>
        </batch:listeners>

        <batch:step id="my-master-step">
            <batch:partition step="my-step" partitioner="my-step-partitioner" handler="my-partitioner-handler">
            </batch:partition>
        </batch:step>
    </batch:job>

    <batch:step id="my-step" >
        <batch:tasklet ref="myTasklet" transaction-manager="transactionManager" >
        </batch:tasklet>
        <batch:listeners>
            <batch:listener ref="myStepListener"></batch:listener>
        </batch:listeners> 
    </batch:step>

我的架构图应该如下图所示：

​

我不确定是否可以使用spring batch.Any的想法，或者我无法实现it.Thank you。

Answer 1

您可以尝试执行以下操作。

 <batch:job id="myJob" restartable="true" job-repository="jobRepository" >
        <batch:listeners>
            <batch:listener ref="myJoblistener"></batch:listener>
        </batch:listeners>

        <batch:step id="my-master-step">
            <batch:partition step="my-step" partitioner="my-step-partitioner" handler="my-partitioner-handler">
            </batch:partition>
        </batch:step>
    </batch:job>

    <batch:step id="my-step" >
        <batch:job ref="MyChildJob" job-launcher="jobLauncher"
                job-parameters-extractor="jobParametersExtractor" />
        <batch:listeners>
            <batch:listener ref="myStepListener"></batch:listener>
        </batch:listeners> 
    </batch:step>

    <batch:job id="MyChildJob" restartable="false"
        xmlns="http://www.springframework.org/schema/batch">
        <batch:step id="MyChildStep1" next="MyChildStep2">
            <batch:tasklet ref="MyChildStep1Tasklet" transaction-manager="transactionManager" >
            </batch:tasklet>
        </batch:step>

        <batch:step id="MyChildStep2" next="MyChildStep3">
            <batch:tasklet ref="MyChildStep2Tasklet" transaction-manager="transactionManager" >
            </batch:tasklet>
        </batch:step>

        <batch:step id="MyChildStep3">
            <batch:tasklet ref="MyChildStep3Tasklet" transaction-manager="transactionManager" >
            </batch:tasklet>
        </batch:step>

    </batch:job>

Answer 2

我有类似的需求，并使用下面的需求解决了它

<batch:job id="cycleJob">
        <batch:step id="zStep" next="gStep">
            <batch:partition partitioner="zPartitioner">
                <batch:step>
                    <batch:tasklet throttle-limit="1">
                        <batch:chunk processor="itemProcessor" reader="zReader" writer="itemWriter" commit-interval="1">
                        </batch:chunk>
                    </batch:tasklet>
                </batch:step>
                <batch:handler task-executor="taskExecutor" grid-size="${maxThreads}" />
            </batch:partition>
        </batch:step>
        <batch:step id="gStep" parent="zStep" next="yStep">
            <batch:partition partitioner="gPartitioner">
                <batch:step>
                    <batch:tasklet throttle-limit="1">
                        <batch:chunk processor="itemProcessor" reader="gReader" writer="itemWriter" commit-interval="1">
                        </batch:chunk>
                    </batch:tasklet>
                </batch:step>
                <batch:handler task-executor="taskExecutor" grid-size="${maxThreads}" />
            </batch:partition>
        </batch:step>
</batch:job>

Answer 3

回答晚了，但我终于找到了我最初在这里寻找的解决方案，使用flow而不是子作业。所以我想我也应该把它贴在这里。

    <job id="myJob">
        <step id="my-master-step">
            <partition partitioner="my-step-partitioner">
                <handler task-executor="my-partitioner-handler" />
                <step>
                    <!-- For each partition, we run the complete flow -->
                    <flow parent="mainFlow" />
                </step>
            </partition>
        </step>
    </job>
    
    <!-- The flow consists of several sequential steps (2 here) -->
    <flow id="mainFlow">
        <step id="MyChildStep1" next="MyChildStep2">
            <!-- Here you can have a tasklet or a chunk of course -->
            <tasklet ref="MyChildStep1Tasklet" />
        </step>
        <step id="MyChildStep2">
            <!-- Same here -->
            <tasklet ref="MyChildStep2Tasklet" />
        </step>
    </flow>
    
    <bean id="MyChildStep1Tasklet" class="..." />
    
    <bean id="MyChildStep1Tasklet" class="..." />

我没有测试过并行运行它，但我看不出它为什么不能工作。