我在一个步骤后查看了Spring批处理流/拆分,发现(a)标记的解决方案由于无法访问拆分而无法解析,(b)我的用例不同,因此答案的意图不同。
我还回顾了如何使用弹簧批处理并行步骤分流配置上述用例?,但解决方案是增加并行化,而不是拆分和连接线程。
这似乎是一个常见的用例,应该是一个FAQ,但我还没有看到解决方案。
我有一个并行的spring-batch作业,我想添加设置和拆卸步骤。设置和拆卸是单线程的,但主体工作是并行的。
随着时间的推移以不同的方式诊断为事件:
start setup
|
split -------
| |
some other
stuff stuff
| |
join -------
|
finish teardown
我开始的春季批处理作业XML是:
<batch:job id="myJob">
<batch:step id="step0001-setup">
<batch:tasklet transaction-manager="jtaTransactionManager"
start-limit="100" allow-start-if-complete="true">
<batch:chunk reader="beforeJobScriptExecutor" writer="dummySinkWriter"
commit-interval="1" />
</batch:tasklet>
</batch:step>
<batch:split id="split1" task-executor="taskExecutor" next="step9999">
<batch:flow>
<batch:step id="step0003-one-thread"
allow-start-if-complete="true">
<batch:tasklet transaction-manager="jtaTransactionManager"
start-limit="100">
<batch:chunk reader="myReader1" writer="myWriter1"
commit-interval="1" />
</batch:tasklet>
</batch:step>
</batch:flow>
<batch:flow>
<batch:step id="step0002-another-thread">
<batch:tasklet transaction-manager="jtaTransactionManager"
start-limit="100">
<batch:chunk reader="myReader2" writer="myWriter2"
commit-interval="1" />
</batch:tasklet>
</batch:step>
</batch:flow>
</batch:split>
<batch:step id="step9999">
<batch:tasklet transaction-manager="jtaTransactionManager"
start-limit="100" allow-start-if-complete="true">
<batch:chunk reader="afterJobScriptExecutor" writer="dummySinkWriter"
commit-interval="1" />
</batch:tasklet>
</batch:step>
</batch:job>
本文主要介绍了解决方案:如何使用Decider 在Spring批量拆分流中终止步骤
春季批处理作业的最终文本是:
<batch:job id="myJob">
<batch:step id="step0001-setup">
<batch:tasklet transaction-manager="jtaTransactionManager"
start-limit="100" allow-start-if-complete="true">
<batch:chunk reader="beforeJobScriptExecutor" writer="dummySinkWriter"
commit-interval="1" />
</batch:tasklet>
<batch:end on="FAILED" />
<batch:next on="*" to="split1" />
</batch:step>
<batch:split id="split1" task-executor="taskExecutor" next="step9999">
<batch:flow>
<batch:step id="step0003-one-thread"
allow-start-if-complete="true">
<batch:tasklet transaction-manager="jtaTransactionManager"
start-limit="100">
<batch:chunk reader="myReader1" writer="myWriter1"
commit-interval="1" />
</batch:tasklet>
</batch:step>
</batch:flow>
<batch:flow>
<batch:step id="step0002-another-thread">
<batch:tasklet transaction-manager="jtaTransactionManager"
start-limit="100">
<batch:chunk reader="myReader2" writer="myWriter2"
commit-interval="1" />
</batch:tasklet>
</batch:step>
</batch:flow>
</batch:split>
<batch:step id="step9999">
<batch:tasklet transaction-manager="jtaTransactionManager"
start-limit="100" allow-start-if-complete="true">
<batch:chunk reader="afterJobScriptExecutor" writer="dummySinkWriter"
commit-interval="1" />
</batch:tasklet>
</batch:step>
</batch:job>