Files
spring-batch/build/reference-work/step.xml
Michael Minella 75ab909314 update
2017-03-23 10:18:33 -05:00

1742 lines
77 KiB
XML

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE chapter PUBLIC "-//OASIS//DTD DocBook XML V4.4//EN"
"http://www.oasis-open.org/docbook/xml/4.4/docbookx.dtd">
<chapter id="configureStep">
<title>Configuring a Step</title>
<para>As discussed in <xref linkend="domain" />, a
<classname>Step</classname> is a domain object that encapsulates an
independent, sequential phase of a batch job and contains all of the
information necessary to define and control the actual batch processing.
This is a necessarily vague description because the contents of any given
<classname>Step</classname> are at the discretion of the developer writing a
<classname>Job</classname>. A Step can be as simple or complex as the
developer desires. A simple <classname>Step</classname> might load data from
a file into the database, requiring little or no code. (depending upon the
implementations used) A more complex <classname>Step</classname> may have
complicated business rules that are applied as part of the
processing.</para>
<mediaobject>
<imageobject role="html">
<imagedata align="center" fileref="images/step.png" scale="60" />
</imageobject>
<imageobject role="fo">
<imagedata align="center" fileref="images/step.png" scale="50" />
</imageobject>
</mediaobject>
<section id="chunkOrientedProcessing">
<title>Chunk-Oriented Processing</title>
<para>Spring Batch uses a 'Chunk Oriented' processing style within its
most common implementation. Chunk oriented processing refers to reading
the data one at a time, and creating 'chunks' that will be written out,
within a transaction boundary. One item is read in from an
<classname>ItemReader</classname>, handed to an
<classname>ItemProcessor</classname>, and aggregated. Once the number of
items read equals the commit interval, the entire chunk is written out via
the ItemWriter, and then the transaction is committed.</para>
<mediaobject>
<imageobject role="html">
<imagedata align="center"
fileref="images/chunk-oriented-processing.png" scale="100" />
</imageobject>
<imageobject role="fo">
<imagedata align="center"
fileref="images/chunk-oriented-processing.png" scale="55" />
</imageobject>
</mediaobject>
<para>Below is a code representation of the same concepts shown
above:</para>
<programlisting language="java">List items = new Arraylist();
for(int i = 0; i &lt; commitInterval; i++){
Object item = itemReader.read()
Object processedItem = itemProcessor.process(item);
items.add(processedItem);
}
itemWriter.write(items);</programlisting>
<section id="configuringAStep">
<title>Configuring a Step</title>
<para>Despite the relatively short list of required dependencies for a
<classname>Step</classname>, it is an extremely complex class that can
potentially contain many collaborators. In order to ease configuration,
the Spring Batch namespace can be used:</para>
<programlisting language="xml">&lt;job id="sampleJob" job-repository="jobRepository"&gt;
&lt;step id="step1"&gt;
&lt;tasklet transaction-manager="transactionManager"&gt;
&lt;chunk reader="itemReader" writer="itemWriter" commit-interval="10"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;
&lt;/job&gt;</programlisting>
<para>The configuration above represents the only required dependencies
to create a item-oriented step:<itemizedlist>
<listitem>
<para>reader - The <classname>ItemReader</classname> that provides
items for processing.</para>
</listitem>
<listitem>
<para>writer - The <classname>ItemWriter</classname> that
processes the items provided by the
<classname>ItemReader</classname>.</para>
</listitem>
<listitem>
<para>transaction-manager - Spring's
<classname>PlatformTransactionManager</classname> that will be
used to begin and commit transactions during processing.</para>
</listitem>
<listitem>
<para>job-repository - The <classname>JobRepository</classname>
that will be used to periodically store the
<classname>StepExecution</classname> and
<classname>ExecutionContext</classname> during processing (just
before committing). For an in-line &lt;step/&gt; (one defined
within a &lt;job/&gt;) it is an attribute on the &lt;job/&gt;
element; for a standalone step, it is defined as an attribute of
the &lt;tasklet/&gt;.</para>
</listitem>
<listitem>
<para>commit-interval - The number of items that will be processed
before the transaction is committed.</para>
</listitem>
</itemizedlist></para>
<para>It should be noted that, job-repository defaults to
"jobRepository" and transaction-manager defaults to "transactionManger".
Furthermore, the <classname>ItemProcessor</classname> is optional, not
required, since the item could be directly passed from the reader to the
writer.</para>
</section>
<section id="InheritingFromParentStep">
<title>Inheriting from a Parent Step</title>
<para>If a group of <classname>Step</classname>s share similar
configurations, then it may be helpful to define a "parent"
<classname>Step</classname> from which the concrete
<classname>Step</classname>s may inherit properties. Similar to class
inheritance in Java, the "child" <classname>Step</classname> will
combine its elements and attributes with the parent's. The child will
also override any of the parent's <classname>Step</classname>s.</para>
<para>In the following example, the <classname>Step</classname>
"concreteStep1" will inherit from "parentStep". It will be instantiated
with 'itemReader', 'itemProcessor', 'itemWriter', startLimit=5, and
allowStartIfComplete=true. Additionally, the commitInterval will be '5'
since it is overridden by the "concreteStep1":</para>
<programlisting language="xml">&lt;step id="parentStep"&gt;
&lt;tasklet allow-start-if-complete="true"&gt;
&lt;chunk reader="itemReader" writer="itemWriter" commit-interval="10"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;
&lt;step id="concreteStep1" parent="parentStep"&gt;
&lt;tasklet start-limit="5"&gt;
&lt;chunk processor="itemProcessor" commit-interval="5"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
<para>The id attribute is still required on the step within the job
element. This is for two reasons:<orderedlist>
<listitem>
<para>The id will be used as the step name when persisting the
StepExecution. If the same standalone step is referenced in more
than one step in the job, an error will occur.</para>
</listitem>
<listitem>
<para>When creating job flows, as described later in this chapter,
the next attribute should be referring to the step in the flow,
not the standalone step.</para>
</listitem>
</orderedlist></para>
<section id="abstractStep">
<title>Abstract Step</title>
<para>Sometimes it may be necessary to define a parent
<classname>Step</classname> that is not a complete
<classname>Step</classname> configuration. If, for instance, the
reader, writer, and tasklet attributes are left off of a
<classname>Step </classname>configuration, then initialization will
fail. If a parent must be defined without these properties, then the
"abstract" attribute should be used. An "abstract"
<classname>Step</classname> will not be instantiated; it is used only
for extending.</para>
<para>In the following example, the <classname>Step</classname>
"abstractParentStep" would not instantiate if it were not declared to
be abstract. The <classname>Step</classname> "concreteStep2" will have
'itemReader', 'itemWriter', and commitInterval=10.</para>
<programlisting language="xml">&lt;step id="abstractParentStep" abstract="true"&gt;
&lt;tasklet&gt;
&lt;chunk commit-interval="10"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;
&lt;step id="concreteStep2" parent="abstractParentStep"&gt;
&lt;tasklet&gt;
&lt;chunk reader="itemReader" writer="itemWriter"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
</section>
<section id="mergingListsOnStep">
<title>Merging Lists</title>
<para>Some of the configurable elements on
<classname>Step</classname>s are lists; the &lt;listeners/&gt;
element, for instance. If both the parent and child
<classname>Step</classname>s declare a &lt;listeners/&gt; element,
then the child's list will override the parent's. In order to allow a
child to add additional listeners to the list defined by the parent,
every list element has a "merge" attribute. If the element specifies
that merge="true", then the child's list will be combined with the
parent's instead of overriding it.</para>
<para>In the following example, the <classname>Step</classname>
"concreteStep3" will be created will two listeners:
<classname>listenerOne</classname> and
<classname>listenerTwo</classname>:</para>
<programlisting language="xml">&lt;step id="listenersParentStep" abstract="true"&gt;
&lt;listeners&gt;
&lt;listener ref="listenerOne"/&gt;
&lt;listeners&gt;
&lt;/step&gt;
&lt;step id="concreteStep3" parent="listenersParentStep"&gt;
&lt;tasklet&gt;
&lt;chunk reader="itemReader" writer="itemWriter" commit-interval="5"/&gt;
&lt;/tasklet&gt;
&lt;listeners merge="true"&gt;
&lt;listener ref="listenerTwo"/&gt;
&lt;listeners&gt;
&lt;/step&gt;</programlisting>
</section>
</section>
<section id="commitInterval">
<title>The Commit Interval</title>
<para>As mentioned above, a step reads in and writes out items,
periodically committing using the supplied
<classname>PlatformTransactionManager</classname>. With a
commit-interval of 1, it will commit after writing each individual item.
This is less than ideal in many situations, since beginning and
committing a transaction is expensive. Ideally, it is preferable to
process as many items as possible in each transaction, which is
completely dependent upon the type of data being processed and the
resources with which the step is interacting. For this reason, the
number of items that are processed within a commit can be
configured.</para>
<programlisting language="xml">&lt;job id="sampleJob"&gt;
&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="itemReader" writer="itemWriter" <emphasis
role="bold">commit-interval="10"</emphasis>/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;
&lt;/job&gt;</programlisting>
<para>In the example above, 10 items will be processed within each
transaction. At the beginning of processing a transaction is begun, and
each time <markup>read</markup> is called on the
<classname>ItemReader</classname>, a counter is incremented. When it
reaches 10, the list of aggregated items is passed to the
<classname>ItemWriter</classname>, and the transaction will be
committed.</para>
</section>
<section id="stepRestart">
<title>Configuring a Step for Restart</title>
<para>In <xref linkend="configureJob" />, restarting a
<classname>Job</classname> was discussed. Restart has numerous impacts
on steps, and as such may require some specific configuration.</para>
<section id="startLimit">
<title>Setting a StartLimit</title>
<para>There are many scenarios where you may want to control the
number of times a <classname>Step</classname> may be started. For
example, a particular <classname>Step</classname> might need to be
configured so that it only runs once because it invalidates some
resource that must be fixed manually before it can be run again. This
is configurable on the step level, since different steps may have
different requirements. A <classname>Step</classname> that may only be
executed once can exist as part of the same <classname>Job</classname>
as a <classname>Step</classname> that can be run infinitely. Below is
an example start limit configuration:</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet start-limit="1"&gt;
&lt;chunk reader="itemReader" writer="itemWriter" commit-interval="10"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
<para>The simple step above can be run only once. Attempting to run it
again will cause an exception to be thrown. It should be noted that
the default value for the start-limit is
<classname>Integer.MAX_VALUE</classname>.</para>
</section>
<section id="allowStartIfComplete">
<title>Restarting a completed step</title>
<para>In the case of a restartable job, there may be one or more steps
that should always be run, regardless of whether or not they were
successful the first time. An example might be a validation step, or a
<classname>Step</classname> that cleans up resources before
processing. During normal processing of a restarted job, any step with
a status of 'COMPLETED', meaning it has already been completed
successfully, will be skipped. Setting allow-start-if-complete to
"true" overrides this so that the step will always run:</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet allow-start-if-complete="true"&gt;
&lt;chunk reader="itemReader" writer="itemWriter" commit-interval="10"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
</section>
<section id="stepRestartExample">
<title>Step Restart Configuration Example</title>
<programlisting language="xml">&lt;job id="footballJob" restartable="true"&gt;
&lt;step id="playerload" next="gameLoad"&gt;
&lt;tasklet&gt;
&lt;chunk reader="playerFileItemReader" writer="playerWriter"
commit-interval="10" /&gt;
&lt;/tasklet&gt;
&lt;/step&gt;
&lt;step id="gameLoad" next="playerSummarization"&gt;
&lt;tasklet allow-start-if-complete="true"&gt;
&lt;chunk reader="gameFileItemReader" writer="gameWriter"
commit-interval="10"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;
&lt;step id="playerSummarization"&gt;
&lt;tasklet start-limit="3"&gt;
&lt;chunk reader="playerSummarizationSource" writer="summaryWriter"
commit-interval="10"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;
&lt;/job&gt;</programlisting>
<para>The above example configuration is for a job that loads in
information about football games and summarizes them. It contains
three steps: playerLoad, gameLoad, and playerSummarization. The
playerLoad <classname>Step</classname> loads player information from a
flat file, while the gameLoad <classname>Step</classname> does the
same for games. The final <classname>Step</classname>,
playerSummarization, then summarizes the statistics for each player
based upon the provided games. It is assumed that the file loaded by
'playerLoad' must be loaded only once, but that 'gameLoad' will load
any games found within a particular directory, deleting them after
they have been successfully loaded into the database. As a result, the
playerLoad <classname>Step</classname> contains no additional
configuration. It can be started almost limitlessly, and if complete
will be skipped. The 'gameLoad' <classname>Step</classname>, however,
needs to be run every time in case extra files have been dropped since
it last executed. It has 'allow-start-if-complete' set to 'true' in
order to always be started. (It is assumed that the database tables
games are loaded into has a process indicator on it, to ensure new
games can be properly found by the summarization step). The
summarization <classname>Step</classname>, which is the most important
in the <classname>Job</classname>, is configured to have a start limit
of 3. This is useful because if the step continually fails, a new exit
code will be returned to the operators that control job execution, and
it won't be allowed to start again until manual intervention has taken
place.</para>
<note>
<para>This job is purely for example purposes and is not the same as
the footballJob found in the samples project.</para>
</note>
<para>Run 1:</para>
<orderedlist>
<listitem>
<para>playerLoad is executed and completes successfully, adding
400 players to the 'PLAYERS' table.</para>
</listitem>
<listitem>
<para>gameLoad is executed and processes 11 files worth of game
data, loading their contents into the 'GAMES' table.</para>
</listitem>
<listitem>
<para>playerSummarization begins processing and fails after 5
minutes.</para>
</listitem>
</orderedlist>
<para>Run 2:</para>
<orderedlist>
<listitem>
<para>playerLoad is not run, since it has already completed
successfully, and allow-start-if-complete is 'false' (the
default).</para>
</listitem>
<listitem>
<para>gameLoad is executed again and processes another 2 files,
loading their contents into the 'GAMES' table as well (with a
process indicator indicating they have yet to be processed)</para>
</listitem>
<listitem>
<para>playerSummarization begins processing of all remaining game
data (filtering using the process indicator) and fails again after
30 minutes.</para>
</listitem>
</orderedlist>
<para>Run 3:</para>
<orderedlist>
<listitem>
<para>playerLoad is not run, since it has already completed
successfully, and allow-start-if-complete is 'false' (the
default).</para>
</listitem>
<listitem>
<para>gameLoad is executed again and processes another 2 files,
loading their contents into the 'GAMES' table as well (with a
process indicator indicating they have yet to be processed)</para>
</listitem>
<listitem>
<para>playerSummarization is not start, and the job is immediately
killed, since this is the third execution of playerSummarization,
and its limit is only 2. The limit must either be raised, or the
<classname>Job</classname> must be executed as a new
<classname>JobInstance</classname>.</para>
</listitem>
</orderedlist>
</section>
</section>
<section id="configuringSkip">
<title>Configuring Skip Logic</title>
<para>There are many scenarios where errors encountered while processing
should not result in <classname>Step</classname> failure, but should be
skipped instead. This is usually a decision that must be made by someone
who understands the data itself and what meaning it has. Financial data,
for example, may not be skippable because it results in money being
transferred, which needs to be completely accurate. Loading a list of
vendors, on the other hand, might allow for skips. If a vendor is not
loaded because it was formatted incorrectly or was missing necessary
information, then there probably won't be issues. Usually these bad
records are logged as well, which will be covered later when discussing
listeners.
</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="flatFileItemReader" writer="itemWriter"
commit-interval="10" <emphasis role="bold">skip-limit="10"</emphasis>&gt;
<emphasis role="bold">&lt;skippable-exception-classes&gt;
&lt;include class="org.springframework.batch.item.file.FlatFileParseException"/&gt;
&lt;/skippable-exception-classes&gt;</emphasis>
&lt;/chunk&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
<para>In this example, a <classname>FlatFileItemReader</classname> is
used, and if at any point a
<classname>FlatFileParseException</classname> is thrown, it will be
skipped and counted against the total skip limit of 10. Separate counts
are made of skips on read, process and write inside the step execution,
and the limit applies across all. Once the skip limit is reached, the
next exception found will cause the step to fail.</para>
<para>One problem with the example above is that any other exception
besides a <classname>FlatFileParseException</classname> will cause the
<classname>Job</classname> to fail. In certain scenarios this may be the
correct behavior. However, in other scenarios it may be easier to
identify which exceptions should cause failure and skip everything
else:
</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="flatFileItemReader" writer="itemWriter"
commit-interval="10" <emphasis role="bold">skip-limit="10"</emphasis>&gt;
<emphasis role="bold"> &lt;skippable-exception-classes&gt;
&lt;include class="java.lang.Exception"/&gt;
&lt;exclude class="java.io.FileNotFoundException"/&gt;
&lt;/skippable-exception-classes&gt;
</emphasis> &lt;/chunk&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
<para>By 'including' <classname>java.lang.Exception</classname> as a
skippable exception class, the configuration indicates that all
<classname>Exception</classname>s are skippable. However, by 'excluding'
<classname>java.io.FileNotFoundException</classname>, the configuration
refines the list of skippable exception classes to be all
<classname>Exception</classname>s <emphasis>except</emphasis>
<classname>FileNotFoundException</classname>. Any excluded exception
classes will be fatal if encountered (i.e. not skipped).</para>
<para>For any exception encountered, the skippability will be determined
by the nearest superclass in the class hierarchy. Any unclassifed
exception will be treated as 'fatal'. The order of the
<code>&lt;include/&gt;</code> and <code>&lt;exclude/&gt;</code> elements
does not matter.</para>
</section>
<section id="retryLogic">
<title>Configuring Retry Logic</title>
<para>In most cases you want an exception to cause either a skip or
<classname>Step</classname> failure. However, not all exceptions are
deterministic. If a <classname>FlatFileParseException</classname> is
encountered while reading, it will always be thrown for that record;
resetting the <classname>ItemReader</classname> will not help. However,
for other exceptions, such as a
<classname>DeadlockLoserDataAccessException</classname>, which indicates
that the current process has attempted to update a record that another
process holds a lock on, waiting and trying again might result in
success. In this case, retry should be configured:</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="itemReader" writer="itemWriter"
commit-interval="2" <emphasis role="bold">retry-limit="3"</emphasis>&gt;
<emphasis role="bold">&lt;retryable-exception-classes&gt;
&lt;include class="org.springframework.dao.DeadlockLoserDataAccessException"/&gt;
&lt;/retryable-exception-classes&gt;</emphasis>
&lt;/chunk&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
<para>The <classname>Step</classname> allows a limit for the number of
times an individual item can be retried, and a list of exceptions that
are 'retryable'. More details on how retry works can be found in <xref
linkend="retry" />.</para>
</section>
<section id="controllingRollback">
<title>Controlling Rollback</title>
<para>By default, regardless of retry or skip, any exceptions thrown
from the <classname>ItemWriter</classname> will cause the transaction
controlled by the <classname>Step</classname> to rollback. If skip is
configured as described above, exceptions thrown from the
<classname>ItemReader</classname> will not cause a rollback. However,
there are many scenarios in which exceptions thrown from the
<classname>ItemWriter</classname> should not cause a rollback because no
action has taken place to invalidate the transaction. For this reason,
the <classname>Step</classname> can be configured with a list of
exceptions that should not cause rollback.</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="itemReader" writer="itemWriter" commit-interval="2"/&gt;
&lt;no-rollback-exception-classes&gt;
&lt;include class="org.springframework.batch.item.validator.ValidationException"/&gt;
&lt;/no-rollback-exception-classes&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
<section id="transactionalReaders">
<title>Transactional Readers</title>
<para>The basic contract of the <classname>ItemReader</classname> is
that it is forward only. The step buffers reader input, so that in the
case of a rollback the items don't need to be re-read from the reader.
However, there are certain scenarios in which the reader is built on
top of a transactional resource, such as a JMS queue. In this case,
since the queue is tied to the transaction that is rolled back, the
messages that have been pulled from the queue will be put back on. For
this reason, the step can be configured to not buffer the
items:</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="itemReader" writer="itemWriter" commit-interval="2"
<emphasis role="bold"> is-reader-transactional-queue="true"</emphasis>/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
</section>
</section>
<section id="transactionAttributes">
<title>Transaction Attributes</title>
<para>Transaction attributes can be used to control the isolation,
propagation, and timeout settings. More information on setting
transaction attributes can be found in the spring core
documentation.</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="itemReader" writer="itemWriter" commit-interval="2"/&gt;
&lt;transaction-attributes isolation="DEFAULT"
propagation="REQUIRED"
timeout="30"/&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
</section>
<section id="registeringItemStreams">
<title>Registering ItemStreams with the Step</title>
<para>The step has to take care of <classname>ItemStream</classname>
callbacks at the necessary points in its lifecycle. (for more
information on the <classname>ItemStream</classname> interface, please
refer to <xref linkend="itemStream" />) This is vital if a step fails,
and might need to be restarted, because the
<classname>ItemStream</classname> interface is where the step gets the
information it needs about persistent state between executions.</para>
<para>If the <classname>ItemReader</classname>,
<classname>ItemProcessor</classname>, or
<classname>ItemWriter</classname> itself implements the
<classname>ItemStream</classname> interface, then these will be
registered automatically. Any other streams need to be registered
separately. This is often the case where there are indirect dependencies
such as delegates being injected into the reader and writer. A stream
can be registered on the <classname>Step</classname> through the
'streams' element, as illustrated below:</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="itemReader" writer="compositeWriter" commit-interval="2"&gt;
<emphasis role="bold">&lt;streams&gt;
&lt;stream ref="fileItemWriter1"/&gt;
&lt;stream ref="fileItemWriter2"/&gt;
&lt;/streams&gt;</emphasis>
&lt;/chunk&gt;
&lt;/tasklet&gt;
&lt;/step&gt;
&lt;beans:bean id="compositeWriter"
class="org.springframework.batch.item.support.CompositeItemWriter"&gt;
&lt;beans:property name="delegates"&gt;
&lt;beans:list&gt;
&lt;beans:ref bean="fileItemWriter1" /&gt;
&lt;beans:ref bean="fileItemWriter2" /&gt;
&lt;/beans:list&gt;
&lt;/beans:property&gt;
&lt;/beans:bean&gt;</programlisting>
<para>In the example above, the
<classname>CompositeItemWriter</classname> is not an
<classname>ItemStream</classname>, but both of its delegates are.
Therefore, both delegate writers must be explicitly registered as
streams in order for the framework to handle them correctly. The
<classname>ItemReader</classname> does not need to be explicitly
registered as a stream because it is a direct property of the
<classname>Step</classname>. The step will now be restartable and the
state of the reader and writer will be correctly persisted in the event
of a failure.</para>
</section>
<section id="interceptingStepExecution">
<title>Intercepting Step Execution</title>
<para>Just as with the <classname>Job</classname>, there are many events
during the execution of a <classname>Step</classname> where a user may
need to perform some functionality. For example, in order to write out
to a flat file that requires a footer, the
<classname>ItemWriter</classname> needs to be notified when the
<classname>Step</classname> has been completed, so that the footer can
written. This can be accomplished with one of many
<classname>Step</classname> scoped listeners.</para>
<para>Any class that implements one of the extensions
of <classname>StepListener</classname> (but not that interface
itself since it is empty) can be applied to a step via the
listeners element. The listeners element is valid inside a
step, tasklet or chunk declaration. It is recommended that you
declare the listeners at the level which its function applies,
or if it is multi-featured
(e.g. <classname>StepExecutionListener</classname>
and <classname>ItemReadListener</classname>) then declare it at
the most granular level that it applies (chunk in the example
given).</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet&gt;
&lt;chunk reader="reader" writer="writer" commit-interval="10"/&gt;
&lt;listeners&gt;
&lt;listener ref="chunkListener"/&gt;
&lt;/listeners&gt;
&lt;/tasklet&gt;
&lt;/step&gt;</programlisting>
<para>An <classname>ItemReader</classname>,
<classname>ItemWriter</classname> or
<classname>ItemProcessor</classname> that itself implements one of the
<classname>StepListener</classname> interfaces will be registered
automatically with the <classname>Step</classname> if using the
namespace <literal>&lt;step&gt;</literal> element, or one of the the
<classname>*StepFactoryBean</classname> factories. This only applies to
components directly injected into the <classname>Step</classname>: if
the listener is nested inside another component, it needs to be
explicitly registered (as described above).</para>
<para>In addition to the <classname>StepListener</classname> interfaces,
annotations are provided to address the same concerns. Plain old Java
objects can have methods with these annotations that are then converted
into the corresponding <classname>StepListener</classname> type. It is
also common to annotate custom implementations of chunk components like
<classname>ItemReader</classname> or <classname>ItemWriter</classname>
or <classname>Tasklet</classname>. The annotations are analysed by the
XML parser for the <code>&lt;listener/&gt;</code> elements, so all you
need to do is use the XML namespace to register the listeners with a
step.</para>
<section id="stepExecutionListener">
<title>StepExecutionListener</title>
<para><classname>StepExecutionListener</classname> represents the most
generic listener for <classname>Step</classname> execution. It allows
for notification before a <classname>Step</classname> is started and
after it has ends, whether it ended normally or failed:</para>
<programlisting language="java">public interface StepExecutionListener extends StepListener {
void beforeStep(StepExecution stepExecution);
ExitStatus afterStep(StepExecution stepExecution);
}</programlisting>
<para><classname>ExitStatus</classname> is the return type of
<methodname>afterStep</methodname> in order to allow listeners the
chance to modify the exit code that is returned upon completion of a
<classname>Step</classname>.</para>
<para>The annotations corresponding to this interface are:</para>
<itemizedlist>
<listitem>
<para><classname>@BeforeStep</classname></para>
</listitem>
<listitem>
<para><classname>@AfterStep</classname></para>
</listitem>
</itemizedlist>
</section>
<section id="chunkListener">
<title>ChunkListener</title>
<para>A chunk is defined as the items processed within the scope of a
transaction. Committing a transaction, at each commit interval,
commits a 'chunk'. A <classname>ChunkListener</classname> can be
useful to perform logic before a chunk begins processing or after a
chunk has completed successfully:</para>
<programlisting language="java">public interface ChunkListener extends StepListener {
void beforeChunk();
void afterChunk();
}</programlisting>
<para>The <methodname>beforeChunk</methodname> method is called after
the transaction is started, but before <methodname>read</methodname>
is called on the <classname>ItemReader</classname>. Conversely,
<methodname>afterChunk</methodname> is called after the chunk has been
committed (and not at all if there is a rollback).</para>
<para>The annotations corresponding to this interface are:</para>
<itemizedlist>
<listitem>
<para><classname>@BeforeChunk</classname></para>
</listitem>
<listitem>
<para><classname>@AfterChunk</classname></para>
</listitem>
</itemizedlist>
<para>A <classname>ChunkListener</classname> can be applied
when there is no chunk declaration: it is
the <classname>TaskletStep</classname> that is responsible for
calling the <classname>ChunkListener</classname> so it applies
to a non-item-oriented tasklet as well (called before and
after the tasklet).</para>
</section>
<section id="itemReadListener">
<title>ItemReadListener</title>
<para>When discussing skip logic above, it was mentioned that it may
be beneficial to log the skipped records, so that they can be deal
with later. In the case of read errors, this can be done with an
<classname>ItemReaderListener:</classname>
</para>
<programlisting language="java">public interface ItemReadListener&lt;T&gt; extends StepListener {
void beforeRead();
void afterRead(T item);
void onReadError(Exception ex);
}</programlisting>
<para>The <methodname>beforeRead</methodname> method will be called
before each call to <methodname>read</methodname> on the
<classname>ItemReader</classname>. The
<methodname>afterRead</methodname> method will be called after each
successful call to <methodname>read</methodname>, and will be passed
the item that was read. If there was an error while reading, the
<classname>onReadError</classname> method will be called. The
exception encountered will be provided so that it can be
logged.</para>
<para>The annotations corresponding to this interface are:</para>
<itemizedlist>
<listitem>
<para><classname>@BeforeRead</classname></para>
</listitem>
<listitem>
<para><classname>@AfterRead</classname></para>
</listitem>
<listitem>
<para><classname>@OnReadError</classname></para>
</listitem>
</itemizedlist>
</section>
<section id="itemProcessListener">
<title>ItemProcessListener</title>
<para>Just as with the <classname>ItemReadListener</classname>, the
processing of an item can be 'listened' to:</para>
<programlisting language="java">public interface ItemProcessListener&lt;T, S&gt; extends StepListener {
void beforeProcess(T item);
void afterProcess(T item, S result);
void onProcessError(T item, Exception e);
}</programlisting>
<para>The <methodname>beforeProcess</methodname> method will be called
before <methodname>process</methodname> on the
<classname>ItemProcessor</classname>, and is handed the item that will
be processed. The <methodname>afterProcess</methodname> method will be
called after the item has been successfully processed. If there was an
error while processing, the <methodname>onProcessError</methodname>
method will be called. The exception encountered and the item that was
attempted to be processed will be provided, so that they can be
logged.</para>
<para>The annotations corresponding to this interface are:</para>
<itemizedlist>
<listitem>
<para><classname>@BeforeProcess</classname></para>
</listitem>
<listitem>
<para><classname>@AfterProcess</classname></para>
</listitem>
<listitem>
<para><classname>@OnProcessError</classname></para>
</listitem>
</itemizedlist>
</section>
<section id="itemWriteListener">
<title>ItemWriteListener</title>
<para>The writing of an item can be 'listened' to with the
<classname>ItemWriteListener</classname>:</para>
<programlisting language="java">public interface ItemWriteListener&lt;S&gt; extends StepListener {
void beforeWrite(List&lt;? extends S&gt; items);
void afterWrite(List&lt;? extends S&gt; items);
void onWriteError(Exception exception, List&lt;? extends S&gt; items);
}</programlisting>
<para>The <methodname>beforeWrite</methodname> method will be called
before <methodname>write</methodname> on the
<classname>ItemWriter</classname>, and is handed the item that will be
written. The <methodname>afterWrite</methodname> method will be called
after the item has been successfully written. If there was an error
while writing, the <methodname>onWriteError</methodname> method will
be called. The exception encountered and the item that was attempted
to be written will be provided, so that they can be logged.</para>
<para>The annotations corresponding to this interface are:</para>
<itemizedlist>
<listitem>
<para><classname>@BeforeWrite</classname></para>
</listitem>
<listitem>
<para><classname>@AfterWrite</classname></para>
</listitem>
<listitem>
<para><classname>@OnWriteError</classname></para>
</listitem>
</itemizedlist>
</section>
<section id="skipListener">
<title>SkipListener</title>
<para><classname>ItemReadListener</classname>,
<classname>ItemProcessListener</classname>, and
<classname>ItemWriteListner</classname> all provide mechanisms for
being notified of errors, but none will inform you that a record has
actually been skipped. <methodname>onWriteError</methodname>, for
example, will be called even if an item is retried and successful. For
this reason, there is a separate interface for tracking skipped
items:</para>
<programlisting language="java">public interface SkipListener&lt;T,S&gt; extends StepListener {
void onSkipInRead(Throwable t);
void onSkipInProcess(T item, Throwable t);
void onSkipInWrite(S item, Throwable t);
}</programlisting>
<para><methodname>onSkipInRead</methodname> will be called whenever an
item is skipped while reading. It should be noted that rollbacks may
cause the same item to be registered as skipped more than once.
<methodname>onSkipInWrite</methodname> will be called when an item is
skipped while writing. Because the item has been read successfully
(and not skipped), it is also provided the item itself as an
argument.</para>
<para>The annotations corresponding to this interface are:</para>
<itemizedlist>
<listitem>
<para><classname>@OnSkipInRead</classname></para>
</listitem>
<listitem>
<para><classname>@OnSkipInWrite</classname></para>
</listitem>
<listitem>
<para><classname>@OnSkipInProcess</classname></para>
</listitem>
</itemizedlist>
<section id="skipListenersAndTransactions">
<title>SkipListeners and Transactions</title>
<para>One of the most common use cases for a
<classname>SkipListener</classname> is to log out a skipped item, so
that another batch process or even human process can be used to
evaluate and fix the issue leading to the skip. Because there are
many cases in which the original transaction may be rolled back,
Spring Batch makes two guarantees:</para>
<orderedlist>
<listitem>
<para>The appropriate skip method (depending on when the error
happened) will only be called once per item.</para>
</listitem>
<listitem>
<para>The <classname>SkipListener</classname> will always be
called just before the transaction is committed. This is to
ensure that any transactional resources call by the listener are
not rolled back by a failure within the
<classname>ItemWriter</classname>.</para>
</listitem>
</orderedlist>
</section>
</section>
</section>
</section>
<section id="taskletStep">
<title>TaskletStep</title>
<para>Chunk-oriented processing is not the only way to process in a
<classname>Step</classname>. What if a <classname>Step</classname> must
consist as a simple stored procedure call? You could implement the call as
an <classname>ItemReader</classname> and return null after the procedure
finishes, but it is a bit unnatural since there would need to be a no-op
<classname>ItemWriter</classname>. Spring Batch provides the
<classname>TaskletStep</classname> for this scenario.</para>
<para>The <classname>Tasklet</classname> is a simple interface that has
one method, <methodname>execute</methodname>, which will be a called
repeatedly by the <classname>TaskletStep</classname> until it either
returns <literal>RepeatStatus.FINISHED</literal> or throws an exception to
signal a failure. Each call to the <classname>Tasklet</classname> is
wrapped in a transaction. <classname>Tasklet</classname> implementors
might call a stored procedure, a script, or a simple SQL update statement.
To create a <classname>TaskletStep</classname>, the 'ref' attribute of the
&lt;tasklet/&gt; element should reference a bean defining a
<classname>Tasklet</classname> object; no &lt;chunk/&gt; element should be
used within the &lt;tasklet/&gt;:</para>
<programlisting language="xml">&lt;step id="step1"&gt;
&lt;tasklet ref="myTasklet"/&gt;
&lt;/step&gt;</programlisting>
<note>
<para><classname>TaskletStep</classname> will automatically register the
tasklet as <classname>StepListener</classname> if it implements this
interface</para>
</note>
<section id="taskletAdapter">
<title>TaskletAdapter</title>
<para>As with other adapters for the <classname>ItemReader</classname>
and <classname>ItemWriter</classname> interfaces, the
<classname>Tasklet</classname> interface contains an implementation that
allows for adapting itself to any pre-existing class:
<classname>TaskletAdapter</classname>. An example where this may be
useful is an existing DAO that is used to update a flag on a set of
records. The <classname>TaskletAdapter</classname> can be used to call
this class without having to write an adapter for the
<classname>Tasklet</classname> interface:</para>
<programlisting language="xml">&lt;bean id="myTasklet" class="o.s.b.core.step.tasklet.MethodInvokingTaskletAdapter"&gt;
&lt;property name="targetObject"&gt;
&lt;bean class="org.mycompany.FooDao"/&gt;
&lt;/property&gt;
&lt;property name="targetMethod" value="updateFoo" /&gt;
&lt;/bean&gt;</programlisting>
</section>
<section id="exampleTaskletImplementation">
<title>Example Tasklet Implementation</title>
<para>Many batch jobs contain steps that must be done before the main
processing begins in order to set up various resources or after
processing has completed to cleanup those resources. In the case of a
job that works heavily with files, it is often necessary to delete
certain files locally after they have been uploaded successfully to
another location. The example below taken from the Spring Batch samples
project, is a <classname>Tasklet</classname> implementation with just
such a responsibility:</para>
<programlisting language="java">public class FileDeletingTasklet implements Tasklet, InitializingBean {
private Resource directory;
public RepeatStatus execute(StepContribution contribution,
ChunkContext chunkContext) throws Exception {
File dir = directory.getFile();
Assert.state(dir.isDirectory());
File[] files = dir.listFiles();
for (int i = 0; i &lt; files.length; i++) {
boolean deleted = files[i].delete();
if (!deleted) {
throw new UnexpectedJobExecutionException("Could not delete file " +
files[i].getPath());
}
}
return RepeatStatus.FINISHED;
}
public void setDirectoryResource(Resource directory) {
this.directory = directory;
}
public void afterPropertiesSet() throws Exception {
Assert.notNull(directory, "directory must be set");
}
}</programlisting>
<para>The above <classname>Tasklet</classname> implementation will
delete all files within a given directory. It should be noted that the
<methodname>execute</methodname> method will only be called once. All
that is left is to reference the <classname>Tasklet</classname> from the
<classname>Step</classname>:</para>
<programlisting language="xml">&lt;job id="taskletJob"&gt;
&lt;step id="deleteFilesInDir"&gt;
&lt;tasklet ref="fileDeletingTasklet"/&gt;
&lt;/step&gt;
&lt;/job&gt;
&lt;beans:bean id="fileDeletingTasklet"
class="org.springframework.batch.sample.tasklet.FileDeletingTasklet"&gt;
&lt;beans:property name="directoryResource"&gt;
&lt;beans:bean id="directory"
class="org.springframework.core.io.FileSystemResource"&gt;
&lt;beans:constructor-arg value="target/test-outputs/test-dir" /&gt;
&lt;/beans:bean&gt;
&lt;/beans:property&gt;
&lt;/beans:bean&gt;</programlisting>
</section>
</section>
<section id="controllingStepFlow">
<title>Controlling Step Flow</title>
<para>With the ability to group steps together within an owning job comes
the need to be able to control how the job 'flows' from one step to
another. The failure of a <classname>Step</classname> doesn't necessarily
mean that the <classname>Job</classname> should fail. Furthermore, there
may be more than one type of 'success' which determines which
<classname>Step</classname> should be executed next. Depending upon how a
group of Steps is configured, certain steps may not even be processed at
all.</para>
<section id="SequentialFlow">
<title>Sequential Flow</title>
<para>The simplest flow scenario is a job where all of the steps execute
sequentially:</para>
<mediaobject>
<imageobject role="html">
<imagedata align="center" fileref="images/sequential-flow.png"
scale="20" />
</imageobject>
<imageobject role="fo">
<imagedata align="center" fileref="images/sequential-flow.png"
scale="45" />
</imageobject>
</mediaobject>
<para>This can be achieved using the 'next' attribute of the step
element:</para>
<programlisting language="xml">&lt;job id="job"&gt;
&lt;step id="stepA" parent="s1" next="stepB" /&gt;
&lt;step id="stepB" parent="s2" next="stepC"/&gt;
&lt;step id="stepC" parent="s3" /&gt;
&lt;/job&gt;</programlisting>
<para>In the scenario above, 'step A' will execute
first because it is the first <classname>Step</classname> listed. If
'step A' completes normally, then 'step B' will execute, and so on.
However, if 'step A' fails, then the entire <classname>Job</classname>
will fail and 'step B' will not execute.</para>
<note>
<para>With the Spring Batch namespace, the first step listed in the
configuration will <emphasis>always</emphasis> be the first step
executed by the <classname>Job</classname>. The order of the other
step elements does not matter, but the first step must always appear
first in the xml.</para>
</note>
</section>
<section id="conditionalFlow">
<title>Conditional Flow</title>
<para>In the example above, there are only two possibilities:</para>
<orderedlist>
<listitem>
<para>The <classname>Step</classname> is successful and the next
<classname>Step</classname> should be executed.</para>
</listitem>
<listitem>
<para>The <classname>Step</classname> failed and thus the
<classname>Job</classname> should fail.</para>
</listitem>
</orderedlist>
<para>In many cases, this may be sufficient. However, what about a
scenario in which the failure of a <classname>Step</classname> should
trigger a different <classname>Step</classname>, rather than causing
failure? <mediaobject>
<imageobject role="html">
<imagedata align="center" fileref="images/conditional-flow.png"
scale="50" />
</imageobject>
<imageobject role="fo">
<imagedata align="center" fileref="images/conditional-flow.png"
scale="50" />
</imageobject>
</mediaobject></para>
<para id="nextElement">In order to handle more complex scenarios, the
Spring Batch namespace allows transition elements to be defined within
the step element. One such transition is the "next" element. Like the
"next" attribute, the "next" element will tell the
<classname>Job</classname> which <classname>Step</classname> to execute
next. However, unlike the attribute, any number of "next" elements are
allowed on a given <classname>Step</classname>, and there is no default
behavior the case of failure. This means that if transition elements are
used, then all of the behavior for the <classname>Step</classname>'s
transitions must be defined explicitly. Note also that a single step
cannot have both a "next" attribute and a transition element.</para>
<para>The next element specifies a pattern to match and the step to
execute next:</para>
<programlisting language="xml">&lt;job id="job"&gt;
&lt;step id="stepA" parent="s1"&gt;
&lt;next on="*" to="stepB" /&gt;
&lt;next on="FAILED" to="stepC" /&gt;
&lt;/step&gt;
&lt;step id="stepB" parent="s2" next="stepC" /&gt;
&lt;step id="stepC" parent="s3" /&gt;
&lt;/job&gt;</programlisting>
<para>The "on" attribute of a transition element uses a simple
pattern-matching scheme to match the <classname>ExitStatus</classname>
that results from the execution of the <classname>Step</classname>. Only
two special characters are allowed in the pattern:</para>
<itemizedlist>
<listitem>
<para>"*" will zero or more characters</para>
</listitem>
<listitem>
<para>"?" will match exactly one character</para>
</listitem>
</itemizedlist>
<para>For example, "c*t" will match "cat" and "count", while "c?t" will
match "cat" but not "count".</para>
<para>While there is no limit to the number of transition elements on a
<classname>Step</classname>, if the <classname>Step</classname>'s
execution results in an <classname>ExitStatus</classname> that is not
covered by an element, then the framework will throw an exception and
the <classname>Job</classname> will fail. The framework will
automatically order transitions from most specific to
least specific. This means that even if the elements were swapped for
"stepA" in the example above, an <classname>ExitStatus</classname> of
"FAILED" would still go to "stepC".</para>
<section id="batchStatusVsExitStatus">
<title>Batch Status vs. Exit Status</title>
<para>When configuring a <classname>Job</classname> for conditional
flow, it is important to understand the difference between
<classname>BatchStatus</classname> and
<classname>ExitStatus</classname>. <classname>BatchStatus</classname>
is an enumeration that is a property of both
<classname>JobExecution</classname> and
<classname>StepExecution</classname> and is used by the framework to
record the status of a <classname>Job</classname> or
<classname>Step</classname>. It can be one of the following values:
COMPLETED, STARTING, STARTED, STOPPING, STOPPED, FAILED, ABANDONED or
UNKNOWN. Most of them are self explanatory: COMPLETED is the status
set when a step or job has completed successfully, FAILED is set when
it fails, and so on. The example above contains the following 'next'
element:</para>
<programlisting language="xml">&lt;next on="FAILED" to="stepB" /&gt;</programlisting>
<para>At first glance, it would appear that the 'on' attribute
references the <classname>BatchStatus</classname> of the
<classname>Step</classname> to which it belongs. However, it actually
references the <classname>ExitStatus</classname> of the
<classname>Step</classname>. As the name implies,
<classname>ExitStatus</classname> represents the status of a
<classname>Step</classname> after it finishes execution. More
specifically, the 'next' element above references the exit code of the
<classname>ExitStatus</classname>. To write it in English, it says:
"go to stepB if the exit code is FAILED". By default, the exit code is
always the same as the <classname>BatchStatus</classname> for the
Step, which is why the entry above works. However, what if the exit
code needs to be different? A good example comes from the skip sample
job within the samples project:</para>
<programlisting language="xml">&lt;step id="step1" parent="s1"&gt;
&lt;end on="FAILED" /&gt;
&lt;next on="COMPLETED WITH SKIPS" to="errorPrint1" /&gt;
&lt;next on="*" to="step2" /&gt;
&lt;/step&gt;</programlisting>
<para>The above step has three possibilities:</para>
<orderedlist>
<listitem>
<para>The <classname>Step</classname> failed, in which case the
job should fail.</para>
</listitem>
<listitem>
<para>The <classname>Step</classname> completed
successfully.</para>
</listitem>
<listitem>
<para>The <classname>Step</classname> completed successfully, but
with an exit code of 'COMPLETED WITH SKIPS'. In this case, a
different step should be run to handle the errors.</para>
</listitem>
</orderedlist>
<para>The above configuration will work. However, something needs to
change the exit code based on the condition of the execution having
skipped records:</para>
<programlisting language="java">public class SkipCheckingListener extends StepExecutionListenerSupport {
public ExitStatus afterStep(StepExecution stepExecution) {
String exitCode = stepExecution.getExitStatus().getExitCode();
if (!exitCode.equals(ExitStatus.FAILED.getExitCode()) &amp;&amp;
stepExecution.getSkipCount() &gt; 0) {
return new ExitStatus("COMPLETED WITH SKIPS");
}
else {
return null;
}
}
}</programlisting>
<para>The above code is a <classname>StepExecutionListener</classname>
that first checks to make sure the <classname>Step</classname> was
successful, and next if the skip count on the
<classname>StepExecution</classname> is higher than 0. If both
conditions are met, a new <classname>ExitStatus</classname> with an
exit code of "COMPLETED WITH SKIPS" is returned.</para>
</section>
</section>
<section id="configuringForStop">
<title>Configuring for Stop</title>
<para>After the discussion of <link
linkend="batchStatusVsExitStatus"><classname>BatchStatus</classname> and
<classname>ExitStatus</classname></link>, one might wonder how the
<classname>BatchStatus</classname> and <classname>ExitStatus</classname>
are determined for the <classname>Job</classname>. While these statuses
are determined for the <classname>Step</classname> by the code that is
executed, the statuses for the <classname>Job</classname> will be
determined based on the configuration.</para>
<para>So far, all of the job configurations discussed have had at least
one final <classname>Step</classname> with no transitions. For example,
after the following step executes, the <classname>Job</classname> will
end:</para>
<programlisting language="xml">&lt;step id="stepC" parent="s3"/&gt;</programlisting>
<para>If no transitions are defined for a <classname>Step</classname>,
then the <classname>Job</classname>'s statuses will be defined as
follows:</para>
<itemizedlist>
<listitem>
<para>If the <classname>Step</classname> ends with
<classname>ExitStatus</classname> FAILED, then the
<classname>Job</classname>'s <classname>BatchStatus</classname> and
<classname>ExitStatus</classname> will both be FAILED.</para>
</listitem>
<listitem>
<para>Otherwise, the <classname>Job</classname>'s
<classname>BatchStatus</classname> and
<classname>ExitStatus</classname> will both be COMPLETED.</para>
</listitem>
</itemizedlist>
<para>While this method of terminating a batch job is sufficient for
some batch jobs, such as a simple sequential step job, custom defined
job-stopping scenarios may be required. For this purpose, Spring Batch
provides three transition elements to stop a <classname>Job</classname>
(in addition to the <link linkend="nextElement">"next" element</link>
that we discussed previously). Each of these stopping elements will stop
a <classname>Job</classname> with a particular
<classname>BatchStatus</classname>. It is important to note that the
stop transition elements will have no effect on either the
<classname>BatchStatus</classname> or <classname>ExitStatus</classname>
of any <classname>Step</classname>s in the <classname>Job</classname>:
these elements will only affect the final statuses of the
<classname>Job</classname>. For example, it is possible for every step
in a job to have a status of FAILED but the job to have a status of
COMPLETED, or vise versa.</para>
<section id="endElement">
<title>The 'End' Element</title>
<para>The 'end' element instructs a <classname>Job</classname> to stop
with a <classname>BatchStatus</classname> of COMPLETED. A
<classname>Job</classname> that has finished with status COMPLETED
cannot be restarted (the framework will throw a
<classname>JobInstanceAlreadyCompleteException</classname>). The 'end'
element also allows for an optional 'exit-code' attribute that can be
used to customize the <classname>ExitStatus</classname> of the
<classname>Job</classname>. If no 'exit-code' attribute is given, then
the <classname>ExitStatus</classname> will be "COMPLETED" by default,
to match the <classname>BatchStatus</classname>.</para>
<para>In the following scenario, if step2 fails, then the
<classname>Job</classname> will stop with a
<classname>BatchStatus</classname> of COMPLETED and an
<classname>ExitStatus</classname> of "COMPLETED" and step3 will not
execute; otherwise, execution will move to step3. Note that if step2
fails, the <classname>Job</classname> will not be restartable (because
the status is COMPLETED).</para>
<programlisting language="xml">&lt;step id="step1" parent="s1" next="step2"&gt;
&lt;step id="step2" parent="s2"&gt;
&lt;end on="FAILED"/&gt;
&lt;next on="*" to="step3"/&gt;
&lt;/step&gt;
&lt;step id="step3" parent="s3"&gt;</programlisting>
</section>
<section id="failElement">
<title>The 'Fail' Element</title>
<para>The 'fail' element instructs a <classname>Job</classname> to
stop with a <classname>BatchStatus</classname> of FAILED. Unlike the
'end' element, the 'fail' element will not prevent the
<classname>Job</classname> from being restarted. The 'fail' element
also allows for an optional 'exit-code' attribute that can be used to
customize the <classname>ExitStatus</classname> of the
<classname>Job</classname>. If no 'exit-code' attribute is given, then
the <classname>ExitStatus</classname> will be "FAILED" by default, to
match the <classname>BatchStatus</classname>.</para>
<para>In the following scenario, if step2 fails, then the
<classname>Job</classname> will stop with a
<classname>BatchStatus</classname> of FAILED and an
<classname>ExitStatus</classname> of "EARLY TERMINATION" and step3
will not execute; otherwise, execution will move to step3.
Additionally, if step2 fails, and the <classname>Job</classname> is
restarted, then execution will begin again on step2.</para>
<programlisting language="xml">&lt;step id="step1" parent="s1" next="step2"&gt;
&lt;step id="step2" parent="s2"&gt;
&lt;fail on="FAILED" exit-code="EARLY TERMINATION"/&gt;
&lt;next on="*" to="step3"/&gt;
&lt;/step&gt;
&lt;step id="step3" parent="s3"&gt;</programlisting>
</section>
<section id="stopElement">
<title>The 'Stop' Element</title>
<para>The 'stop' element instructs a <classname>Job</classname> to
stop with a <classname>BatchStatus</classname> of STOPPED. Stopping a
<classname>Job</classname> can provide a temporary break in processing
so that the operator can take some action before restarting the
<classname>Job</classname>. The 'stop' element requires a 'restart'
attribute that specifies the step where execution should pick up when
the <classname>Job is restarted</classname>.</para>
<para>In the following scenario, if step1 finishes with COMPLETE, then
the job will then stop. Once it is restarted, execution will begin on
step2.</para>
<programlisting language="xml">&lt;step id="step1" parent="s1"&gt;
&lt;stop on="COMPLETED" restart="step2"/&gt;
&lt;/step&gt;
&lt;step id="step2" parent="s2"/&gt;</programlisting>
</section>
</section>
<section id="programmaticFlowDecisions">
<title>Programmatic Flow Decisions</title>
<para>In some situations, more information than the
<classname>ExitStatus</classname> may be required to decide which step
to execute next. In this case, a
<classname>JobExecutionDecider</classname> can be used to assist in the
decision.</para>
<programlisting language="java">public class MyDecider implements JobExecutionDecider {
public FlowExecutionStatus decide(JobExecution jobExecution, StepExecution stepExecution) {
if (someCondition) {
return "FAILED";
}
else {
return "COMPLETED";
}
}
}</programlisting>
<para>In the job configuration, a "decision" tag will specify the
decider to use as well as all of the transitions.</para>
<programlisting language="xml">&lt;job id="job"&gt;
&lt;step id="step1" parent="s1" next="decision" /&gt;
&lt;decision id="decision" decider="decider"&gt;
&lt;next on="FAILED" to="step2" /&gt;
&lt;next on="COMPLETED" to="step3" /&gt;
&lt;/decision&gt;
&lt;step id="step2" parent="s2" next="step3"/&gt;
&lt;step id="step3" parent="s3" /&gt;
&lt;/job&gt;
&lt;beans:bean id="decider" class="com.MyDecider"/&gt;</programlisting>
</section>
<section id="split-flows">
<title>Split Flows</title>
<para>Every scenario described so far has involved a
<classname>Job</classname> that executes its
<classname>Step</classname>s one at a time in a linear fashion. In
addition to this typical style, the Spring Batch namespace also allows
for a job to be configured with parallel flows using the 'split'
element. As is seen below, the 'split' element contains one or more
'flow' elements, where entire separate flows can be defined. A 'split'
element may also contain any of the previously discussed transition
elements such as the 'next' attribute or the 'next', 'end', 'fail', or
'pause' elements.</para>
<programlisting language="xml">&lt;split id="split1" next="step4"&gt;
&lt;flow&gt;
&lt;step id="step1" parent="s1" next="step2"/&gt;
&lt;step id="step2" parent="s2"/&gt;
&lt;/flow&gt;
&lt;flow&gt;
&lt;step id="step3" parent="s3"/&gt;
&lt;/flow&gt;
&lt;/split&gt;
&lt;step id="step4" parent="s4"/&gt;</programlisting>
</section>
<section id="external-flows">
<title>Externalizing Flow Definitions and Dependencies Between
Jobs</title>
<para>Part of the flow in a job can be externalized as a separate bean
definition, and then re-used. There are two ways to do this, and the
first is to simply declare the flow as a reference to one defined
elsewhere:</para>
<programlisting language="xml">&lt;job id="job"&gt;
&lt;flow id="job1.flow1" parent="flow1" next="step3"/&gt;
&lt;step id="step3" parent="s3"/&gt;
&lt;/job&gt;
&lt;flow id="flow1"&gt;
&lt;step id="step1" parent="s1" next="step2"/&gt;
&lt;step id="step2" parent="s2"/&gt;
&lt;/flow&gt;</programlisting>
<para>The effect of defining an external flow like this is simply to
insert the steps from the external flow into the job as if they had been
declared inline. In this way many jobs can refer to the same template
flow and compose such templates into different logical flows. This is
also a good way to separate the integration testing of the individual
flows.</para>
<para>The other form of an externalized flow is to use a
<classname>JobStep</classname>. A <classname>JobStep</classname> is
similar to a <classname>FlowStep</classname>, but actually creates and
launches a separate job execution for the steps in the flow specified.
Here is an example:</para>
<programlisting language="xml">&lt;job id="jobStepJob" restartable="true"&gt;
&lt;step id="jobStepJob.step1"&gt;
&lt;job ref="<emphasis role="bold">job</emphasis>" job-launcher="jobLauncher"
job-parameters-extractor="jobParametersExtractor"/&gt;
&lt;/step&gt;
&lt;/job&gt;
&lt;job id="<emphasis role="bold">job</emphasis>" restartable="true"&gt;...&lt;/job&gt;
&lt;bean id="jobParametersExtractor" class="org.spr...DefaultJobParametersExtractor"&gt;
&lt;property name="keys" value="input.file"/&gt;
&lt;/bean&gt;</programlisting>
<para>The job parameters extractor is a strategy that determines how a
the <classname>ExecutionContext</classname> for the
<classname>Step</classname> is converted into
<classname>JobParameters</classname> for the Job that is executed. The
<classname>JobStep</classname> is useful when you want to have some more
granular options for monitoring and reporting on jobs and steps. Using
<classname>JobStep</classname> is also often a good answer to the
question: "How do I create dependencies between jobs?". It is a good way
to break up a large system into smaller modules and control the flow of
jobs.</para>
</section>
</section>
<section id="late-binding" xreflabel="Late Binding">
<title>Late Binding of Job and Step Attributes</title>
<para>Both the XML and Flat File examples above use the Spring
<classname>Resource</classname> abstraction to obtain a file. This works
because <classname>Resource</classname> has a <markup>getFile</markup>
method, which returns a <classname>java.io.File</classname>. Both XML and
Flat File resources can be configured using standard Spring
constructs:</para>
<programlisting language="xml">&lt;bean id="flatFileItemReader"
class="org.springframework.batch.item.file.FlatFileItemReader"&gt;
&lt;property name="resource"
value="file://outputs/20070122.testStream.CustomerReportStep.TEMP.txt" /&gt;
&lt;/bean&gt;</programlisting>
<para>The above <classname>Resource</classname> will load the file from
the file system location specified. Note that absolute locations have to
start with a double slash ("//"). In most spring applications, this
solution is good enough because the names of these are known at compile
time. However, in batch scenarios, the file name may need to be determined
at runtime as a parameter to the job. This could be solved using '-D'
parameters, i.e. a system property:</para>
<programlisting language="xml">&lt;bean id="flatFileItemReader"
class="org.springframework.batch.item.file.FlatFileItemReader"&gt;
&lt;property name="resource" value="${input.file.name}" /&gt;
&lt;/bean&gt;</programlisting>
<para>All that would be required for this solution to work would be a
system argument (-Dinput.file.name="file://file.txt"). (Note that although
a <classname>PropertyPlaceholderConfigurer</classname> can be used here,
it is not necessary if the system property is always set because the
<classname>ResourceEditor</classname> in Spring already filters and does
placeholder replacement on system properties.)</para>
<para>Often in a batch setting it is preferable to parameterize the file
name in the <link
linkend="jobParameters"><classname>JobParameters</classname></link> of the
job, instead of through system properties, and access them that way. To
accomplish this, Spring Batch allows for the late binding of various Job
and Step attributes:</para>
<programlisting language="xml">&lt;bean id="flatFileItemReader" scope="step"
class="org.springframework.batch.item.file.FlatFileItemReader"&gt;
&lt;property name="resource" value="<emphasis role="bold">#{jobParameters['input.file.name']}</emphasis>" /&gt;
&lt;/bean&gt;</programlisting>
<para>Both the <classname>JobExecution</classname> and
<classname>StepExecution</classname> level
<classname>ExecutionContext</classname> can be accessed in the same
way:</para>
<programlisting language="xml">&lt;bean id="flatFileItemReader" scope="step"
class="org.springframework.batch.item.file.FlatFileItemReader"&gt;
&lt;property name="resource" value="<emphasis role="bold">#{jobExecutionContext['input.file.name']}</emphasis>" /&gt;
&lt;/bean&gt;</programlisting>
<programlisting language="xml">&lt;bean id="flatFileItemReader" scope="step"
class="org.springframework.batch.item.file.FlatFileItemReader"&gt;
&lt;property name="resource" value="<emphasis role="bold">#{stepExecutionContext['input.file.name']}</emphasis>" /&gt;
&lt;/bean&gt;</programlisting>
<note>
<para>Any bean that uses late-binding must be declared with
scope="step". See for <xref linkend="step-scope" /> more
information.</para>
</note>
<note>
<para>If you are using Spring 3.0 (or above) the expressions in
step-scoped beans are in the Spring Expression Language, a powerful
general purpose language with many interesting features. To provide
backward compatibility, if Spring Batch detects the presence of older
versions of Spring it uses a native expression language that is less
powerful, and has slightly different parsing rules. The main difference
is that the map keys in the example above do not need to be quoted with
Spring 2.5, but the quotes are mandatory in Spring 3.0.</para>
</note>
<section id="step-scope">
<title>Step Scope</title>
<para>All of the late binding examples from above have a scope of "step"
declared on the bean definition:</para>
<programlisting language="xml">&lt;bean id="flatFileItemReader" <emphasis role="bold">scope="step"</emphasis>
class="org.springframework.batch.item.file.FlatFileItemReader"&gt;
&lt;property name="resource" value="#{jobParameters[input.file.name]}" /&gt;
&lt;/bean&gt;</programlisting>
<para>Using a scope of <classname>Step</classname> is required in order
to use late binding since the bean cannot actually be instantiated until
the <classname>Step</classname> starts, which allows the attributes to
be found. Because it is not part of the Spring container by default, the
scope must be added explicitly, either by using the
<literal>batch</literal> namespace:</para>
<programlisting language="xml">&lt;beans xmlns="http://www.springframework.org/schema/beans"
xmlns:batch="http://www.springframework.org/schema/batch"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="..."&gt;
&lt;batch:job .../&gt;
...
&lt;/beans&gt;</programlisting>
<para>or by including a bean definition explicitly for the<classname>
StepScope</classname> (but not both):</para>
<programlisting language="xml">&lt;bean class="org.springframework.batch.core.scope.StepScope" /&gt;</programlisting>
</section>
<section id="job-scope">
<title>Job Scope</title>
<para>Job scope, introduced in Spring Batch 3.0 is similar to Step scope
in configuration but is a Scope for the Job context so there is only one
instance of such a bean per executing job. Additionally, support is provided
for late binding of references accessible from the JobContext using
#{..} placeholders. Using this feature, bean properties can be pulled from
the job or job execution context and the job parameters. E.g.
</para>
<programlisting language="xml">&lt;bean id=&quot;...&quot; class=&quot;...&quot; <emphasis role="bold">scope=&quot;job&quot;</emphasis>&gt;
&lt;property name=&quot;name&quot; value=&quot;#{jobParameters[input]}&quot; /&gt;
&lt;/bean&gt;
</programlisting>
<programlisting language="xml">&lt;bean id=&quot;...&quot; class=&quot;...&quot; <emphasis role="bold">scope=&quot;job&quot;</emphasis>&gt;
&lt;property name=&quot;name&quot; value=&quot;#{jobExecutionContext['input.name']}.txt&quot; /&gt;
&lt;/bean&gt;
</programlisting>
<para>Because it is not part of the Spring container by default, the scope
must be added explicitly, either by using the <literal>batch</literal> namespace:</para>
<programlisting language="xml">&lt;beans xmlns="http://www.springframework.org/schema/beans"
xmlns:batch="http://www.springframework.org/schema/batch"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="..."&gt;
&lt;batch:job .../&gt;
...
&lt;/beans&gt;</programlisting>
<para>Or by including a bean definition explicitly for the <classname>JobScope</classname> (but not both):</para>
<programlisting language="xml">&lt;bean class="org.springframework.batch.core.scope.JobScope" /&gt;</programlisting>
</section>
</section>
</chapter>