Scope

• Components for dependency handling
• Download & Deployment information

Download & Deployment

• Patch

  • This patch is available from the attached file com.sos.csi.patch.jar

  • To deploy this patch

    1. copy the jar file to the directory $SCHEDULER_HOME/patch (create this folder if is does not existi)
    2. open the file $SCHEDULER_DATA/config/factory.ini
    3. navigate to the section java
    4. change the class_path configuration to
       class_path = ${SCHEDULER_HOME}/patch/com.sos.csi.patch.jar:${SCHEDULER_HOME}/lib:....

    5. restart JobScheduler

• Dependency
  • Attached archive contains the below mentioned files for the dependency folder: dependency.zip
  • The archive should be extracted to a subfolder of the JobScheduler live directory, e.g. $SCHEDULER_DATA/config/live/csi.

Patch: com.sos.csi.patch.jar

• This patch contains the new option scheduler.ignore_sync_jobs_in_stopped_jobchains

• Enabling this option by setting the scheduler variable in the configuration file $SCHEDULER_DATA/scheduler.xml to true will ignore synchronization nodes that are part of stopped job chains for all instances of synchronization jobs.
  
  

  Configuration in file scheduler.xml

  <spooler>
      <config>
          <param name="scheduler.ignore_sync_jobs_in_stopped_jobchains" value="true"/>
  ...

•  It is also possible to enable this option per job by adding a job parameter like this:
  
  

  Configuration by job parameter

  <job  order="yes">
      <params >
          <param  name="scheduler.ignore_sync_jobs_in_stopped_jobchains" value="true"/>
      </params>
  ...

Folder: dependency

• This folder provides the components required to implement dependency handling.
• This folder should be located below the JobScheduler live directory.

Function Set: dependency.js

• This function set provides basic capabilities for dependency checking
• This function set is used by the following job Monitor Scripts:
  • jobnet_check_predecessor.js
  • jobnet_check_successor.js

Job: jobnet_check_predecessor.job.xml

Usage

• Sample job for a job step that provides dependency checking.
• This job is added to the beginning of a job chain.
• This job step is added the job node state jobnet_start.

Sample

Job Implementation

• The job implements an arbitrary job body.
• Dependency handling is implemented by the Monitor Script.
  • The Monitor Script includes two files:
    • jobnet_check_predecessor.js
    • dependency.js
  • When creating individual job implementations then no invidual implementation of the Monitor Script should be effected but the below-mentioned script files should be included as they are subject to future improvement.

<?xml version="1.0" encoding="ISO-8859-1"?>
<job  order="yes" stop_on_error="no" title="check jobnet predecessors">
    <script  language="javax.script:rhino">
        <![CDATA[
spooler_log.info("checking jobnet predecessors");
        ]]>
    </script>
    <monitor  ordering="0" name="monitor">
        <script  language="javax.script:rhino">
            <include  file="config/live/csi/dependency/jobnet_check_predecessor.js"/>
            <include  file="config/live/csi/dependency/dependency.js"/>
        </script>
    </monitor>
    <run_time />
</job>

• The Monitor Script makes use of the function set provided by the file dependency.js in order to handle dependencies for predecessor job chains.
• The Monitor Script checks predecessor job chains in order to decide if the current job chain should be executed or if its execution shoud be postponed..

function spooler_process_after( spooler_process_result ) {

    spooler_log.info( "spooler_process_after(): checking jobnet predecessor" );


    if ( spooler_process_result ) {
        // check if predecessor job chains exists and will start in period
        if ( isPredecessorJobChainInPeriod() ) {
            spooler_log.info( "spooler_process_after(): predecessor job chain will start later in period, order execution postponed, moving order to final state: " + getStateSkip() );
            var order = spooler_task.order();
            order.set_state( getStateSkip() );
        } else {
            spooler_log.info( "spooler_process_after(): no predecessor job chain found for consideration that will start later in period, order is continued" );
		}
    } else {
        spooler_log.info( "spooler_process_after(): no predecessor checking due to failed job processing" );
    }


    return spooler_process_result;
}

Job: jobnet_check_successor.job.xml

Usage

• Sample job for a job step that provides dependency checking.
• This job is added to the end of a job chain.
• This job step is added the job node state jobnet_end.

Sample

• see job jobnet_check_successor from the above screenshot

Job Implementation

• This job implements an arbitrary job body.
• Dependency handling is implemented by the Monitor Script.
  
  • The Monitor Script includes two files:
    • jobnet_check_successor.js
    • dependency.js
  • When creating individual job implementations then no invidual implementation of the Monitor Script should be effected but the below-mentioned script files should be included as they are subject to future improvement.

<?xml version="1.0" encoding="ISO-8859-1"?>
<job  order="yes" stop_on_error="no" title="check jobnet successors">
    <script  language="javax.script:rhino">
        <![CDATA[
spooler_log.info("checking jobnet successors");
        ]]>
    </script>
    <monitor  ordering="0" name="monitor">
        <script  language="javax.script:rhino">
            <include  file="config/live/csi/dependency/jobnet_check_successor.js"/>
            <include  file="config/live/csi/dependency/dependency.js"/>
        </script>
    </monitor>
    <run_time />
</job>

• The Monitor Script makes use of the function set provided by the file dependency.js in order to handle dependencies for successor job chains.
• The Monitor Scripts starts successor job chains if applicable.

function spooler_process_after( spooler_process_result ) {

    spooler_log.info( "spooler_process_after(): checking jobnet successor" );


    if ( spooler_process_result ) {
        // start orders for successor job chains
        var numOfOrders = startSuccessorJobChains();
        if ( numOfOrders > 0 ) {
            spooler_log.info( "spooler_process_after(): " + numOfOrders + " orders for successor job chains started" );
        } else {
            spooler_log.info( "spooler_process_after(): no successor job chain found for consideration, no orders started" );
        }
    } else {
        spooler_log.info( "spooler_process_after(): no successor checking due to failed job processing" );
    }


    return spooler_process_result;
}

Components for Job Chain Resets

• The following components are included with the folder dependency

Job Chain: jobnet_reset_job_chains.job_chain.xml

• The job chain makes use of only one job jobnet_reset_job_chains.job.xml, see below.

Job: jobnet_reset_job_chains.job.xml

Usage

• Standard job that is used to toggle the state of job chains for dependency checking. Job chains present one of these states:
  • running: the job chain is active and dependency checking is performed
  • stopped: the job chain is inactive, no dependency checking is performed.

Job Implementation

• This job implements the operations to toggle the state of one or multiple job chains.
• The effective processing is implemented by JavaScript.

<?xml version="1.0" encoding="ISO-8859-1"?>

<job  order="yes" stop_on_error="no" title="reset job chaines to unstopped or stopped state">
    <script  language="javax.script:rhino">
        <include  file="config/live/csi/dependency/jobnet_reset_job_chains.js"/>
        <include  file="config/live/csi/dependency/dependency.js"/>
    </script>
    <run_time />
</job>

• The JavaScript implementation:

function spooler_process( ) {
    spooler_log.info( "spooler_process(): reset job chains in jobnet" );
    
    var numOfModifiedJobChains = toggleJobChainState();
    
    spooler_log.info( "spooler_process(): " + numOfModifiedJobChains + " job chains modified" );
    return true;
}

Order: jobnet_reset_job_chains.order.xml

• This is a sample order that shows how job chains could be selected for activation and deactivation.
• Parameters
  • operation: the toggling operation
    • Values
      
      • activate: modify the state of stopped job chains to running
      • deactivate: modify the state of running job chains to stopped
    • Default Value
      • activate
  • input_list:
    • Values
      • A comma separated list of job chain paths
    • Default Value
      • none
  • input_file: 
    • Values
      • the location of a file that contains one job chain path per line
    • Default Value
      • none
• Behaviour
  • Without parameters being specified all job chains will be activated.
  • Without the parameters input_list and input_file being specified the operation is applied to all job chains.
  • Should in addition to the parameter input_list the  parameter an input_file be specified then both specifications of job chains are merged.

<?xml version="1.0" encoding="ISO-8859-1"?>

<order  title="reset job chaines to unstopped or stopped state">
    <params >
        <param  name="operation" value="deactivate"/>
        <param  name="input_list" value="csi/examples/JobChain_B/job_chain_B,csi/examples/JobChain_D/job_chain_D"/>
        <param  name="input_file" value="csi/jobnet_reset_job_chains.txt"/>
    </params>
    <run_time  let_run="no">
        <period  single_start="24:00"/>
    </run_time>
</order>

Limitations

• Dependency Hnandling Scripts and Monitor Scripts are implemented for the Rhino scripting engine up to Java 1.7. Adjustments might be required for the Nashorn engine that is introduced with Java 1.8.