Big-Data-HPC-Platform/docs/ECLProgrammersGuide/PRG_Mods/PrG_Job_Failure.xml

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE sect1 PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
"http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd">
<sect1 id="Job_Failure">
  <title><emphasis role="bold">Job “Failure”</emphasis></title>

  <para>Sometimes jobs fail. And sometimes that behavior is by design.</para>

  <para>For example, attempting to send an entire output result back to the
  ECL IDE program when that result contains more than 10 megabytes of data
  will cause the job to "fail" with the error “Dataset too large to output to
  workunit (limit 10 megabytes).” This job “failure” is deliberate on the part
  of the system (and you can reset this particular limit on a per-workunit
  basis using #OPTION), because any time you are writing that amount of data
  out you should be writing it to a file to de-spray. Otherwise, you will
  rapidly fill your system data store.</para>

  <para>Other examples of this type of deliberate system "failure" is
  exceeding skew limits or exceeding any other runtime limit. For some of
  these limits there are ways to reset the limit itself (which is usually NOT
  the best solution). Otherwise, the deliberate “failure” is a signal that
  there is something inherently wrong with the job and perhaps the approach
  you are using needs to be re-thought.</para>

  <para>Contact Technical Support whenever such an issue arises and we will
  help you formulate a strategy to accomplish what you need to without
  incurring these deliberate system "failures."</para>
</sect1>