DataStage parallel job fails with: Internal Error: (index < m_outputLinkObjArraySize)

Resolving The Problem

If the job contains column names using punctuation characters such as '.' or other special characters in a parallel job, you will need to change those column names.

Some special characters are reserved for other uses. For example, the period character '.' is used as a separator character between subrecord components: "asub.b" refers to the column "b" within subrecord "asub".

A parallel engine column name consists of one or more alphabetic, digits (0-9), or underscore characters. Alphabetic characters are not restricted to Roman characters; Unicode (for example, Kanji) or accented characters are also allowed. Also, the name cannot start with a digit.

If the failing job does not have columns names with special characters, then the job may contain an orphan stage -- a stage that is not connected to any other stage in the job via input or output links. Often this stage is on the outer edge of Designer palette and not visible unless fully zoom out, so the state may have been overlooked. Some older releases may have compiled the job without error and ignored that stage at run time, so this could be a job which worked in the past but began to fail after either information server updates, or after migrating the job stages with the Connector Migration Tool.

To resolve this issue, use the zoom out icon on Designer toolbar to show the full job palette and confirm there are no stage icons without links. If you find one, confirm whether the stage should actually be connected to rest of job and then either delete the stage if not needed, or update job design to correctly link that stage, then recompile.

In the case of error when using the new Unstructured Data stage, review the input data to determine if the first row contains special characters.