Jobs are the execution unit in Synthetic Data. A job defines how synthetic data is generated and where it is delivered by combining a dataset with connector mappings and data volume settings.
Each job is associated with a single application and dataset. When a job runs, it uses the dataset's schema and generation rules to produce synthetic data and writes the output to the configured target connectors.
Jobs can be run on demand using the Run Now action. After each run, the job records its last run status (for example, Success or Failed) so you can monitor data generation activity at a glance.
Dataset jobs are also listed in the Jobs view on the dataset details page. If a data-generation execution is interrupted by a syngen-compute restart, the job is moved out of the running state and the failure details are recorded.
When a field-generator assignment is invalid, fields with valid assignments can still produce partial output. The field with the invalid assignment is reported with its own error.
Reference-generator validation handles child tables with multiple foreign keys to different parent tables without incorrectly reporting a referenced table as unavailable.
Composite-key validation
If a composite foreign-key constraint is missing a generator assignment, the job fails before generation and identifies the constraint, table, and missing column. If a composite junction table does not have enough available key combinations for the requested row count, reduce the requested row count or increase the parent-table row counts before retrying.
Before you can create a job, the following must already exist:
-
A dataset must be configured within the application, and
-
At least one reference connector and one target connector must be available to map as the data source and destination.