Pipelines

Validate a Pipeline

Validation checks your mappings for problems that would produce a broken destination schema or broken load code. You must validate with no errors before you can finalize.

In the app: Data Model › Pipelines › your Pipeline › Validate

[screenshot]
The Validation Results dialog showing grouped errors, a warning, and informational notes.

Run validation

  1. Open the Pipeline’s mapping editor.
  2. Click Validate.
  3. Read the Validation Results dialog, then click Close.

Finalize becomes available once a validation in this session passes with no errors. If you leave the editor and come back, click Validate again before you finalize.

Read the results

The top line says Validation Passed or Validation Failed. Below it:

  • Errors block finalizing. They are grouped so related fixes sit together: Stale mappings, Key requirements, Type compatibility, and Names & other.
  • Warnings do not block, but read them. They often predict a load that will fail later.
  • Notes confirm what went well, such as N columns mapped successfully and No naming collisions detected.

When there are errors, the dialog ends with Fix all errors before finalizing the pipeline.

What is checked

Errors

Check Example message
A source view has no key column View "CustomerSummary" has no primary key designated
Two included tables share a destination name Duplicate destination table name "Customer" (from sources: …)
A table or column has no destination name Table "cust" has no destination name
SCD Type 2 has no change detection method Table "Customer": SCD2 requires a change detection method (Hash, Watermark, or Batch)
A key-based strategy on a table with no key column Table "Order": load strategy IncrementalHash requires at least one key column
A destination type the source cannot convert to Table "Order", column "Amount": … with the reason
A text column with no length, on SQL Server, Fabric Warehouse, MySQL, or Redshift Column "Customer.Email" has no length for NVARCHAR — …
A mapping whose source no longer exists Table "Legacy": … — run Sync from Source → Resolve stale mappings.

A table with no strategy is checked as if it used Truncate & Reload.

Warnings

  • A table with no key column whose strategy does not need one (… has no primary key selected).
  • Incremental Load or Incremental Batch without its column configured (… — procedure generation will fail).
  • A narrowing type change that may lose data.
  • A source schema that is also the destination (Source and destination schemas overlap — TruncateAndReload will destroy source data).
  • AI name suggestions you have not accepted or rejected (N AI suggestions have not been reviewed).

Notes

  • A hash-based table without a source hash column: a view will be created to compute the comparison hash. This is expected.

Fix and re-validate

  1. Close the dialog.
  2. Fix the issue: mark a key column in the mapping editor, rename a duplicate table, configure the strategy on the destination schema’s Pipeline tab, or resolve stale mappings.
  3. Click Validate again.

For fixes by message, see Troubleshooting: my Pipeline will not validate.