Configure a Writeback table

A Writeback table receives rows from DataMind OS and writes them into a table of a destination. Configure it to set where its rows come from and which workflow runs it. Use this page for the General step, the Dependencies step and the Preview tab.

To open the editor and for its tabs, steps and saving mechanics, see Create and edit a Writeback table.

Set the workflow and the source data

  1. Open the table. In the Writeback tree, double-click it, or right-click it and choose Edit.
  2. On the Table tab, stay on General.
  3. Under Workflow, check the workflow. An imported table starts on the destination's (Standard) workflow. The list shows the Writeback workflows of this table's destination.
  4. To write a whole table, keep Table Selection on. Under Table Type, choose Lake or Warehouse, then pick the table under Table Name.
  5. To write the result of a query, turn Table Selection off. Write the query and click Validate.
  6. Click Apply. The editor saves the step, and then moves to Columns.

The Target section shows the name of the table in the destination. You cannot change it here.

SettingWhat it doesDefault
Table SelectionOn: the table takes a whole deployed Lake or Warehouse table. Off: the table takes the result of a query.On for a new table
Table TypeChooses which tables Table Name offers: Lake or Warehouse.Lake
Table NameThe deployed table to write. Options follow the Table Type and are sorted by name. Type in the list to search.Shows Select a table
WorkflowThe workflow that runs this table. Apply saves General only when one is chosen.The destination's (Standard) workflow, set at import

On a RabbitMQ destination, General also shows a Publish Window. See Publish to RabbitMQ.

To use a Lake or Warehouse table as a source, it must be deployed. See Configure a table.

Write and validate a query

When Table Selection is off, the editor tile shows these controls:

If the query is empty, Validate stops with Query cannot be empty. When validation fails, the reason from the database appears above the editor; a failure with no readable reason reads Query validation failed. A successful validation opens the Query Data Preview window with the query's result.

Validation runs the query for up to 30 minutes. A query that runs longer fails. SQL dialect and data scope covers the SQL you can use.

While the tile says Loading editor, the editor is still starting.

Check the dependencies

Dependencies is read-only. It lists the upstream tables this table reads, grouped by kind, with a count on each group. The list refreshes when you save General.

If the table reads nothing registered, the step shows No dependencies with the text "This write-back has no upstream tables registered as dependencies."

Preview what will be written

Open the Preview tab. It shows up to 1,000 rows of the source data, with the mapped columns renamed to the target column names. Use the toolbar to search, show or hide columns, page through the rows and export them.

If the preview cannot load, it shows Preview Error with the reason and a Retry button.

Save your changes

Saving, reverting and unsaved changes are covered in full on Save and revert your changes. If the step cannot be saved, a Validation Error toast tells you what is missing; check the Workflow, the table or the query. See Writeback troubleshooting for the message list.

Permissions

To do thisYou need
Open the editor and see the Preview tabView WriteBack Objects
Click Validate, Apply or FinishAdd/Modify WriteBack Objects

See Writeback permissions for the full list.

Warning

Changing the source rebuilds your column mapping. Each of these rebuilds it from matching column names: saving a different source table, switching between table and query, and changing the query so it no longer returns a column you mapped. Columns you mapped by hand are replaced, and a mapped column whose name is gone is dropped. If no column matches, the table is no longer deployed and does not run. Check Columns after every change to the source.

Limits and edge cases

Questions, answered

When should I use a query instead of a table?

Use a query when you need to filter, join or reshape data before it is written. Turn Table Selection off, write the query and click Validate.

Why is my Lake table not in the Table Name list?

Only deployed tables are offered. Deploy the Lake table first, using Configure a table.

Why does Apply refuse my General step?

Apply checks the step before it saves. Choose a Workflow, pick a table or, in query mode, click Validate after your last change.

Why did my column mapping reset?

The mapping is rebuilt from matching column names when you save a different source table, switch between table and query, or change the query so it no longer returns a column you mapped. Check Map columns after each change.

Why are some columns already mapped when I first open Columns?

Columns with the same name in the source and the target are matched automatically, ignoring case, when the configuration is saved. Columns with no match stay unmapped.

Can I edit the query in the full Editor?

Yes. Click Open in Editor. The query opens in the Editor, in a new console named after the table. Copy your final version back into the Writeback editor and click Validate.

How long can validation take?

Validation runs your query for up to 30 minutes. A query that runs longer fails, and the editor shows the error.

Does the preview show exactly what will be written?

The preview shows the source values with the mapped columns renamed to the target names, up to 1,000 rows. When the table runs, each value is converted to the data type of its target column, so a value can look different in the destination.

Can I move a table to another workflow?

Yes. On General, choose another Workflow and click Apply. To move several tables at once, see Transfer tables between workflows.

Can two Writeback tables write into the same destination table?

No. Each destination table has a single configuration, so it appears once in the tree and is configured once.