Get hands-on experience with 20+ free Google Cloud products and $300 in free credit for new customers.

vertex search import from big query errors out with missing ID field

Hi - just trying to import a BQ table into vertex search as a data source, but it's erroring out with the following error:

BQ table (project_id: pink-stage, dataset_id: pinkdocstest, table_id: fiann) Missing field id with type STRING and is in REQUIRED mode

Screenshot 2024-01-09 at 9.09.39 AM.png

Where my BQ table does have an id field... 

Screenshot 2024-01-09 at 9.00.46 AM.png

 

any thoughts? 

3 3 1,394
3 REPLIES 3

Figured it out. 
Within a Data Store, you can't combine structured and unstructured data sources. I created a new datastore with structured data only and it works.

Well, it kinda works - the search is very basic - it looks like it only matches whole words in the value. So you can't do anything like "Accounts with Oil in the name"; instead you have to just type "Oil" and it will match if it finds that exact word in the value field. Strange how unstructured search is so awesome, but structured search is bare bones. 

Hi

For anyone seeing this try create the table and set the id to "REQUIRED" mode you can't do this from UI for some reason

So try re-create the table with NOT NULL on the id work for me

Reference https://stackoverflow.com/questions/54465074/revert-nullable-field-to-required for the method

 

The error message suggests that the vertex search is expecting the id field in the BigQuery table to be of type STRING and in REQUIRED mode. Perhaps your table schema shows that the id field is NULLABLE, not REQUIRED.