Why are my PDTs building on multiple connections?

[sam8]

Last tested: June 2021

A PDT will build on every model that it is included in. So, if you have several models with include "*.view", then each of those models will include all PDTs that you've defined.

A good solution is to adopt a naming convention for your PDT views. For example, you could make sure your BigQuery PDTs all begin with bq so that you can write:
include "bq*.view" in your model that includes a BigQuery connection.

Then, once you have your PDTs being included that way, you can adopt a similar naming convention for files you want to include in everything (like everywhere_* ) or include them individually.

Here are some other common reasons PDTs would build for multiple models:
 - The PDT is included in multiple models that have different connections.
- The PDT uses a datagroup that is defined differently in different models.
- The PDT is included in multiple models with different case sensitivity settings
- Differences in a dependency PDT between one model and another.

This content is subject to limited support.                

[mitchellposluns]

hey sam  hoping you might be able to assist here…

We have a PDT set on a SQL trigger that builds everyday at 9am. The view is included in multiple models that have different connections. We see that the PDT is built on both models in both the PDT pane and the system activity logs. However, we can see that when running a query with the PDT on an explore using the second connection to build, it returns stale data. Is it possible Looker might be getting confused, using the stale version of the PDT on the second connection instead of the new one?

This problem remedies itself when we run a manually rebuild derived tables from the explore link using the second connection...