dbt Configurations

Learn more about how Datafold integrates dbt in CI

Prerequisites

Make sure Datafold GitHub integration is set up.

Tag primary keys in dbt model

Datafold needs to know which column is the primary key of the table to perform the diff. When Datafold cannot determine the primary key of the two tables to diff, and will produce an error:

When setting up the CI integration, one of the steps is proving the primary-key tag:

This tag we can use in the dbt metadata to let Datafold know which column can be used to perform the diff. Datafold supports composite primary keys, meaning that you can assign multiple columns that make up the primary key together. There are three ways of doing this, which we'll discuss next:

Metadata

The first one is setting the tag in the dbt metadata. We set the primary key tag to primary-keyso we use this in the metadata.

Table metadata can also be used to specify per-model diff options. In the example below diff is configured to compare only rows matching user_id > 2350. The expression in the filter is an SQL expression and can be anything you could put into where clause when selecting from the tables. Also you can include columns (include_columns) which will be used when diffing, and exclude (exclude_columns) some of them. For time dimension you can set a time column of `time_travel` field. If you want to add time travelling, you can define prod_time_travel for travelling of a production model, pr_time_travel of a PR model but, please, keep in mind that not all databases support time traveling.

models:
  - name: users
    meta:
      datafold:
        datadiff:
          filter: "user_id > 2350"
          include_columns:
            - user_id
            - created_at
            - name
          exclude_columns:
            - full_name
          prod_time_travel:
            - 2022-02-07T00:00:00
          pr_time_travel:
            - 2022-02-07T00:00:00
          time_column:
            - created_at
          
    columns:
      - name: user_id
        meta:
          primary-key: true
                    
          
# In the case of a compound Primary Key, you can assign a second column (or more)
      - name: version_id
        meta:
          primary-key: true

Tags

If the primary key is not found in the metadata, it will go through the tags.

models:
  - name: users
    columns:
      - name: user_id
        tags:
          - primary-key
# In the case of a compound Primary Key, you can assign a second column (or more)
      - name: version_id
        tags:
          - primary-key

Inferred

If the primary key isn't provided explicitly, Datafold will try to assume a pk from dbt's uniqueness tests. If you have a single column uniqueness test defined, it will use this column as the PK:

models:
  - name: users
    columns:
      - name: user_id
        tests:
          - unique

Also, model level uniqueness tests are used for inferring the PK:

models:
  - name: sales
    columns:
      - name: col1
      - name: col2
      ...
    tests:
      - unique:
          column_name: "col1 || col2"

  - name: sales
    columns:
      - name: col1
      - name: col2
      ...
    tests:
      - unique:
          column_name: "CONCAT(col1, col2)"

Finally, we also support unique_combination_of_columns from the dbt_utils package:

models:
  - name: users
    columns:
      - name: order_no
      - name: order_line
      ...
    tests:
      - dbt_utils.unique_combination_of_columns:
          combination_of_columns:
            - order_no
            - order_line

Keep in mind that this is a failover mechanism. If you change the uniqueness test, this will also impact the way Datafold performs the diff.

Checking primary key annotations

You can check what models in your dbt repo already have primary key annotations, and which need more attention. You'll need to install Datafold SDK and configure access parameters:

$ pip3 install 'datafold-sdk>=0.0.6'

# skip this step if you are using app.datafold.com
$ export DATAFOLD_HOST=https://<hostname>

# get your API key in Datafold UI -> Edit Profile -> API Key
$ export DATAFOLD_APIKEY=RSSQrpfddSEEK8WVtc0zd27f9nsdhPU3AxZ

After that you need to compile manifest.json and you are ready to do the check:

# Lookup your CI configuration id in URL when you go to Settings -> CI settings -> <name>:
# https://app.datafold.com/settings/ci_integrations/14

$ datafold dbt check-primary-keys --ci-config-id 14 manifest.json 
meta        dbt_snowflake.service_calls              INCIDENT_NUMBER            models/service_calls.sql              models/schema.yml 
meta        dbt_snowflake.supply_of_ones             ID                         models/supply_of_ones.sql             models/schema.yml 
none        dbt_snowflake.fokko.boom                                            models/fokko/boom.sql                                   
none        dbt_snowflake.new_service_calls                                     models/new_service_calls.sql          models/schema.yml 
tags        dbt_snowflake.ephemeral_supply_of_twos   ID                         models/ephemeral_supply_of_twos.sql   models/schema.yml 
uniqueness  dbt_snowflake.new_service_calls_concat2  CAL_YEAR, INCIDENT_NUMBER  models/new_service_calls_concat2.sql  models/schema.yml 
uniqueness  dbt_snowflake.supply_of_twos             ID                         models/supply_of_twos.sql             models/schema.yml 

The first column shows how the key was inferred:

• none - Datafold was unable to find any PKs,

• uniqueness - primary keys were derived from uniqueness tests,

• tags - PKs were specified with column-level tags,

• meta - column-level metadata was used,

• meta_table - table-level metadata.

Out of those, none and possibly uniqueness require further actions.

The other fields in the printout are:

• fully qualified name of dbt model,

• list of primary keys,

• sql file that contains model definition,

• "patch" yml file that has dbt configuration of the model.

dbt metadata synchronization

Datafold integrates very well with dbt, and also has the ability to ingest the metadata provided by dbt automatically. dbt models has metadata that can be synchronized from the production branch into the Datafold catalog. When a table has metadata being synchronized using dbt, user editing is no longer permitted for that entire table. This is to ensure that there is a single source of truth.

Metadata can be applied both on a table and column level:

models:
  - name: users
    description: "Description of the table"
    meta:
      owner: [e-mail]
      foo: bar
    tags:
      - pii
      - abc
    columns:
      - name: user_id
        tags:
          - pk
          - id
        meta:
          pk: true
      - name: email
        description: "The user's email"
        tags:
          - pii
        meta:
          type: email

There are two special meta types:

• owner: Used to specify the owner of the table and applies the owner of the table in the catalog view

• <pk_tag>: The tag/name that is configured to identify primary columns is not synchronized into the meta-information, but it is synchronized as a tag if it exists.

So for the above table:

• description is synchronized into the description field of the table in the catalog.

• The owner of the table is set to the user identified by the [e-mail] field. This user must exist in Datafold with that email.

• The foo meta information is added to the description field with the value bar

• The tags pii and bar are applied to the table as tags.

For the columns above:

• The column user_id has two tags applied: pk and id

• The metadata for user_id is ignored, because it reflects the primary key tag.

• The email column has the description applied.

• The email column has the tag pii applied

• The email column has extra metadata information in the description field: type with the value email.

Metadata synchronization occurs in one of two methods:

• The meta_schedule is set for the dbt cloud integration. This will run according to the specified cron schedule, find the most recent dbt cloud production run, and synchronize the metadata from there.

• It can also be configured to synchronize metadata whenever a push to production happens.