Creating Data Ops Builds
Trigger a Build
A build will connect to your repository, detect all of the Glean configuration files in your repository and create, update, and delete resources to make glean consistent with the configuration:
- Navigate to the
DataOpspage using the link on the navigation side bar
- Click the
Buildbutton on the top right to create a new build
- Configure your build and click Build. If any fields are left empty, Glean will use the default values specified on the
Version Controlsettings page.
- You can view the summary of errors, warnings and changes by clicking the build from the build list.
We recommend always first creating a Preview Build to see which resources will be affected prior to running a Deploy. The build process will automatically delete and create resources in Glean to leave your project in a consistent state. For example, if you delete a column in a model configuration file, all saved views that refer to that column will be automatically deleted (as happens if you delete a column through the user interface).
The build process
Here is a high-level description of what actually happens during the build process:
- Create resources. For models that you check in that do not currently exist, the model is created from scratch from your configuration file and from the database. This means that the base table and all the columns must still exist in the database for the build to succeed.
- Update resources. If a column is changed, it will be updated in all dependent items (regardless of whether the dependent items were also checked in) - so things like name changes should be correctly reflected.
- Clean up resources. If any columns or models are deleted, all dependent items that relied on that column or model will also be deleted.
- Smoke Test. Once the model is built, a single aggregate analytical query will be run against the database to make sure that data aggregation can still be run against the model.
If there is an error while deploying a Data Model, the build process will revert to the previous version of that model, and all dependent resources of that model will not be updated.
Glean Configuration Files
Glean supports configuration files written in YAML or JSON. A Glean Build will attempt to validate every file in the specified path with a
Full documentation of all configuration files can be found for each resource type:
While this feature is in beta, it is possible we may need to introduce breaking changes to the Glean configuration file format. In these cases we will give advanced warning and assist beta partners with any necessary migrations.
Exporting configuration files
An easy way to get started building a configuration file is to export it from an existing Glean resource.
- Data Models:
- Navigate to the Data Models page
Editfor one of your Data Models
- Click the
⋮in the top-right corner
- Saved Views: On any Explore page, click the
⋮in the top-right corner of the chart area, and then click
Export Saved View Config File.
- Dashboards: On any Dashboard page, click the
⋮in the top-right corner, and then click
View DataOps Config File.
You will likely want to adjust some values in the configuration files after exporting to match your team's preferences and local environment, such as column names and paths to other resources.