Dry Runs
Published 16 November 2022
Dry Runs
When Flyway migrates a database, it looks for migrations that need to be applied, sorts them and applies them in order directly against the database.
This default behavior is great for the vast majority of the cases.
There are however situations where you may want to
- preview the changes Flyway will make to the database
- submit the SQL statements for review to a DBA before applying them
- use Flyway to determine what needs updating, yet use a different tool to apply the actual database changes
Flyway Teams Edition gives you a way to achieve all these scenarios using Dry Runs.
Usage
API/Maven/Gradle users will need to include flyway-proprietary
as a dependency in order to use Dry run.
See API (Java)
How it works
When doing a Dry Run, Flyway sets up a read-only connection to the database. It assesses what migrations need to run and generates a single SQL file containing all statements it would have executed in case of a regular migration run. This SQL file can then be reviewed. If satisfactory, Flyway can then be instructed to migrate the database and all changes will be applied. This SQL file also contains the necessary statements to create and update Flyway's schema history table, ensuring that all schema changes are tracked the usual way.
It is not advised to change a dry run script after it's been generated. Instead, any changes should be made to the migrations and a new dry run script generated. This is to ensure the changes executed match what's in your migrations.
It is not recommended to execute the dry run script directly without using Flyway, as there are some Flyway features which are not handled by the dry run script. These limitations are listed under Prepare, which is the command to use if you do need to do this.
Intercepted in Dry Run
These changes are intercepted and written into a file as explained above.
- SQL versioned migrations
- SQL repeatable migrations
- SQL callbacks
New in V9:
These are no longer executed during a dry run. Instead, their file names are logged in the dry run output.
Configuration
When using the Flyway command-line tool, Maven plugin or
Gradle plugin, a SQL file contained the output of the dry run can be configured using the
flyway.dryRunOutput
property. This can be on the local file
system, or in AWS S3 / Google Cloud Storage.
When using the API directly, the dry run output can be configured using a java.io.OutputStream
, giving you additional
flexibility.
As soon as this property is set, Flyway kicks in dry run mode. The database is no longer modified and all SQL statements that would have been applied are sent to the dry run output instead.
Tutorial
Click here to see a tutorial on using dry runs.