Liquibase guidelines

Liquibase

https://www.liquibase.org/get-started/best-practices

Guidelines

• changelogs organized by release, db.changelog-master.xml includes the changelog for the releases in the correct order. db.changelog-master.xml is the changelog you pass to all Liquibase calls.

• Datatypes: https://dba-presents.com/index.php/liquibase/216-liquibase-3-6-x-data-types-mapping-table

• Naming NVARCHAR for SQL Server

  - property:
      name: varcharDataType
      value: NVARCHAR
      dbms: mssql
  - property:
      name: varcharDataType
      value: VARCHAR
      dbms: mysql
  - property:
      name: varcharDataType
      value: VARCHAR2"	
  

• CSV or SQL Files for data, we need folder structure convention (generated-sql, manual-sql, changelogs)

• Sometimes we need to write plain SQL and add to change log: example partitioning etc

    <changeSet author="backbase" id="2_20_0_001">
        <createTable tableName="test">
            <column name="id" type="INT">
                <constraints nullable="false"/>
            </column>
            <column name="NAME" type="VARCHAR(30)"/>
            <column name="borned" type="TIMESTAMP"/>
        </createTable>
        <modifySql dbms="postgresql">
            <append value=" PARTITION BY RANGE (UNIX_TIMESTAMP(borned)) (
    PARTITION p0 VALUES LESS THAN (UNIX_TIMESTAMP('2000-01-01 00:00:00')),
    PARTITION p1 VALUES LESS THAN (MAXVALUE))"/>
        </modifySql>
    </changeSet>
  

• ID and AUTHOR guidelines:

  • ID = version+sequence e.g: 2_20_0_001
  • Author: backbase

• Comments: must have, can be used for documentation

• Patching

• Sequence per table, naming conventions for the sequences explained in Confluence guidelines

• Context naming convention: for the creation changesets: initial for the upgrade ones: upgrade_X_Y_Z_to_X1_Y1_Z1 e.g: upgrade_1_0_0_to_1_1_0

• Labels: can be used for the same purpose as the contexts Contexts vs labels

• Packaging of sql scripts, follow the folder conventions and guidelines, e.g: no H2