数据库迁移Flyway

为什么需要Flyway

日常开发常常会遇到一些这样的场景

1. 小红开发一个模块在本地数据库增加了两个字段，并且改动了dao层的代码提交到git。这时候小黄拉取了代码Run很可能报错。
2. 如果在上线正式环境的时候，忘记在正式数据库执行sql脚本可能造成严重的问题。
3. 传统的解决方式是在一个固定的地方添加sql脚本，开发人员相互沟通执行哪个sql脚本

Flyway可以将这一类问题解决，在项目编译期就将改动写入数据库。只要启动成功就没有问题。

---

Flyway 导入

如果是Gradle，在build.gradle添加依赖

compile "org.flywaydb:flyway-core:4.1.2"

Maven 在pom.xml添加依赖

　　org.flywaydb
　　flyway-core
　　4.0.3

---

实际使用

使用Spring Boot，Gradle 构建项目，添加依赖。

如果是已经开发一段时间的项目需要开启 baselineOnMigrate 否则抛出异常

Found non-empty schema(s) `reshelf2` without metadata table! Use baseline() or set baselineOnMigrate to true to initialize the metadata table. 

在 application.properties 添加配置

flyway.baselineOnMigrate=true

当然也可以实现接口

@Component
public class BaselineOnMigrateMigrationStrategy implements FlywayMigrationStrategy {
    @Override
    public void migrate(Flyway flyway) {
        flyway.setBaselineOnMigrate(true);
        flyway.migrate();
    }
}

数据库中的schema_version为存储对比脚本版本的表

sql脚本默认放置在 classpath:db/migration

文件以.sql结尾，命名V字开头，后面数字为版本号 例如 V1__init.sql

摘自 org.flywaydb.core.Flyway

 /**
     * The locations to scan recursively for migrations.
     * 

     * 
The location type is determined by its prefix.
     * Unprefixed locations or locations starting with {@code classpath:} point to a package on the classpath and may
     * contain both sql and java-based migrations.
     * Locations starting with {@code filesystem:} point to a directory on the filesystem and may only contain sql
     * migrations.
     * 

     * (default: db/migration)
     */
    private Locations locations = new Locations("db/migration");

    /**
     * The encoding of Sql migrations. (default: UTF-8)
     */
    private String encoding = "UTF-8";

    /**
     * The schemas managed by Flyway.  These schema names are case-sensitive. (default: The default schema for the datasource connection)
     * 
Consequences:
     * 

     * 
The first schema in the list will be automatically set as the default one during the migration.
     * 
The first schema in the list will also be the one containing the metadata table.
     * 
The schemas will be cleaned in the order of this list.
     * 
     */
    private String[] schemaNames = new String[0];

    /**
     * 
The name of the schema metadata table that will be used by Flyway. (default: schema_version)
 By default
     * (single-schema mode) the metadata table is placed in the default schema for the connection provided by the
     * datasource. 
 
 When the flyway.schemas property is set (multi-schema mode), the metadata table is
     * placed in the first schema of the list. 
     */
    private String table = "schema_version";

    /**
     * The target version up to which Flyway should consider migrations. Migrations with a higher version number will
     * be ignored. The special value {@code current} designates the current version of the schema (default: the latest version)
     */
    private MigrationVersion target = MigrationVersion.LATEST;

    /**
     * Whether placeholders should be replaced. (default: true)
     */
    private boolean placeholderReplacement = true;

    /**
     * The map of <placeholder, replacementValue> to apply to sql migration scripts.
     */
    private Map placeholders = new HashMap();

    /**
     * The prefix of every placeholder. (default: ${ )
     */
    private String placeholderPrefix = "${";

    /**
     * The suffix of every placeholder. (default: } )
     */
    private String placeholderSuffix = "}";

    /**
     * The file name prefix for sql migrations. (default: V)
     * 

     * 
Sql migrations have the following file name structure: prefixVERSIONseparatorDESCRIPTIONsuffix ,
     * which using the defaults translates to V1_1__My_description.sql
     */
    private String sqlMigrationPrefix = "V";

    /**
     * The file name prefix for repeatable sql migrations. (default: R)
     * 

     * 
Repeatable sql migrations have the following file name structure: prefixSeparatorDESCRIPTIONsuffix ,
     * which using the defaults translates to R__My_description.sql
     */
    private String repeatableSqlMigrationPrefix = "R";

    /**
     * The file name separator for sql migrations. (default: __)
     * 

     * 
Sql migrations have the following file name structure: prefixVERSIONseparatorDESCRIPTIONsuffix ,
     * which using the defaults translates to V1_1__My_description.sql
     */
    private String sqlMigrationSeparator = "__";

    /**
     * The file name suffix for sql migrations. (default: .sql)
     * 

     * 
Sql migrations have the following file name structure: prefixVERSIONseparatorDESCRIPTIONsuffix ,
     * which using the defaults translates to V1_1__My_description.sql
     */
    private String sqlMigrationSuffix = ".sql";

可以自定义配置，但是建议不要。

注意：sql脚本需要有相应的版本号，例如如果想让 V2__init.sql执行 需要有V1__init.sql作为一个基准对比，然后flyway才会执行相应的sql脚本。

---

一些链接

https://flywaydb.org/documentation/gradle/

http://stackoverflow.com/questions/33029311/setting-flyway-baselineonmigrate-and-baselineversion-using-spring-boot-prope

http://stackoverflow.com/questions/30013953/how-to-use-jdbc-authentication-of-spring-boot-spring-security-with-flyway

https://github.com/spark-jobserver/spark-jobserver/issues/503

---