zoukankan      html  css  js  c++  java
  • 数据库迁移框架Flyway介绍

    官方文档

    https://flywaydb.org/getstarted/firststeps/api[https://flywaydb.org/getstarted/firststeps/api]

    入门示例

    Java代码

    package foobar;
    

    import org.flywaydb.core.Flyway;

    public class App {
    public static void main(String[] args) {
    Flyway flyway = new Flyway();
    // 指定数据源
    flyway.setDataSource("jdbc:mysql://localhost/test", "root", "root");
    // 开始数据迁移
    flyway.migrate();
    }
    }

    在classpath下添加SQL文件 db/migration/V1__Create_person_table.sql

    create table PERSON (
        ID int not null,
        NAME varchar(100) not null
    );
    

    运行程序,在test数据库会自动创建PERSON表。

    后续新增表和字段,只需在db/migration目录下新增SQL文件,格式为V${version}__${name}.sql,version值依次增加,比如V2__name.sql,V3__name.sql。

    原理介绍

    Flyway的数据库迁移的实现原理是,从classpath或文件系统中找到符合规则的数据库迁移脚本,比如db/migration目录下命名规则为V${version}__${name}.sql的文件,将脚本按照version进行排序,依次执行。执行过的脚本会作为一条记录,存储在schema_version表中。当下次执行迁移时,判断脚本已经执行,则跳过。

    MigrationResolver接口负责查找数据库迁移脚本,方法为resolveMigrations(),数据库迁移脚本用ResolvedMigration对象表示。MigrationResolver包含多种实现类,比如SqlMigrationResolver会从classpath下查找sql文件。查询通过Scanner类实现,Location类指定查询路径,sql文件的命名规则需要符合V${version}__${name}.sql,规则中的前缀V、后缀.sql、分隔符__均在FlywayConfiguration接口中定义。另一种实现类JdbcMigrationResolver会从classpath下查找实现JdbcMigration接口的类,类的命名规则需要符合V${version}__${name}。需要扩展自己的实现类,可以继承BaseMigrationResolver。

    MetaDataTable接口负责查找已执行的数据库迁移脚本,方法为findAppliedMigrations(),已执行的数据库迁移脚本用AppliedMigration对象表示。MetaDataTable只有一种实现类MetaDataTableImpl,从数据库schema_version表查询所有记录。

    ResolvedMigration集合包含了已经执行的AppliedMigration集合,在执行ResolvedMigration前,需要对比AppliedMigration,找到已执行和未执行的ResolvedMigration,对比通过MigrationInfoServiceImpl.refresh()实现。已执行的ResolvedMigration需要校验文件有没有发生变化,有变更则提示错误。未执行的ResolvedMigration依次执行,执行结果记录在schema_version表中。

    Flyway的主要方法:

    public class Flyway {
       /**数据库迁移*/
       public int migrate();
       /**校验已执行的迁移操作的变更情况*/
       public void validate();
       /**清理数据库*/
       public void clean();
       /**设置数据库的基准版本*/
       public void baseline();
       /**删除执行错误的迁移记录*/
       public void repair();
       /**准备执行环境,并执行Command操作,以上方法都调用了execute()来执行操作*/
       <T> T execute(Command<T> command);
    }
    

    接下来我们分析Flyway.migrate()代码执行的主逻辑。

    public void migrate() {
      //由Flyway.execute()准备Command.execute()执行所需要的参数
      return execute(new Command<Integer>() {
        public Integer execute(Connection connectionMetaDataTable,
          MigrationResolver migrationResolver,  MetaDataTable metaDataTable, 
          DbSupport dbSupport, Schema[] schemas, FlywayCallback[] flywayCallbacks) {
          //为了简化代码,忽略参数传递
          doMigrate();
    
    }
    

    });
    }

    private void doMigrate() {
    //校验已执行的迁移操作的变更情况
    if (validateOnMigrate) {
    doValidate(connectionMetaDataTable, dbSupport, migrationResolver,
    metaDataTable, schemas, flywayCallbacks, true);
    }

    //如果尚未进行数据迁移,即schema_version表中不存在数据,
    //并且数据库不为空,则插入一条baseline信息
    if (!metaDataTable.exists()) {
    //数据库不为空
    if (!nonEmptySchemas.isEmpty()) {
    //插入一条baseline信息
    new DbBaseline(connectionMetaDataTable, dbSupport, metaDataTable,
    schemas[0], baselineVersion, baselineDescription, flywayCallbacks).baseline();
    }
    }

    //进行数据迁移
    DbMigrate dbMigrate = new DbMigrate(connectionUserObjects, dbSupport,
    metaDataTable,schemas[0], migrationResolver, ignoreFailedFutureMigration,
    Flyway.this);
    return dbMigrate.migrate();
    }

    接下来看DbMigrate.migrate()的代码片段。

    MigrationInfoServiceImpl对比ResolvedMigration和AppliedMigration对象,找出需要执行数据库迁移脚本,通过pending()方法返回。最后执行数据库迁移脚本。

    public int migrate() {
      int migrationSuccessCount = 0;
      while (true) {
        int count = metaDataTable.lock(new Callable<Integer>() {
    
      <span class="hljs-meta">@Override</span>
      <span class="hljs-function"><span class="hljs-keyword">public</span> Integer <span class="hljs-title">call</span><span class="hljs-params">()</span> </span>{
        <span class="hljs-comment">//为了简化代码,忽略参数传递</span>
        <span class="hljs-keyword">return</span> doMigrate();
      }
    }
    <span class="hljs-keyword">if</span> (count == <span class="hljs-number">0</span>) {
      <span class="hljs-comment">// No further migrations available</span>
      <span class="hljs-keyword">break</span>;
    }
    migrationSuccessCount += count;
    

    }
    return migrationSuccessCount;
    }

    private int doMigrate() {
    //收集已经入库的数据库迁移记录,和以文件形式存在的数据库迁移脚本
    MigrationInfoServiceImpl infoService = new MigrationInfoServiceImpl(
    migrationResolver, metaDataTable, configuration.getTarget(),
    configuration.isOutOfOrder(), true, true, true);
    infoService.refresh();

    //infoService.pending()记录将要执行的数据库迁移脚本
    LinkedHashMap<MigrationInfoImpl, Boolean> group =
    groupMigrationInfo(infoService.pending());
    if (!group.isEmpty()) {
    //执行数据库迁移操作
    applyMigrations(group);
    }
    }

    DbMigrate.doMigrateGroup() 执行数据库迁移脚本

    private void doMigrateGroup() {
      //执行迁移脚本
      migration.getResolvedMigration().getExecutor().execute(connectionUserObjects);
    

    //存入数据库
    AppliedMigration appliedMigration = new AppliedMigration(migration.getVersion(),
    migration.getDescription(), migration.getType(), migration.getScript(),
    migration.getResolvedMigration().getChecksum(), executionTime, true);
    metaDataTable.addAppliedMigration(appliedMigration);
    }

    扩展练习

    在一个已经上线的游戏项目中引入Flyway框架,对于已经存在的游戏服,只执行新增的sql语句,对于新搭建的游戏服,需要创建数据库,并执行新增的sql语句。

    在这个需求中,需要做的是对于新搭建的游戏服,需要创建数据库。我们可以通过FlywayCallback实现这一点,如果指定数据库为空,则执行初始化数据库的语句。

    代码如下:

    public static void main(String[] args) {
        final Flyway flyway = new Flyway();
        flyway.setDataSource("jdbc:mysql://localhost/test", "root", "root");
    
    FlywayCallback flywayCallback = <span class="hljs-keyword">new</span> BaseFlywayCallback() {
        
        <span class="hljs-meta">@Override</span>
        <span class="hljs-function"><span class="hljs-keyword">public</span> <span class="hljs-keyword">void</span> <span class="hljs-title">beforeMigrate</span><span class="hljs-params">(Connection connection)</span> </span>{
            DbSupport dbSupport = DbSupportFactory.createDbSupport(connection, <span class="hljs-keyword">false</span>);
            <span class="hljs-keyword">if</span>(!hasTable(dbSupport)) {
                initDb(dbSupport);
            }
        }
    
        <span class="hljs-function"><span class="hljs-keyword">private</span> <span class="hljs-keyword">boolean</span> <span class="hljs-title">hasTable</span><span class="hljs-params">(DbSupport dbSupport)</span> </span>{
            Schema&lt;?&gt; schema = dbSupport.getOriginalSchema();
            Table[] tables = schema.allTables();
            <span class="hljs-keyword">if</span>(tables == <span class="hljs-keyword">null</span> || tables.length == <span class="hljs-number">0</span>) {
                <span class="hljs-keyword">return</span> <span class="hljs-keyword">false</span>;
            }
            <span class="hljs-comment">//忽略表 schema_version</span>
            <span class="hljs-keyword">if</span>(tables.length == <span class="hljs-number">1</span> &amp;&amp; 
                tables[<span class="hljs-number">0</span>].getName().equalsIgnoreCase(<span class="hljs-string">"schema_version"</span>)) {
                <span class="hljs-keyword">return</span> <span class="hljs-keyword">false</span>;
            }
            <span class="hljs-keyword">return</span> <span class="hljs-keyword">true</span>;
        }
        
    
        <span class="hljs-function"><span class="hljs-keyword">private</span>  <span class="hljs-keyword">void</span> <span class="hljs-title">initDb</span><span class="hljs-params">(DbSupport dbSupport)</span> </span>{
            Scanner scanner = <span class="hljs-keyword">new</span> Scanner(
              Thread.currentThread().getContextClassLoader());
            Resource[] resources = scanner.scanForResources(
              <span class="hljs-keyword">new</span> Location(<span class="hljs-string">"db/init"</span>), <span class="hljs-string">""</span>, <span class="hljs-string">".sql"</span>);
            <span class="hljs-keyword">if</span>(resources == <span class="hljs-keyword">null</span> || resources.length == <span class="hljs-number">0</span>) {
                <span class="hljs-keyword">throw</span> <span class="hljs-keyword">new</span> RuntimeException(
                  <span class="hljs-string">"db/init/*.sql not found in the classpath. "</span>);
            }
            <span class="hljs-keyword">for</span>(Resource resource : resources) {
              SqlMigrationExecutor executor = <span class="hljs-keyword">new</span> SqlMigrationExecutor(
                  dbSupport, resource, PlaceholderReplacer.NO_PLACEHOLDERS, flyway);
              executor.execute(dbSupport.getJdbcTemplate().getConnection());
            }
        }
    };
    
    List&lt;FlywayCallback&gt; callbacks = <span class="hljs-keyword">new</span> ArrayList&lt;FlywayCallback&gt;(
      Arrays.asList(flyway.getCallbacks()));
    callbacks.add(flywayCallback);
    flyway.setCallbacks(callbacks.toArray(<span class="hljs-keyword">new</span> FlywayCallback[callbacks.size()]));
    
    flyway.setBaselineOnMigrate(<span class="hljs-keyword">true</span>);
    flyway.repair();
    flyway.migrate();
    

    }

          </div>
  • 相关阅读:
    利用 Makefile 写的小程序
    linux内核学习之进程管理------task_struct结构体
    智能指针
    explicit 关键字
    ant脚本编写
    FROM_UNIXTIME 格式化MYSQL时间戳函数
    Dubbo架构设计详解-转
    Dubbo与Zookeeper、SpringMVC整合和使用(负载均衡、容错)转
    SiteMesh详解
    sitemesh使用步骤
  • 原文地址:https://www.cnblogs.com/jpfss/p/9967700.html
Copyright © 2011-2022 走看看