SpringBoot+AOP实现多数据源动态切换

• 背景
• 设计总体思路
• 步骤

背景

系统后端需要访问多个数据库,现有的数据库连接配置写入配置文件中。后端需要从一个数据库的配置表里动态的读取其它mysql数据库的链接配置信息，并根据链接信息动态创建数据库链接，发起请求，而且还要能使用现在的一些连接池

设计总体思路

SpringBoot+AOP方式实现多数据源切换，继承AbstractRoutingDataSource实现数据源动态的获取，在service层使用注解指定数据源

步骤

对数据源库表进行设计，相关SQL语句如下所示：CREATE TABLE "YUDB"."DOM_DATABASE" ("DB_ID" NUMBER NOT NULL ENABLE,"SRC_ID" NUMBER,"DB_NAME" VARCHAR2 ( 30 ),"DB_C_NAME" VARCHAR2 ( 60 ),"NOTE" VARCHAR2 ( 60 ),"TAB_NUM" NUMBER,"DB_SIZE" NUMBER,"UD_TIME" DATE NOT NULL ENABLE,"MD_FILE_ID" VARCHAR2 ( 32 ),CONSTRAINT "PK_DATABASE" PRIMARY KEY ( "DB_ID" ) USING INDEX PCTFREE 10 INITRANS 2 MAXTRANS 255 COMPUTE STATISTICS STORAGE ( INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645 PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1 BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT ) TABLESPACE "USERS" ENABLE,CONSTRAINT "SYS_C0011295" CHECK ( "DB_ID" IS NOT NULL ) ENABLE,CONSTRAINT "SYS_C0011296" CHECK ( "UD_TIME" IS NOT NULL ) ENABLE,CONSTRAINT "FK_DATABASE" FOREIGN KEY ( "SRC_ID" ) REFERENCES "YUDB"."DOM_DATASOURCE" ( "SRC_ID" ) ON DELETE CASCADE ENABLE ) SEGMENT CREATION IMMEDIATE PCTFREE 10 PCTUSED 40 INITRANS 1 MAXTRANS 255 NOCOMPRESS LOGGING STORAGE ( INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645 PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1 BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT ) TABLESPACE "YU" CREATE TABLE "YUDB"."DOM_DATASOURCE" ("SRC_ID" NUMBER NOT NULL ENABLE,"CHG_ID" NUMBER,"SRC_SID" VARCHAR2 ( 20 ),"TABLE_SPACE" VARCHAR2 ( 20 ),"IP_ADDR" VARCHAR2 ( 30 ),"SRC_PORT" VARCHAR2 ( 10 ),"SRC_USER" VARCHAR2 ( 20 ),"SRC_PSW" VARCHAR2 ( 20 ),"ORGID" NUMBER,"DBTYPE" VARCHAR2 ( 100 ),CONSTRAINT "PK_DATASOURCE" PRIMARY KEY ( "SRC_ID" ) USING INDEX PCTFREE 10 INITRANS 2 MAXTRANS 255 COMPUTE STATISTICS STORAGE ( INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645 PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1 BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT ) TABLESPACE "USERS" ENABLE, CONSTRAINT "SYS_C0011298" CHECK ( "SRC_ID" IS NOT NULL ) ENABLE ) SEGMENT CREATION IMMEDIATE PCTFREE 10 PCTUSED 40 INITRANS 1 MAXTRANS 255 NOCOMPRESS LOGGING STORAGE ( INITIAL 65536 NEXT 1048576 MINEXTENTS 1 MAXEXTENTS 2147483645 PCTINCREASE 0 FREELISTS 1 FREELIST GROUPS 1 BUFFER_POOL DEFAULT FLASH_CACHE DEFAULT CELL_FLASH_CACHE DEFAULT ) TABLESPACE "YU" DataBase库表设计

DataSource库表设计：

ER图如下图所示：

往yu数据库中的表dataSource里增加数据库的相关配置信息。

在配置文件application-dev.yml中，进行多数据源配置spring:devtools:restart:enabled: true #设置开启热部署additional-paths: src/main/java #重启目录exclude: WEB-INF/**datasource:main:username: YUDBpassword: YUDB_HHurl: jdbc:oracle:thin:@xxx.xxx.xxx.xxx:1521/RAC1driver-class-name: oracle.jdbc.driver.OracleDriverzyml:username: xxxxxxpassword: xxxxxxurl: jdbc:oracle:thin:@xxx.xxx.xxx.xxx:1521/HHUdriver-class-name: oracle.jdbc.driver.OracleDriverbjobj:username: xxxxxxpassword: xxxxxxurl: jdbc:oracle:thin:@xxx.xxx.xxx.xxx:1521/BJOBJdriver-class-name: oracle.jdbc.driver.OracleDriver#####DruidDataSource配置#####################type: com.alibaba.druid.pool.DruidDataSourceinitialSize: 5minIdle: 5maxActive: 20# 配置获取连接等待超时的时间maxWait: 60000# 配置间隔多久才进行一次检测，检测需要关闭的空闲连接，单位是毫秒timeBetweenEvictionRunsMillis: 60000# 配置一个连接在池中最小生存的时间，单位是毫秒minEvictableIdleTimeMillis: 300000validationQuery: SELECT 1 FROM DUALtestWhileIdle: truetestOnBorrow: falsetestOnReturn: false# 打开PSCache，并且指定每个连接上PSCache的大小poolPreparedStatements: truemaxPoolPreparedStatementPerConnectionSize: 20# 配置监控统计拦截的filters，去掉后监控界面sql无法统计，'wall'用于防火墙filters: stat,wall,log4j# 通过connectProperties属性来打开mergeSql功能；慢SQL记录connectionProperties: druid.stat.mergeSql=true;druid.stat.slowSqlMillis=5000#合并多个DruidDataSource的监控数据useGlobalDataSourceStat: true###############以上为配置druid添加的配置########################################

创建实体类DataSource和DataBase
DataBase表和DataSource表通过外键src_id链接，先创建DataBase实体类package org.hhu.yu.system.rdbms.entity;import lombok.Data;import javax.persistence.*; import java.util.Date;@Data public class DataBase {private Long db_id;@JoinColumn(name="SRC_ID")@ManyToOne(cascade = CascadeType.ALL)//数据库名private String db_name;//数据库中文名private String db_c_nname;//数据库描述private String db_desc;//表数量private Long table_num;//数据库大小private Long data_size;//数据库更新时间private Date update_time;private String metaId; } 然后创建DataSource实体类import io.swagger.annotations.ApiModel; import io.swagger.annotations.ApiModelProperty; import lombok.*;@ApiModel(description = "数据源对象Model") @Data @AllArgsConstructor @NoArgsConstructor @RequiredArgsConstructor public class DataSource{@NonNull@ApiModelProperty(value = "数据源id",name ="src_id" )private Long src_id;@NonNull@ApiModelProperty(value= "数据源实例名",name="src_sid",required=true)private String src_sid;//数据源表空间private String table_space;@ApiModelProperty(value= "数据源ip地址",name="ip_addr",required=true)private String ip_addr;@ApiModelProperty(value= "数据源端口",name="src_port",required=true)private String src_port;@ApiModelProperty(value= "数据源用户名",name="src_user",required=true)private String src_user;@ApiModelProperty(value= "数据源密码",name="src_pwd",required=true)private String src_pwd;@ApiModelProperty(value= "数据库类型",name="db_type",required=true)private String db_type;@NonNullprivate DataBase dataBase; }

创建DruidDBConfig.java，配置Druid数据库连接池

@Configuration public class DruidDBConfig {@Bean@Qualifier("mainDataSource")@ConfigurationProperties(prefix = "spring.datasource.main")DataSource mainConfig() throws SQLException{DruidDataSource build = DruidDataSourceBuilder.create().build();List<Filter> filters = new ArrayList<>();filters.add(statFilter());filters.add(logFilter());build.setProxyFilters(filters);return build;}@Bean@Qualifier("zymlDataSource")@ConfigurationProperties(prefix = "spring.datasource.zyml")DataSource zymlConfig(){return DruidDataSourceBuilder.create().build();}@Bean@Qualifier("bjobjDataSource")@ConfigurationProperties(prefix = "spring.datasource.bjobj")DataSource bjobjConfig(){return DruidDataSourceBuilder.create().build();}@Bean(name = "dynamicDataSource")@Primary //优先使用，多数据源@Qualifier("dynamicDataSource")public DynamicDataSource dynamicDataSource() throws SQLException {DynamicDataSource dynamicDataSource = new DynamicDataSource();dynamicDataSource.setDebug(false);// 默认数据源配置 DefaultTargetDataSourcedynamicDataSource.setDefaultTargetDataSource(mainConfig());Map<Object, Object> targetDataSources = new HashMap<Object, Object>();//多数据源配置 TargetDataSourcestargetDataSources.put("mainDataSource", mainConfig());targetDataSources.put("zymlDataSource", zymlConfig());targetDataSources.put("bjobjDataSource", mainConfig());dynamicDataSource.setTargetDataSources(targetDataSources);return dynamicDataSource;} }

创建DBContextHolder类，实现数据源的切换服务

@Slf4j public class DBContextHolder {// 对当前线程的操作-线程安全的private static final ThreadLocal<String> contextHolder = new ThreadLocal<String>();// 调用此方法，切换数据源public static void setDataSource(String dataSource) {contextHolder.set(dataSource);log.info("已切换到数据源:{}",dataSource);}// 获取数据源public static String getDataSource() {return contextHolder.get();}// 删除数据源public static void clearDataSource() {contextHolder.remove();log.info("已切换到主数据源");} }

创建核心的动态数据源配置类，该类继承AbstractRoutingDataSource实现数据源动态的获取。

@Slf4j public class DynamicDataSource extends AbstractRoutingDataSource {private boolean debug = true;private Map<Object, Object> dynamicTargetDataSources;private Object dynamicDefaultTargetDataSource;@Overrideprotected Object determineCurrentLookupKey() {String datasource = DBContextHolder.getDataSource();if (!StringUtils.isEmpty(datasource)) {Map<Object, Object> dynamicTargetDataSources2 = this.dynamicTargetDataSources;if (dynamicTargetDataSources2.containsKey(datasource)) {log.info("---当前数据源：" + datasource + "---");} else {log.info("不存在的数据源：");throw new ADIException("不存在的数据源："+datasource,500);return null; }} else {log.info("---当前数据源：默认数据源---");}return datasource;}@Overridepublic void setTargetDataSources(Map<Object, Object> targetDataSources) {super.setTargetDataSources(targetDataSources);this.dynamicTargetDataSources = targetDataSources;}@Overridepublic void setDefaultTargetDataSource(Object defaultTargetDataSource) {super.setDefaultTargetDataSource(defaultTargetDataSource);this.dynamicDefaultTargetDataSource = defaultTargetDataSource;}/*** @param debug* @description the debug to set*/public void setDebug(boolean debug) {this.debug = debug;}/*** @return the debug*/public boolean isDebug() {return debug;}/*** @return the dynamicTargetDataSources*/public Map<Object, Object> getDynamicTargetDataSources() {return dynamicTargetDataSources;}/*** @param dynamicTargetDataSources* the dynamicTargetDataSources to set*/public void setDynamicTargetDataSources(Map<Object, Object> dynamicTargetDataSources) {this.dynamicTargetDataSources = dynamicTargetDataSources;}/*** @return the dynamicDefaultTargetDataSource*/public Object getDynamicDefaultTargetDataSource() {return dynamicDefaultTargetDataSource;}/*** @param dynamicDefaultTargetDataSource* the dynamicDefaultTargetDataSource to set*/public void setDynamicDefaultTargetDataSource(Object dynamicDefaultTargetDataSource) {this.dynamicDefaultTargetDataSource = dynamicDefaultTargetDataSource;}/*** @param dataSource* @throws Exception* @description 创建新数据源时检查数据源是否存在*/public void createDataSourceWithCheck(DataSource dataSource) throws Exception {@NonNull Long src_id = dataSource.getSrc_id();log.info("正在检查数据源："+src_id);Map<Object, Object> currentDynamicTargetDataSources = this.dynamicTargetDataSources;if (currentDynamicTargetDataSources.containsKey(src_id)) {log.info("数据源"+src_id+"之前已经创建，准备测试数据源是否正常...");DruidDataSource druidDataSource = (DruidDataSource) currentDynamicTargetDataSources.get(src_id);boolean rightFlag = true;Connection connection = null;try {log.info(src_id+"数据源的概况->当前闲置连接数："+druidDataSource.getPoolingCount());long activeCount = druidDataSource.getActiveCount();log.info(src_id+"数据源的概况->当前活动连接数："+activeCount);if(activeCount > 0) {log.info(src_id+"数据源的概况->活跃连接堆栈信息："+druidDataSource.getActiveConnectionStackTrace());}log.info("准备获取数据库连接...");connection = druidDataSource.getConnection();log.info("数据源"+src_id+"正常");} catch (Exception e) {log.error(e.getMessage(),e); //把异常信息打印到日志文件rightFlag = false;log.info("缓存数据源"+src_id+"已失效，准备删除...");if(delDatasources(src_id)) {log.info("缓存数据源删除成功");} else {log.info("缓存数据源删除失败");}} finally {if(null != connection) {connection.close();}}if(rightFlag) {log.info("不需要重新创建数据源");return;} else {log.info("准备重新创建数据源...");createDataSource(dataSource);log.info("重新创建数据源完成");}} else {createDataSource(dataSource);}}private void createDataSource(DataSource dataSource) throws Exception {DBUtils dbUtils = SpringUtils.getBean(DBUtils.class);@NonNull Long src_id = dataSource.getSrc_id();log.info("准备创建数据源"+src_id);String db_type = dataSource.getDb_type();String username = dataSource.getSrc_user();String password = dataSource.getSrc_pwd();String ip_addr = dataSource.getIp_addr();String src_port = dataSource.getSrc_port();@NonNull String src_sid = dataSource.getSrc_sid();String url = null;String driveClass = null;if("mysql".equalsIgnoreCase(db_type)) {driveClass = DBUtils.MYSQL_DRIVER;dbUtils.setMySQLYUrl(ip_addr,src_port,src_sid);url = dbUtils.getMySQLYUrl();} else if("oracle".equalsIgnoreCase(db_type)){driveClass = DBUtils.ORACLE_DRIVER;dbUtils.setOracleUrl(ip_addr,src_port,src_sid);url = dbUtils.getOracleUrl();} else if("dm".equalsIgnoreCase(db_type)){driveClass = DBUtils.DM_DRIVER;dbUtils.setDMUrl(ip_addr,src_port,src_sid);url = dbUtils.getDMUrl();} else if("sqlserver".equalsIgnoreCase(db_type)){driveClass = DBUtils.SQLSERVER_DRIVER;dbUtils.setSQLserverUrl(ip_addr,src_port,src_sid);url = dbUtils.getSQLserverUrl();}if(testDatasource(src_id.toString(),driveClass,url,username,password)) {boolean result = this.createDataSource(src_id.toString(), driveClass, url, username, password, db_type);if(!result) {log.error("数据源"+src_id+"配置正确，但是创建失败");throw new ADIException("数据源"+src_id+"配置正确，但是创建失败",500);}} else {log.error("数据源配置有错误");throw new ADIException("数据源配置有错误",500);}}/*** @description 自定义创建数据源* @param key* @param driveClass* @param url* @param username* @param password* @param db_type* @return*/public boolean createDataSource(String key, String driveClass, String url, String username, String password, String db_type) {try {try { // 排除连接不上的错误Class.forName(driveClass);DriverManager.getConnection(url, username, password);//连接数据库} catch (Exception e) {return false;}@SuppressWarnings("resource") // HikariDataSource druidDataSource = new HikariDataSource();DruidDataSource druidDataSource = new DruidDataSource();druidDataSource.setName(key);druidDataSource.setDriverClassName(driveClass);druidDataSource.setUrl(url);druidDataSource.setUsername(username);druidDataSource.setPassword(password);druidDataSource.setInitialSize(1); //初始化时建立物理连接的个数。初始化发生在显示调用init方法，或者第一次getConnection时druidDataSource.setMaxActive(20); //最大连接池数量druidDataSource.setMaxWait(60000); //获取连接时最大等待时间，单位毫秒。当链接数已经达到了最大链接数的时候，应用如果还要获取链接就会出现等待的现象，等待链接释放并回到链接池，如果等待的时间过长就应该踢掉这个等待，不然应用很可能出现雪崩现象druidDataSource.setMinIdle(5); //最小连接池数量String validationQuery = "select 1 from dual";if("mysql".equalsIgnoreCase(db_type)) {driveClass = DBUtils.MYSQL_DRIVER;validationQuery = "select 1";} else if("oracle".equalsIgnoreCase(db_type)){driveClass = DBUtils.ORACLE_DRIVER;druidDataSource.setPoolPreparedStatements(true); //是否缓存preparedStatement,PSCache对支持游标的ORACLE数据库性能提升巨大,在mysql下建议关闭。druidDataSource.setMaxPoolPreparedStatementPerConnectionSize(50);druidDataSource.setConnectionProperties("oracle.net.CONNECT_TIMEOUT=6000;oracle.jdbc.ReadTimeout=300000");//对于耗时长的查询sql，会受限于ReadTimeout的控制，单位毫秒} else if("dm".equalsIgnoreCase(db_type)){driveClass = DBUtils.DM_DRIVER;validationQuery = "select 1";} else if("sqlserver".equalsIgnoreCase(db_type)){driveClass = DBUtils.SQLSERVER_DRIVER;validationQuery = "select 1";}//申请连接时执行validationQuery检测连接是否有效，这里建议配置为TRUE，防止取到的连接不可用druidDataSource.setTestOnBorrow(true);//申请连接的时候检测，如果空闲时间大于timeBetweenEvictionRunsMillis，执行validationQuery检测连接是否有效。druidDataSource.setTestWhileIdle(true);//用来检测连接是否有效的sql，要求是一个查询语句。如果validationQuery为null，testOnBorrow、testOnReturn、testWhileIdle都不会起作用。druidDataSource.setValidationQuery(validationQuery);//属性类型是字符串，通过别名的方式配置扩展插件，常用的插件有：监控统计用的filter:stat日志用的filter:log4j防御sql注入的filter:walldruidDataSource.setFilters("stat");//配置间隔多久才进行一次检测，检测需要关闭的空闲连接，单位是毫秒druidDataSource.setTimeBetweenEvictionRunsMillis(60000);//配置一个连接在池中最小生存的时间，单位是毫秒，这里配置为3分钟180000druidDataSource.setMinEvictableIdleTimeMillis(180000);//打开druid.keepAlive之后，当连接池空闲时，池中的minIdle数量以内的连接，空闲时间超过minEvictableIdleTimeMillis，// 则会执行keepAlive操作，即执行druid.validationQuery指定的查询SQL，一般为select * from dual，只要// minEvictableIdleTimeMillis设置的小于防火墙切断连接时间，就可以保证当连接空闲时自动做保活检测，不会被防火墙切断druidDataSource.setKeepAlive(true);//是否移除泄露的连接/超过时间限制是否回收。druidDataSource.setRemoveAbandoned(true);//泄露连接的定义时间(要超过最大事务的处理时间)；单位为秒。这里配置为1小时druidDataSource.setRemoveAbandonedTimeout(3600);//移除泄露连接发生是是否记录日志druidDataSource.setLogAbandoned(true);druidDataSource.init();this.dynamicTargetDataSources.put(key, druidDataSource);// 将map赋值给父类的TargetDataSourcessetTargetDataSources(this.dynamicTargetDataSources);super.afterPropertiesSet();// 将TargetDataSources中的连接信息放入resolvedDataSources管理log.info(key+"数据源初始化成功");//log.info(key+"数据源的概况："+druidDataSource.dump());return true;} catch (Exception e) {log.error(e + "");return false;}}/*** 删除数据源* @param db_id* @return*/public boolean delDatasources(Long db_id) {Map<Object, Object> currentDynamicTargetDataSources = this.dynamicTargetDataSources;if (currentDynamicTargetDataSources.containsKey(db_id)) {Set<DruidDataSource> druidDataSourceInstances = DruidDataSourceStatManager.getDruidDataSourceInstances();for (DruidDataSource l : druidDataSourceInstances) {if (db_id.equals(l.getName())) {currentDynamicTargetDataSources.remove(db_id);DruidDataSourceStatManager.removeDataSource(l);// 将map赋值给父类的TargetDataSourcessetTargetDataSources(currentDynamicTargetDataSources);// 将TargetDataSources中的连接信息放入resolvedDataSources管理super.afterPropertiesSet();return true;}}return false;} else {return false;}}/*** 测试数据源连接* @param driveClass* @param url* @param username* @param password* @return*/public boolean testDatasource(String key,String driveClass, String url, String username, String password) {try {Class.forName(driveClass);DriverManager.getConnection(url, username, password);return true;} catch (Exception e) {return false;}}} 与50位技术专家面对面20年技术见证，附赠技术全景图

总结

以上是生活随笔为你收集整理的SpringBoot+AOP实现多数据源动态切换的全部内容，希望文章能够帮你解决所遇到的问题。

如果觉得生活随笔网站内容还不错，欢迎将生活随笔推荐给好友。