ShardingSphere实现本地注册中心，动态刷新actualDataNodes

ShardingSphere实现本地注册中心，动态刷新actualDataNodes

看过ShardingSphere4.0.0-RC1实现动态刷新actualDataNodes，不使用zookeeper等第三方注册中心这篇文章，但是发现其可操作性还是写的太差，很多还需要自己琢磨，故此写下这篇文章记录自己的琢磨过程

坑1–RegistryCenter 这个类在哪儿

需要引入sharding的服务编排核心jar包，举例：gradle

compile group: 'org.apache.shardingsphere', name: 'sharding-orchestration-core', version: '4.0.1'

坑2–OrchestrationShardingDataSourceFactory怎么也没有

接着引包

compile group: 'org.apache.shardingsphere', name: 'sharding-jdbc-orchestration-spring-namespace', version: '4.0.1'

工程代码

关于sharding SPI方面的说明请自行查看官方文档

步骤1

• 在resources/META-INF/services下建立文件org.apache.shardingsphere.orchestration.reg.api.RegistryCenter
• 建立类com.changan.carbond.shardingjdbc.LocalRegistryCenter
• 文件中的内容 com.xstudio.shardingjdbc.LocalRegistryCenter和上述类保持一致【SPI注册】

步骤2–LocalRegistryCenter

public class LocalRegistryCenter implements RegistryCenter {
    public static Map<String, DataChangedEventListener> listeners = new ConcurrentHashMap<>();

    private RegistryCenterConfiguration config;
	
    private Properties properties;
    /**
	 * public 是为了在重置节点的时候减少去重新读配置
	 */
    public static Map<String, String> values = new ConcurrentHashMap<>();

    @Override
    public void init(RegistryCenterConfiguration config) {
        this.config = config;
    }

    @Override
    public String get(String key) {
        return values.get(key);
    }

    @Override
    public String getDirectly(String key) {
        return values.get(key);
    }

    @Override
    public boolean isExisted(String key) {
        return values.containsKey(key);
    }

    @Override
    public List<String> getChildrenKeys(String key) {
        return null;
    }

    @Override
    public void persist(String key, String value) {
        values.put(key, value);
    }

    @Override
    public void update(String key, String value) {
        values.put(key, value);
    }

    @Override
    public void persistEphemeral(String key, String value) {
        values.put(key, value);
    }

    @Override
    public void watch(String key, DataChangedEventListener dataChangedEventListener) {
        if (null != dataChangedEventListener) {
            // 将数据改变的事件监听器缓存下来
            listeners.put(key, dataChangedEventListener);
        }
    }

    @Override
    public void close() {
        config = null;
    }

    @Override
    public void initLock(String key) {

    }

    @Override
    public boolean tryLock() {
        return false;
    }

    @Override
    public void tryRelease() {

    }

    @Override
    public String getType() {
        // 【关键点1】，留着文章后续引用
        return "shardingLocalRegisterCenter";
    }

    @Override
    public Properties getProperties() {
        return properties;
    }

    @Override
    public void setProperties(Properties properties) {
        LocalRegistryCenter.properties = properties;
    }
}

步骤3

// 数据源设置
 Map<String, DataSource> dataSourceMap = new LinkedHashMap<>();
 dataSourceMap.put("哈哈", dataSource);
 
 // 属性设置
 Properties properties = new Properties();
 properties.setProperty("sql.show", "false");

 OchestrationConfiguration configuration = new OrchestrationConfiguration("【关键点2】",
         new RegistryCenterConfiguration("【关键点1】的值"),
         false);
 OrchestrationShardingDataSourceFactory.createDataSource(dataSourceMap, shardingRuleConfig, properties, configuration)

步骤4–重新加载节点

这边对服务开了一个api接口，方便外部调用触发重新加载节点

// 获取已有的配置
String rules = LocalRegistryCenter.values.get("/【关键点2】/config/schema/logic_db/rule");
// 省略改rule的逻辑，可以字符串替换什么的，我是这样操作的
String newRule = rules.replace("...","...");

LocalRegistryCenter.listeners.get("/【关键点2】/config/schema")
                .onChange(new DataChangedEvent("/【关键点2】/config/schema/logic_db/rule", newRule, DataChangedEvent.ChangedType.UPDATED));