dubbo集群容错策略

相关参数在

org.apache.dubbo.common.constants.ClusterRules

策略名称	作用	使用场景
failover	失败启动切换 当出现失败会重试其他服务器，会带来更长延迟，通过 retries=2 来设置重试次数（不含第一次）	通常用于读操作
failfast	快速失败 只发起一次调用，失败立即报错	通常用于非幂等写操作，比如新增记录
failsafe	失败安全 出现异常会直接忽略	通常用于写入审计日志等操作
failback	失败自动恢复 后台记录失败的请求，然后定时重发	通常用于消息通知相关的操作
forking	并行调用多个服务器，只要一个成功就返回	通常用于实时性要求较高的读操作，需要浪费更多资源，通过 forks=2 设置最大并行数
broadcast	广播调用所有提供者，逐个调用，任何一台报错则报错	通常用于通知所有提供者更新缓存或日志等本地资源信息

默认值 failover

对应策略的异常信息如下

@DubboReference(cluster = ClusterRules.FAIL_OVER)

org.apache.dubbo.rpc.RpcException: Invoke remote method timeout. method: getHello, provider: DefaultServiceInstance{serviceName='my-fault-tolerance-provider', host='192.168.106.105', port=20881, enabled=true, healthy=true, metadata={dubbo.endpoints=[{"port":20881,"protocol":"dubbo"}], dubbo.metadata-service.url-params={"connections":"1","version":"1.0.0","dubbo":"2.0.2","release":"3.0.7","side":"provider","port":"20881","protocol":"dubbo"}, dubbo.metadata.revision=d9eea05f8a4e8ec289758c67ec03017f, dubbo.metadata.storage-type=local}}, service{name='com.ghy.www.api.Service1',group='null',version='null',protocol='dubbo',params={side=provider, release=3.0.7, methods=getHello, logger=slf4j, deprecated=false, dubbo=2.0.2, interface=com.ghy.www.api.Service1, service-name-mapping=true, generic=false, application=my-fault-tolerance-provider, background=false, dynamic=true, anyhost=true},}, cause: org.apache.dubbo.remoting.TimeoutException: Waiting server-side response timeout by scan timer. start time: 2023-11-05 17:08:32.621, end time: 2023-11-05 17:08:33.648, client elapsed: 0 ms, server elapsed: 1027 ms, timeout: 1000 ms, request: Request [id=14, version=2.0.2, twoway=true, event=false, broken=false, data=null], channel: /192.168.106.105:55533 -> /192.168.106.105:20881

@DubboReference(cluster = ClusterRules.FAIL_FAST)

org.apache.dubbo.remoting.TimeoutException: Waiting server-side response timeout by scan timer. start time: 2023-11-05 17:05:44.807, end time: 2023-11-05 17:05:45.829, client elapsed: 1 ms, server elapsed: 1021 ms, timeout: 1000 ms, request: Request [id=4, version=2.0.2, twoway=true, event=false, broken=false, data=null], channel: /192.168.106.105:55533 -> /192.168.106.105:20881

@DubboReference(cluster = ClusterRules.FAIL_SAFE)

 org.apache.dubbo.rpc.RpcException: Invoke remote method timeout. method: getHello, provider: DefaultServiceInstance{serviceName='my-fault-tolerance-provider', host='192.168.106.105', port=20881, enabled=true, healthy=true, metadata={dubbo.endpoints=[{"port":20881,"protocol":"dubbo"}], dubbo.metadata-service.url-params={"connections":"1","version":"1.0.0","dubbo":"2.0.2","release":"3.0.7","side":"provider","port":"20881","protocol":"dubbo"}, dubbo.metadata.revision=d9eea05f8a4e8ec289758c67ec03017f, dubbo.metadata.storage-type=local}}, service{name='com.ghy.www.api.Service3',group='null',version='null',protocol='dubbo',params={side=provider, release=3.0.7, methods=getHello, logger=slf4j, deprecated=false, dubbo=2.0.2, interface=com.ghy.www.api.Service3, service-name-mapping=true, generic=false, application=my-fault-tolerance-provider, background=false, dynamic=true, anyhost=true},}, cause: org.apache.dubbo.remoting.TimeoutException: Waiting server-side response timeout by scan timer. start time: 2023-11-05 17:11:07.645, end time: 2023-11-05 17:11:08.669, client elapsed: 0 ms, server elapsed: 1024 ms, timeout: 1000 ms, request: Request [id=22, version=2.0.2, twoway=true, event=false, broken=false, data=null], channel: /192.168.106.105:55533 -> /192.168.106.105:20881

@DubboReference(cluster = ClusterRules.FAIL_BACK)

无异常

 @DubboReference(cluster = ClusterRules.FAIL_OVER, retries = 5)

org.apache.dubbo.remoting.TimeoutException: Waiting server-side response timeout by scan timer. start time: 2023-11-05 17:22:17.388, end time: 2023-11-05 17:22:18.422, client elapsed: 0 ms, server elapsed: 1034 ms, timeout: 1000 ms, request: Request [id=3058, version=2.0.2, twoway=true, event=false, broken=false, data=null], channel: /192.168.106.105:55535 -> /192.168.106.105:20883

@DubboReference(cluster = ClusterRules.FAIL_BACK, retries = 5)

 org.apache.dubbo.rpc.RpcException: Invoke remote method timeout. method: getHello, provider: DefaultServiceInstance{serviceName='my-fault-tolerance-provider', host='192.168.106.105', port=20883, enabled=true, healthy=true, metadata={dubbo.endpoints=[{"port":20883,"protocol":"dubbo"}], dubbo.metadata-service.url-params={"connections":"1","version":"1.0.0","dubbo":"2.0.2","release":"3.0.7","side":"provider","port":"20883","protocol":"dubbo"}, dubbo.metadata.revision=d9eea05f8a4e8ec289758c67ec03017f, dubbo.metadata.storage-type=local}}, service{name='com.ghy.www.api.Service9',group='null',version='null',protocol='dubbo',params={side=provider, release=3.0.7, methods=getHello, logger=slf4j, deprecated=false, dubbo=2.0.2, interface=com.ghy.www.api.Service9, service-name-mapping=true, generic=false, application=my-fault-tolerance-provider, background=false, dynamic=true, anyhost=true},}, cause: org.apache.dubbo.remoting.TimeoutException: Waiting server-side response timeout by scan timer. start time: 2023-11-05 17:27:53.514, end time: 2023-11-05 17:27:54.533, client elapsed: 1 ms, server elapsed: 1018 ms, timeout: 1000 ms, request: Request [id=3077, version=2.0.2, twoway=true, event=false, broken=false, data=null], channel: /192.168.106.105:55535 -> /192.168.106.105:20883

@DubboReference(cluster = ClusterRules.FORKING)

无异常

@DubboReference(cluster = ClusterRules.BROADCAST)

org.apache.dubbo.rpc.RpcException: Invoke remote method timeout. method: getHello, provider: DefaultServiceInstance{serviceName='my-fault-tolerance-provider', host='192.168.106.105', port=20881, enabled=true, healthy=true, metadata={dubbo.endpoints=[{"port":20881,"protocol":"dubbo"}], dubbo.metadata-service.url-params={"connections":"1","version":"1.0.0","dubbo":"2.0.2","release":"3.0.7","side":"provider","port":"20881","protocol":"dubbo"}, dubbo.metadata.revision=d9eea05f8a4e8ec289758c67ec03017f, dubbo.metadata.storage-type=local}}, service{name='com.ghy.www.api.Service6',group='null',version='null',protocol='dubbo',params={side=provider, release=3.0.7, methods=getHello, logger=slf4j, deprecated=false, dubbo=2.0.2, interface=com.ghy.www.api.Service6, service-name-mapping=true, generic=false, application=my-fault-tolerance-provider, background=false, dynamic=true, anyhost=true},}, cause: org.apache.dubbo.remoting.TimeoutException: Waiting server-side response timeout by scan timer. start time: 2023-11-05 17:37:01.405, end time: 2023-11-05 17:37:02.429, client elapsed: 0 ms, server elapsed: 1024 ms, timeout: 1000 ms, request: Request [id=3121, version=2.0.2, twoway=true, event=false, broken=false, data=null], channel: /192.168.106.105:55533 -> /192.168.106.105:20881 

FAIL_FAST 和 FAIL_OVER 的异常为 org.apache.dubbo.remoting.TimeoutException

FAIL_OVER、FAIL_SAFE、FAIL_BACK、BROADCAST 的异常为 org.apache.dubbo.rpc.RpcException