关于 jetcd 在 etcd 服务端挂掉又重启的情况下， watch 会抛出异常，有什么优雅的解决方案吗

• 请不要在回答技术问题时复制粘贴 AI 生成的内容

This topic created in 2299 days ago, the information mentioned may be changed or developed.

环境 jectd 版本 0.5

etcd 版本 3.4

本人泥腿子 java 对于使用第三方包，假如是第三方包抛出异常，是不是没办法捕获解决。

如果我手动关闭 etcd 服务端，就会这样，这个异常是 jetcd 包里抛出的

00:22:31.569 [grpc-default-executor-1] INFO com.example.demo.EtcdClient - Watching for key=a
00:22:31.577 [grpc-default-executor-1] INFO com.example.demo.EtcdClient - type=PUT, key=a, value=24444
00:22:58.621 [grpc-nio-worker-ELG-1-2] DEBUG io.grpc.netty.NettyClientHandler - [id: 0x138691d4, L:/127.0.0.1:58759 - R:/127.0.0.1:2379] INBOUND DATA: streamId=3 padding=0 endStream=false length=52 bytes=000000002f0a1a08b298eaf1948386fccd0110cdd2918bd2b881cf8e01181120035a11120f0a01611010181120022a0433333333
00:22:58.622 [grpc-default-executor-1] INFO com.example.demo.EtcdClient - Watching for key=a
00:22:58.622 [grpc-default-executor-1] INFO com.example.demo.EtcdClient - type=PUT, key=a, value=3333
00:23:23.851 [grpc-nio-worker-ELG-1-2] DEBUG io.grpc.netty.NettyClientHandler - [id: 0x138691d4, L:/127.0.0.1:58759 - R:/127.0.0.1:2379] INBOUND GO_AWAY: lastStreamId=2147483647 errorCode=0 length=0 bytes=
00:23:23.856 [grpc-nio-worker-ELG-1-2] DEBUG io.grpc.netty.NettyClientHandler - [id: 0x138691d4, L:/127.0.0.1:58759 - R:/127.0.0.1:2379] INBOUND PING: ack=false bytes=73747577769755401
00:23:23.856 [grpc-nio-worker-ELG-1-2] DEBUG io.grpc.netty.NettyClientHandler - [id: 0x138691d4, L:/127.0.0.1:58759 - R:/127.0.0.1:2379] OUTBOUND PING: ack=true bytes=73747577769755401
00:23:23.857 [grpc-nio-worker-ELG-1-2] DEBUG io.grpc.netty.NettyClientHandler - [id: 0x138691d4, L:/127.0.0.1:58759 - R:/127.0.0.1:2379] OUTBOUND GO_AWAY: lastStreamId=0 errorCode=2 length=24 bytes=436f6e6e656374696f6e2072657365742062792070656572
00:23:23.861 [grpc-nio-worker-ELG-1-2] DEBUG io.netty.handler.codec.http2.Http2ConnectionHandler - [id: 0x138691d4, L:/127.0.0.1:58759 ! R:/127.0.0.1:2379] Sending GOAWAY failed: lastStreamId '0', errorCode '2', debugData 'Connection reset by peer'. Forcing shutdown of the connection.
java.io.IOException: Broken pipe
	at sun.nio.ch.FileDispatcherImpl.writev0(Native Method)
	at sun.nio.ch.SocketDispatcher.writev(SocketDispatcher.java:51)
	at sun.nio.ch.IOUtil.write(IOUtil.java:148)
	at sun.nio.ch.SocketChannelImpl.write(SocketChannelImpl.java:504)
	at io.netty.channel.socket.nio.NioSocketChannel.doWrite(NioSocketChannel.java:423)

然后我重启 etcd 服务端，继续 put key 还是能继续 watch 到的：

00:23:50.692 [grpc-nio-worker-ELG-1-3] DEBUG io.grpc.netty.NettyClientHandler - [id: 0x56feb4a4, L:/127.0.0.1:58801 - R:/127.0.0.1:2379] INBOUND DATA: streamId=3 padding=0 endStream=false length=71 bytes=00000000420a1a08b298eaf1948386fccd0110cdd2918bd2b881cf8e01181320045a11120f0a01611010181220032a04333333335a11120f0a01611010181320...
00:23:50.692 [grpc-default-executor-2] INFO com.example.demo.EtcdClient - Watching for key=a
00:23:50.693 [grpc-default-executor-2] INFO com.example.demo.EtcdClient - type=PUT, key=a, value=3333
00:23:50.693 [grpc-default-executor-2] INFO com.example.demo.EtcdClient - type=PUT, key=a, value=4444

假如我在线上要用这个 java 的 etcd 客户端，我肯定起个线程来 watch，但是问题来了，假如 etcd 挂了抛出了这个异常然后这个异常导致线程退出，我难道还要线程退出的时候重新开始 watch 吗，这样感觉就麻烦了。话说面对这样的情况该如何优雅的解决，诚心请教各位大佬了

项目里跑的 demo 参考 jetcd 的 watch demo

7 replies • 2021-02-23 20:35:05 +08:00