Elasticsearch出现Connection reset by peer分析

 

1.异常：

2024-06-13 13:17:10.539 WARN [http-nio-30411-exec-9]com.longdaotech.config.ESConfig -onFailure node:[host=http://192.168.239.253:9200]

2024/6/13 13:17:10 2024-06-13 13:17:10.541 WARN [http-nio-30411-exec-9]com.longdaotech.config.ESConfig -onFailure node:[host=http://192.168.239.249:9200]

2024/6/13 13:17:10 2024-06-13 13:17:10.542 WARN [http-nio-30411-exec-9]com.longdaotech.config.ESConfig -onFailure node:[host=http://192.168.239.67:9200]

2024/6/13 13:17:10 2024-06-13 13:17:10.543 ERROR [http-nio-30411-exec-9]com.longdaotech.front.modules.service.impl.GoodsSearchServiceImpl -{"monitorCode":"ESnull500601","message":"商品索引查询失败","contextContent":""}

2024/6/13 13:17:10 java.io.IOException: Connection reset by peer

2024/6/13 13:17:10 at org.elasticsearch.client.RestClient.extractAndWrapCause(RestClient.java:793)

2024/6/13 13:17:10 at org.elasticsearch.client.RestClient.performRequest(RestClient.java:218)

2024/6/13 13:17:10 at org.elasticsearch.client.RestClient.performRequest(RestClient.java:221)

2024/6/13 13:17:10 at org.elasticsearch.client.RestClient.performRequest(RestClient.java:221)

2024/6/13 13:17:10 at org.elasticsearch.client.RestClient.performRequest(RestClient.java:205)

2024/6/13 13:17:10 at org.elasticsearch.client.RestHighLevelClient.internalPerformRequest(RestHighLevelClient.java:1454)

2024/6/13 13:17:10 at org.elasticsearch.client.RestHighLevelClient.performRequest(RestHighLevelClient.java:1424)

2024/6/13 13:17:10 at org.elasticsearch.client.RestHighLevelClient.performRequestAndParseEntity(RestHighLevelClient.java:1394)

2024/6/13 13:17:10 at org.elasticsearch.client.RestHighLevelClient.search(RestHighLevelClient.java:930)

2024/6/13 13:17:10 at com.longdaotech.front.modules.service.impl.GoodsSearchServiceImpl.goodsSerarch(GoodsSearchServiceImpl.java:361)

2024/6/13 13:17:10 at com.longdaotech.front.modules.controller.GoodsSearchController.searchGoods(GoodsSearchController.java:107)

2024/6/13 13:17:10 at sun.reflect.GeneratedMethodAccessor326.invoke(Unknown Source)

2024/6/13 13:17:10 at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)

2.分析

  为什么服务端会关闭链接？

服务器会有TCP的Keepalive 经过一段时间如果没有操作就会自动断开连接功能，而ES默认就是取服务器的时长配置

linux查看超时时间默认为两小时

注：当ES服务端的keepAlive短于ES客户端的keepAlive，也会导致：服务端已经关闭了连接，客户端继续复用该连接，抛出异常。

3.解决

手动设置KeepAliveStrategy来配置keepAlive，保证客户端keepAlive小于服务端keepAlive，让客户端先于服务端关闭连接 

# 查看服务器超时时间（单位秒）
cat /proc/sys/net/ipv4/tcp_keepalive_time

默认是7200s

解决办法有两个,一是设置keepalive小于服务器的超时时间，这样HttpClient会自己断开连接，下次再重连，由于SpringBoot没有提供设置改属性的地方，配置起来有点小麻烦，可以参考以下链接

ES ElasticSearch Connection reset by peer问题解决_阿里巴巴首席技术官的博客-CSDN博客

方案二是发送心跳，下面贴出代码

@Scheduled(fixedRate = 60 * 1000 * 5)
public void heartbeatToES() {try {RequestOptions requestOptions = RequestOptions.DEFAULT.toBuilder().build();boolean result = restTemplate.getClient().ping(requestOptions);log.info("es状态: {}", result);} catch (Exception e) {}
}

方案二 设置ES服务端 tcp_keepalive_time时间大于等于客户端时间

https://blog.csdn.net/weixin_42195284/article/details/122399262

                        
ES ElasticSearch Connection reset by peer问题解决_es connection reset by peer-CSDN博客