filebeat demo 配置在商业版 kafka 会出错 connection reset by peer

[fudali113]

我使用 公测版本可以使用，修改了商用版本的之后就会报这样子的错误

172.20.1.142:45936->172.17.102.2:9092: read: connection reset by peer
 2018-07-17T04:09:37.366Z	INFO	kafka/log.go:36	kafka message: client/metadata got error from broker while fetching metadata:%!(EXTRA *net.OpError=read tcp 172.20.1.142:45936->172.17.102.2:9092: read: connection reset by peer)
2018-07-17T04:09:37.366Z	INFO	kafka/log.go:36	kafka message: client/metadata no available broker to send metadata request to
2018-07-17T04:09:37.366Z	INFO	kafka/log.go:36	client/brokers resurrecting 3 dead seed brokers
2018-07-17T04:09:37.366Z	INFO	kafka/log.go:36	client/metadata retrying after 250ms... (1 attempts remaining)
 2018-07-17T04:09:37.616Z	INFO	kafka/log.go:36	client/metadata fetching metadata for all topics from broker 172.17.102.3:9092
 2018-07-17T04:09:37.617Z	INFO	kafka/log.go:36	Failed to connect to broker 172.17.102.3:9092: read tcp 172.20.1.142:49526->172.17.102.3:9092: read: connection reset by peer
 2018-07-17T04:09:37.617Z	INFO	kafka/log.go:36	kafka message: client/metadata got error from broker while fetching metadata:%!(EXTRA *net.OpError=read tcp 172.20.1.142:49526->172.17.102.3:9092: read: connection reset by peer)
2018-07-17T04:09:37.617Z	INFO	kafka/log.go:36	client/metadata fetching metadata for all topics from broker 172.17.102.1:9092
 2018-07-17T04:09:37.617Z	INFO	kafka/log.go:36	Failed to connect to broker 172.17.102.1:9092: read tcp 172.20.1.142:49556->172.17.102.1:9092: read: connection reset by peer
 2018-07-17T04:09:37.617Z	INFO	kafka/log.go:36	kafka message: client/metadata got error from broker while fetching metadata:%!(EXTRA *net.OpError=read tcp 172.20.1.142:49556->172.17.102.1:9092: read: connection reset by peer)
2018-07-17T04:09:37.617Z	INFO	kafka/log.go:36	client/metadata fetching metadata for all topics from broker 172.17.102.2:9092
 2018-07-17T04:09:37.618Z	INFO	kafka/log.go:36	Failed to connect to broker 172.17.102.2:9092: read tcp 172.20.1.142:45942->172.17.102.2:9092: read: connection reset by peer
 2018-07-17T04:09:37.618Z	INFO	kafka/log.go:36	kafka message: client/metadata got error from broker while fetching metadata:%!(EXTRA *net.OpError=read tcp 172.20.1.142:45942->172.17.102.2:9092: read: connection reset by peer)
2018-07-17T04:09:37.618Z	INFO	kafka/log.go:36	kafka message: client/metadata no available broker to send metadata request to
2018-07-17T04:09:37.618Z	INFO	kafka/log.go:36	client/brokers resurrecting 3 dead seed brokers
2018-07-17T04:09:37.618Z	INFO	kafka/log.go:36	kafka message: Closing Client
2018-07-17T04:09:37.618Z	ERROR	kafka/client.go:74	Kafka connect fails with: kafka: client has run out of available brokers to talk to (Is your cluster reachable?)
2018-07-17T04:09:37.618Z	ERROR	pipeline/output.go:74	Failed to connect: kafka: client has run out of available brokers to talk to (Is your cluster reachable?)
2018-07-17T04:09:37.618Z	INFO	kafka/log.go:36	kafka message: Initializing new client
2018-07-17T04:09:37.618Z	INFO	kafka/log.go:36	client/metadata fetching metadata for all topics from broker 172.17.102.2:9092

[wp-jackliu]

https://help.aliyun.com/document_detail/68165.html 在分布式环境下，由于网络等原因，偶尔的发送失败是常见的。导致这种失败的原因有可能是消息已经发送成功，但是 Ack 失败，也有可能是确实没发送成功。

消息队列 Kafka 是 VIP 网络架构，会主动掐掉空闲连接（30 秒没活动），也就是说，不是一直活跃的客户端会经常收到”connection rest by peer”这样的错误，因此建议都考虑重试消息发送。