You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kafka.apache.org by "Tomasz Gąska (JIRA)" <ji...@apache.org> on 2018/03/19 13:51:00 UTC
[jira] [Created] (KAFKA-6682) Kafka reconnection after broker
restart
Tomasz Gąska created KAFKA-6682:
-----------------------------------
Summary: Kafka reconnection after broker restart
Key: KAFKA-6682
URL: https://issues.apache.org/jira/browse/KAFKA-6682
Project: Kafka
Issue Type: Bug
Components: clients
Affects Versions: 1.0.0
Reporter: Tomasz Gąska
I am using kafka producer plugin for logback (danielwegener) with the clients library 1.0.0 and after restart of broker all my JVMs connected to it get tons of the exceptions:
{code:java}
11:22:48.738 [kafka-producer-network-thread | app-logback-relaxed] cid:
clid: E [ @] a: o.a.k.c.p.internals.Sender - [Producer
clientId=id-id-logback-relaxed] Uncaught error in kafka producer I/O
thread: ex:java.lang.NullPointerException: null
at org.apache.kafka.common.network.Selector.pollSelectionKeys(Selector.java:436)
at org.apache.kafka.common.network.Selector.poll(Selector.java:399)
at org.apache.kafka.clients.NetworkClient.poll(NetworkClient.java:460)
at org.apache.kafka.clients.producer.internals.Sender.run(Sender.java:239)
at org.apache.kafka.clients.producer.internals.Sender.run(Sender.java:163)
at java.lang.Thread.run(Thread.java:798){code}
During restart there are still other brokers available behind LB.
Dosen't matter kafka is up again, only restarting JVM helps
{code:java}
<appender name="kafkaLogAppender" class="com.github.danielwegener.logback.kafka.KafkaAppender">
<!-- This is the default encoder that encodes every log message to an utf8-encoded string -->
<encoder>
<pattern>%date{"yyyy-MM-dd'T'HH:mm:ss.SSS'Z'"} ${HOSTNAME}
[%thread] %logger{32} - %message ex:%exf%n</pattern>
</encoder>
<topic>mytopichere</topic>
<!-- we don't care how the log messages will be partitioned -->
<keyingStrategy class="com.github.danielwegener.logback.kafka.keying.HostNameKeyingStrategy" />
<!-- use async delivery. the application threads are not blocked by logging -->
<deliveryStrategy class="com.github.danielwegener.logback.kafka.delivery.AsynchronousDeliveryStrategy" />
<!-- each <producerConfig> translates to regular kafka-client config (format: key=value) -->
<!-- producer configs are documented here:
https://kafka.apache.org/documentation.html#newproducerconfigs -->
<!-- bootstrap.servers is the only mandatory producerConfig -->
<producerConfig>bootstrap.servers=10.99.99.1:9092</producerConfig>
<!-- don't wait for a broker to ack the reception of a batch. -->
<producerConfig>acks=0</producerConfig>
<!-- even if the producer buffer runs full, do not block the application but start to drop messages -->
<producerConfig>block.on.buffer.full=false</producerConfig>
<!-- define a client-id that you use to identify yourself against the kafka broker -->
<producerConfig>client.id=${HOSTNAME}-${CONTEXT_NAME}-logback-relaxed</producerConfig>
<!-- use gzip to compress each batch of log messages. valid values: none, gzip, snappy -->
<!-- do przetestowania narzut -->
<producerConfig>compression.type=none</producerConfig>
<!-- there is no fallback <appender-ref>. If this appender
cannot deliver, it will drop its messages. -->
<producerConfig>max.block.ms=0</producerConfig>
</appender>{code}
I provide loadbalancer address in bootstrap servers here. There are three kafka brokers behind.
{code:java}
java version "1.7.0"
Java(TM) SE Runtime Environment (build pap6470sr9fp60ifix-20161110_01(SR9 FP60)+IV90630+IV90578))
IBM J9 VM (build 2.6, JRE 1.7.0 AIX ppc64-64 Compressed References 20161005_321282 (JIT enabled, AOT enabled)
J9VM - R26_Java726_SR9_20161005_1259_B321282
JIT - tr.r11_20161001_125404
GC - R26_Java726_SR9_20161005_1259_B321282_CMPRSS
J9CL - 20161005_321282)
JCL - 20161021_01 based on Oracle jdk7u121-b15{code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)