Kafka broker配置介绍（四）

这部分内容对了解系统和提高软件性能都有很大的帮助，kafka官网上也给出了比较详细的配置详单，但是我们还是直接从代码来看broker到底有哪些配置需要我们去了解的，配置都有英文注释，所以每一部分是干什么的就不翻译了，都能看懂：

100

101

102

103

104

105

106

107

108

/**
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 * 
 *    http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
 
package kafka.server
 
import java.util.Properties
import kafka.utils.{Utils, ZKConfig}
import kafka.message.Message
 
/**
 * Configuration settings for the kafka server
 */
class KafkaConfig(props: Properties) extends ZKConfig(props) {
  /* the port to listen and accept connections on */
  val port: Int = Utils.getInt(props, "port", 6667)
 
  /* hostname of broker. If not set, will pick up from the value returned from getLocalHost. If there are multiple interfaces getLocalHost may not be what you want. */
  val hostName: String = Utils.getString(props, "hostname", null)
 
  /* the broker id for this server */
  val brokerId: Int = Utils.getInt(props, "brokerid")
   
  /* the SO_SNDBUFF buffer of the socket sever sockets */
  val socketSendBuffer: Int = Utils.getInt(props, "socket.send.buffer", 100*1024)
   
  /* the SO_RCVBUFF buffer of the socket sever sockets */
  val socketReceiveBuffer: Int = Utils.getInt(props, "socket.receive.buffer", 100*1024)
   
  /* the maximum number of bytes in a socket request */
  val maxSocketRequestSize: Int = Utils.getIntInRange(props, "max.socket.request.bytes", 100*1024*1024, (1, Int.MaxValue))
 
  /* the maximum size of message that the server can receive */
  val maxMessageSize = Utils.getIntInRange(props, "max.message.size", 1000000, (0, Int.MaxValue))
 
  /* the number of worker threads that the server uses for handling all client requests*/
  val numThreads = Utils.getIntInRange(props, "num.threads", Runtime.getRuntime().availableProcessors, (1, Int.MaxValue))
   
  /* the interval in which to measure performance statistics */
  val monitoringPeriodSecs = Utils.getIntInRange(props, "monitoring.period.secs", 600, (1, Int.MaxValue))
   
  /* the default number of log partitions per topic */
  val numPartitions = Utils.getIntInRange(props, "num.partitions", 1, (1, Int.MaxValue))
   
  /* the directory in which the log data is kept */
  val logDir = Utils.getString(props, "log.dir")
   
  /* the maximum size of a single log file */
  val logFileSize = Utils.getIntInRange(props, "log.file.size", 1*1024*1024*1024, (Message.MinHeaderSize, Int.MaxValue))
 
  /* the maximum size of a single log file for some specific topic */
  val logFileSizeMap = Utils.getTopicFileSize(Utils.getString(props, "topic.log.file.size", ""))
 
  /* the maximum time before a new log segment is rolled out */
  val logRollHours = Utils.getIntInRange(props, "log.roll.hours", 24*7, (1, Int.MaxValue))
 
  /* the number of hours before rolling out a new log segment for some specific topic */
  val logRollHoursMap = Utils.getTopicRollHours(Utils.getString(props, "topic.log.roll.hours", ""))
 
  /* the number of hours to keep a log file before deleting it */
  val logRetentionHours = Utils.getIntInRange(props, "log.retention.hours", 24*7, (1, Int.MaxValue))
 
  /* the number of hours to keep a log file before deleting it for some specific topic*/
  val logRetentionHoursMap = Utils.getTopicRetentionHours(Utils.getString(props, "topic.log.retention.hours", ""))
   
  /* the maximum size of the log before deleting it */
  val logRetentionSize = Utils.getLong(props, "log.retention.size", -1)
 
  /* the maximum size of the log for some specific topic before deleting it */
  val logRetentionSizeMap = Utils.getTopicRetentionSize(Utils.getString(props, "topic.log.retention.size", ""))
 
  /* the frequency in minutes that the log cleaner checks whether any log is eligible for deletion */
  val logCleanupIntervalMinutes = Utils.getIntInRange(props, "log.cleanup.interval.mins", 10, (1, Int.MaxValue))
   
  /* enable zookeeper registration in the server */
  val enableZookeeper = Utils.getBoolean(props, "enable.zookeeper", true)
 
  /* the number of messages accumulated on a log partition before messages are flushed to disk */
  val flushInterval = Utils.getIntInRange(props, "log.flush.interval", 500, (1, Int.MaxValue))
 
  /* the maximum time in ms that a message in selected topics is kept in memory before flushed to disk, e.g., topic1:3000,topic2: 6000  */
  val flushIntervalMap = Utils.getTopicFlushIntervals(Utils.getString(props, "topic.flush.intervals.ms", ""))
 
  /* the frequency in ms that the log flusher checks whether any log needs to be flushed to disk */
  val flushSchedulerThreadRate = Utils.getInt(props, "log.default.flush.scheduler.interval.ms",  3000)
 
  /* the maximum time in ms that a message in any topic is kept in memory before flushed to disk */
  val defaultFlushIntervalMs = Utils.getInt(props, "log.default.flush.interval.ms", flushSchedulerThreadRate)
 
   /* the number of partitions for selected topics, e.g., topic1:8,topic2:16 */
  val topicPartitionsMap = Utils.getTopicPartitions(Utils.getString(props, "topic.partition.count.map", ""))
 
  /* the maximum length of topic name*/
  val maxTopicNameLength = Utils.getIntInRange(props, "max.topic.name.length", 255, (1, Int.MaxValue))
}

上面这段代码来自kafka.server包下的KafkaConfig类，之前我们就说过，broker就是kafka中的server，所以讲配置放在这个包中也不奇怪。这里我们顺着代码往下读，也顺便看看scala的语法。和java一样也要import相关的包，kafka将同一包内的两个类写在大括号中：

1	`import` `kafka.utils.{Utils, ZKConfig}`

然后我们看类的写法：

1	`class` `KafkaConfig(props:` `Properties)` `extends` `ZKConfig(props)`

我们看到在加载kafkaConfig的时候会加载一个properties对象，同时也会加载有关zookeeper的properties，这个时候我们可以回忆一下，之前我们启动kafka broker的命令：

1. 启动zookeeper server ：bin/zookeeper-server-start.sh ../config/zookeeper.properties & (用&是为了能退出命令行)

2. 启动kafka server: bin/kafka-server-start.sh ../config/server.properties &

所以你能明白，初始化kafka broker的时候程序一定是去加载位于config文件夹下的properties，这个和java都一样没有区别。当然properties我们也可以通过程序来给出，这个我们后面再说，继续看我们的代码。既然找到了对应的properties文件，我们就结合代码和properties一起来看。

Kafka broker的properties中，将配置分为以下六类：

l Server Basics：关于brokerid，hostname等配置

l Socket Server Settings：关于传输的配置，端口、buffer的区间等。

l Log Basics：配置log的位置和partition的数量。

l Log Flush Policy：这部分是kafka配置中最重要的部分，决定了数据flush到disk的策略。

l Log Retention Policy：这部分主要配置日志处理时的策略。

l Zookeeper：配置zookeeper的相关信息。

在文件properties中的配置均出现在kafkaConfig这个类中，我们再看看kafkaConfig中的代码：

/* the broker id for this server */
  val brokerId: Int = Utils.getInt(props, "brokerid")
   
  /* the SO_SNDBUFF buffer of the socket sever sockets */
  val socketSendBuffer: Int = Utils.getInt(props, "socket.send.buffer", 100*1024)

凡是参数中有三个的，最后一个是default，而参数只有两个的则要求你一定要配置，否则的话则报错。当然在这么多参数中肯定是有一些经验参数的，至于这些参数怎么配置我确实没有一个特别的推荐，需要在不断的测试中才能磨合出来。

当然你也可以将配置写在程序里，然后通过程序去启动broker，这样kafka的配置就可以像下面一样写：

Properties props = new Properties();
props.setProperty("port","9093");
props.setProperty("log.dir","/home/kafka/data1");

我倒是觉得配置还是直接写在配置文件中比较好，如果需要修改也不会影响正在运行的服务，写在内存中，总是会有些不方便的地方。所以还是建议大家都写配置好了，后面讲到的producer和consumer都一样。

这里再提两个参数一个是brokerid，每个broker的id必须要区分；第二个参数是hostname，这个是broker和producer、consumer联系的关键，这里记住一定要改成你的地址和端口，否则永远连得都是localhost。

--------------------------------------------------------

下一篇将写producer和consumer的配置了，涉及到这部分就要开始编程了，写着写着又往源码里看进去了，下篇会先讲如何搭建开发环境，然后再写两个简单那的例子去熟悉配置。

Kafka broker配置介绍 （四）

Kafka broker配置介绍（四）