我们知道hadoop1.x之前的namenode存在两个主要的问题:1、namenode内存瓶颈的问题,2、namenode的单点故障的问题。针对这两个问题,hadoop2.x都对它进行改进和解决。其中,问题1中对namenode内存瓶颈的问题采用扩展namenode的方式来解决。对于问题2中的namenode的单点故障问题hadoop2.x采用的是HA的解决方案。apache hadoop 官方网站上提供了两种解决HDFS High Availability Using the Quorum Journal Manager 和High Availability with NFS。 本文是采用HDFS High Availability Using the Quorum Journal Manager 方案来实现HA。并且实现namenode单点故障自动切换的功能,这就需要借助与zookeeper集群来实现。下面详细的讲解一下通过zookeeper来实现HDFS High Availability Using the Quorum Journal Manager 单点故障自动切换的方案。 在介绍之前,首先说明一下我的集群规模:2个namenode(hadoop1,hadoop5),3个datanode(hadoop2,hadoop3,hadoop4)。 ------------------------------------------------------------------------------------------------ | IP地址 | 主机名 | NameNode | journalNode | DataNode | zookeeper | | 192.168.1.21 | hadoop1 | 是 | 是 | 否 | 是 | | 192.168.1.22 | hadoop2 | 否 | 是 | 是 | 是 | | 192.168.1.23 | hadoop3 | 否 | 是 | 是 | 是 | | 192.168.1.24 | hadoop4 | 否 | 是 | 是 | 是 | | 192.168.1.25 | hadoop5 | 是 | 是 | 否 | 是 | -------------------------------------------------------------------------------------------------- 1、首先当然是安装zookeeper的集群了 对于该集群的安装可以参考另一篇文章: http://www.cnblogs.com/ljy2013/p/4510143.html 。这篇文章详细介绍了zookeeper的安装 2、安装好了zookeeper集群之后,下一步就需要部署你自己的hadoop2.x的集群了。 对于hadoop2.x的集群,我部署的是hadoop2.6.0的集群,部署方法可以参考文章:http://www.cnblogs.com/ljy2013/articles/4345172.html 。这篇文章当中详细介绍了如何安装和部署简单的hadoop的集群。 3、这里对journalnode进行说明一下,这个节点是一个轻量级的进行,可以与hadoop的集群部署在同一台机器上,并且,它的部署只需要添加hadoop相应的配置参数即可。 4、修改hadoop集群的配置文件,这里需要修改的配置文件较多,修改的参数更多,并且比较重要。 (1)修改core-site.xml文件 <?xml version="1.0" encoding="UTF-8"?> <?xml-stylesheet type="text/xsl" href="configuration.xsl"?> <!-- Licensed 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. See accompanying LICENSE file. --> <!-- Put site-specific property overrides in this file. --> <configuration> <property> <name>fs.defaultFS</name> <value>hdfs://mycluster</value> </property> <property> <name>ha.zookeeper.quorum</name> <value>hadoop1:2181,hadoop2:2181,hadoop3:2181,hadoop4:2181,hadoop5:2181</value> </property> <!-- <property> <name>dfs.ha.automatic-failover.enabled</name> <value>true</value> </property> <property> <name>dfs.ha.fencing.methods</name> <value>sshfence</value> </property> <property> <name>dfs.ha.fencing.ssh.private-key-files</name> <value>/home/grid/.ssh/id_rsa_nn</value> </property> --> <property> <name>ha.zookeeper.session-timeout.ms</name> <value>60000</value> </property> <property> <name>ha.failover-controller.cli-check.rpc-timeout.ms</name> <value>60000</value> </property> <property> <name>ipc.client.connect.timeout</name> <value>20000</value> </property> </configuration> (2)修改hdfs-site.xml文件 <?xml version="1.0" encoding="UTF-8"?> <?xml-stylesheet type="text/xsl" href="configuration.xsl"?> <!-- Licensed 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. See accompanying LICENSE file. --> <!-- Put site-specific property overrides in this file. --> <configuration> <property> <name>dfs.nameservices</name> <value>mycluster</value> </property> <property> <name>dfs.ha.namenodes.mycluster</name> <value>nn1,nn2</value> </property> <property> <name>dfs.namenode.rpc-address.mycluster.nn1</name> <value>hadoop1:9000</value> </property> <property> <name>dfs.namenode.rpc-address.mycluster.nn2</name> <value>hadoop5:9000</value> </property> <property> <name>dfs.namenode.http-address.mycluster.nn1</name> <value>hadoop1:50070</value> </property> <property> <name>dfs.namenode.http-address.mycluster.nn2</name> <value>hadoop5:50070</value> </property> <property> <name>dfs.namenode.servicerpc-address.mycluster.nn1</name> <value>hadoop1:53310</value> </property> <property> <name>dfs.namenode.servicerpc-address.mycluster.nn2</name> <value>hadoop5:53310</value> </property> <property> <name>dfs.ha.automatic-failover.enabled</name> <value>true</value> </property> <property> <name>dfs.namenode.shared.edits.dir</name> <value>qjournal://hadoop1:8485;hadoop2:8485;hadoop3:8485;hadoop4:8485;hadoop5:8485/mycluster</value> </property> <property> <name>dfs.journalnode.edits.dir</name> <value>/home/grid/hadoop-2.6.0/journal/data</value> </property> <property> <name>dfs.client.failover.proxy.provider.mycluster</name> <value>org.apache.hadoop.hdfs.server.namenode.ha.ConfiguredFailoverProxyProvider</value> </property> <property> <name>dfs.replication</name> <value>3</value> </property> <property> <name>dfs.namenode.name.dir</name> <value>file:///home/grid/hadoop-2.6.0/dfs/name</value> </property> <property> <name>dfs.datanode.data.dir</name> <value>file:///home/grid/hadoop-2.6.0/dfs/data</value> </property> <property> <name>dfs.webhdfs.enabled</name> <value>true</value> </property> <property> <name>dfs.permissions.enable</name> <value>false</value> </property> <property> <name>dfs.permissions</name> <value>false</value> </property> <property> <name>dfs.image.transfer.bandwidthPerSec</name> <value>1048576</value> </property> <property> <name>dfs.ha.fencing.methods</name> <value>shell(/bin/true)</value> </property> <!-- <property> <name>dfs.ha.fencing.ssh.private-key-files</name> <value>/home/grid/.ssh/id_rsa_nn</value> </property> --> </configuration> 在这个文件中需要说明的有两点: 第一、在官网上,我查看了hdfs-default.xml文件中,并没有参数dfs.ha.fencing.methods 这个参数,范围这个参数是在core-default.xml文件中有,那么按照官网上的意思是dfs.ha.fencing.methods 这个参数的配置是通过core-site.xml文件来设置的。但是实际上,这个参数是需要在hdfs-site.xml文件中设置的。否则就会出错,错误就是hadoop-daemon.sh start zkfc 启动DFSZKFailoverController进程时,无法启动。 第二、官网上都是通过设置下面两个参数来实现,出现故障时,通过哪种方式登录到另一个namenode上进行接管工作。如果采用下面的参数的话,我设置集群就会报错。显示错误信息的是无法连接,也就是梁一个namenode连接被拒绝了。 <property> <name>dfs.ha.fencing.methods</name> <value>sshfence</value> </property> <property> <name>dfs.ha.fencing.ssh.private-key-files</name> <value>/home/grid/.ssh/id_rsa_nn</value> </property> 所以,我换用了另一个值,如下: <property> <name>dfs.ha.fencing.methods</name> <value>shell(/bin/true)</value> </property> 此时针对HDFS的HA的配置工作已经完成,对应的yarn-site.xml和mapred-site.xml可以采用 http://www.cnblogs.com/ljy2013/articles/4345172.html 一文中的方式来设置即可。在这里我还设置了ResourceManager进行了热备。于是我的文件如下: (3)Yarn-site.xml文件的修改,该文件的配置对于不同的机器需要做出相应的修改工作。 <?xml version="1.0"?> <!-- Licensed 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. See accompanying LICENSE file. --> <configuration> <!-- Site specific YARN configuration properties --> <property> <name>yarn.resourcemanager.connect.retry-interval.ms</name> <value>60000</value> </property> <property> <name>yarn.resourcemanager.ha.enabled</name> <value>true</value> </property> <property> <name>yarn.resourcemanager.cluster-id</name> <value>rm-cluster</value> </property> <property> <name>yarn.resourcemanager.ha.rm-ids</name> <value>rm1,rm2</value> </property> <property> <name>yarn.resourcemanager.ha.id</name> //不同的节点只需要对这个参数做出相应的修改即可,也就是在热备的另一个节点上,该参数设置为rm2.即两个备份机器上的yarn-site.xml文件就是该参数不同。 <value>rm1</value> </property> <property> <name>yarn.resourcemanager.hostname.rm1</name> <value>hadoop1</value> </property> <property> <name>yarn.resourcemanager.hostname.rm2</name> <value>hadoop5</value> </property> <property> <name>yarn.resourcemanager.recovery.enabled</name> <value>true</value> </property> <property> <name>yarn.resourcemanager.store.class</name> <value>org.apache.hadoop.yarn.server.resourcemanager.recovery.ZKRMStateStore</value> </property> <property> <name>yarn.resourcemanager.zk-address</name> <value>hadoop1:2181,hadoop2:2181,hadoop3:2181,hadoop4:2181,hadoop5:2181</value> </property> <property> <name>yarn.resourcemanager.address.rm1</name> <value>${yarn.resourcemanager.hostname.rm1}:23140</value> </property> <property> <name>yarn.resourcemanager.scheduler.address.rm1</name> <value>${yarn.resourcemanager.hostname.rm1}:23130</value> </property> <property> <name>yarn.resourcemanager.webapp.https.address.rm1</name> <value>${yarn.resourcemanager.hostname.rm1}:23189</value> </property> <property> <name>yarn.resourcemanager.webapp.address.rm1</name> <value>${yarn.resourcemanager.hostname.rm1}:23188</value> </property> <property> <name>yarn.resourcemanager.resource-tracker.address.rm1</name> <value>${yarn.resourcemanager.hostname.rm1}:23125</value> </property> <property> <name>yarn.resourcemanager.admin.address.rm1</name> <value>${yarn.resourcemanager.hostname.rm1}:23141</value> </property> <property> <name>yarn.resourcemanager.address.rm2</name> <value>${yarn.resourcemanager.hostname.rm2}:23140</value> </property> <property> <name>yarn.resourcemanager.scheduler.address.rm2</name> <value>${yarn.resourcemanager.hostname.rm2}:23130</value> </property> <property> <name>yarn.resourcemanager.webapp.https.address.rm2</name> <value>${yarn.resourcemanager.hostname.rm2}:23189</value> </property> <property> <name>yarn.resourcemanager.webapp.address.rm2</name> <value>${yarn.resourcemanager.hostname.rm2}:23188</value> </property> <property> <name>yarn.resourcemanager.resource-tracker.address.rm2</name> <value>${yarn.resourcemanager.hostname.rm2}:23125</value> </property> <property> <name>yarn.resourcemanager.admin.address.rm2</name> <value>${yarn.resourcemanager.hostname.rm2}:23141</value> </property> <property> <name>yarn.nodemanager.local-dirs</name> <value>/home/hadoop/logs/yarn_local</value> </property> <property> <name>yarn.nodemanager.log-dirs</name> <value>/home/hadoop/logs/yarn_log</value> </property> <property> <name>yarn.nodemanager.remote-app-log-dir</name> <value>/home/hadoop/logs/yarn_remotelog</value> </property> <property> <name>yarn.log-aggregation-enable</name> <value>true</value> </property> <property> <name>yarn.nodemanager.resource.memory-mb</name> <value>2048</value> </property> <property> <name>yarn.nodemanager.vmem-pmem-ratio</name> <value>4.2</value> </property> <property> <name>yarn.nodemanager.resource.cpu-vcores</name> <value>2</value> </property> <property> <name>yarn.nodemanager.aux-services</name> <value>mapreduce_shuffle</value> </property> <property> <name>yarn.nodemanager.aux-services.mapreduce.shuffle.class</name> <value>org.apache.hadoop.mapred.ShuffleHandler</value> </property> </configuration> (4)mapred-site.xml文件的修改 <?xml version="1.0"?> <?xml-stylesheet type="text/xsl" href="configuration.xsl"?> <!-- Licensed 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. See accompanying LICENSE file. --> <!-- Put site-specific property overrides in this file. --> <configuration> <property> <name>mapreduce.framework.name</name> <value>yarn</value> </property> <property> <name>mapreduce.jobhistory.address</name> <value>hadoop1:10020,hadoop5:10020</value> </property> <property> <name>mapreduce.jobhistory.webapp.address</name> <value>hadoop1:19888,hadoop5:19888</value> </property> <property> <name>yarn.app.mapreduce.am.staging-dir</name> <value>/tmp/hadoop-yarn/staging</value> </property> <property> <name>mapreduce.jobhistory.done-dir</name> <value>${yarn.app.mapreduce.am.staging-dir}/history/done</value> </property> <property> <name>mapreduce.jobhistory.intermediate-done-dir</name> <value>${yarn.app.mapreduce.am.staging-dir}/history/done_intermediate</value> </property> <property> <name>mapreduce.task.io.sort.factor</name> <value>100</value> </property> <property> <name>mapreduce.reduce.shuffle.parallelcopies</name> <value>10</value> </property> </configuration> Ok了!至此,所有的配置文件修改工作都完成了。下面就是介绍一下如何启动了 5、启动 (1)首先启动zookeeper集群 由于我的节点是5个,所以我是将所有的节点都用来作为zookeeper来作为zookeeper的集群。因此在各节点上执行如下命令即可。 zkServer.sh start 所有的节点都启动zookeeper服务之后,zookeeper集群就已经启动了。 (2)对zookeeper集群进行格式化 hdfs zkfc -formatZK (3)启动JournalNode进程,注意这个在第一次的时候必须要按照这个顺序执行。否则后面hdfs格式化不了。 同样,我也是将所有的节点都作为了journalnode的节点,于是在所有的节点上执行下面的命令来启动journalnode。 hadoop-daemon.sh start journalnode (4)格式化hadoop的集群,注意,第一次格式化必须首先启动上面的journalnode进程。并且,hadoop格式化的执行在某一个namenode节点上进行,在这里我选择的是hadoop1上执行。 hdfs namenode -format mycluster (5)启动第(4)步格式化之后的namenode。 也就是说在第(4)步上面格式化后的namenode节点上启动namenode进程。 hadoop-daemon.sh start namenode (6)在另外一个namenode节点上按顺序执行如下两个命令来启动namenode进程。(本文中是hadoop5上执行) hdfs namenode -bootstrapStandby hadoop-daemon.sh start namenode (7)在一个namenode节点上执行一下两个命令启动所有的进程: start-dfs.sh start-yarn.sh (8)此时启动完之后,我们可以通过下面的命令来查看两个namenode的状态是否是standby或者是active hdfs haadmin -getServiceState nn1 standby hdfs haadmin -getServiceState nn2 active 这里的nn1和nn2就是上面的配置文件中所设置的。nn1对应的就是hadoop1,nn2对应的就是hadoop5。 6、检验自动切换,通过kill active的namenode来验证namenode是否能自动切换。 (1)通过上面步骤(8)查看到对应的那个namenode的状态是active,在该namenode节点上查看所有的进程。如下所示: (2)在active的namenode节点上,执行 kill -9 7048 。实际上这一步可以直接将该节点重启也可以。 (3)在standby的namenode节点上查看其状态的改变。 hdfs haadmin -getServiceState nn1 我们可以看到其对应的状态从standby的状态转变为active的状态了。 7、通过上传文件来检测HDFS的健康状态 执行 :hadoop fs -put /hadoop-2.6.0/etc/hadoop/hdfs-site.xml / 然后可以通过web查看hdfs-site.xml 8、测试在作业运行时,namendoe挂掉了是否能自动切换,并且还能正常执行作业? 准备一个2G的文件,我准备了一不电影zr.MP4,2.13G准备上传上HDFS中,在上传的过程中kill 掉active的namenode 查看最终的运行结果。 通过在在standby的namenode节点上执行:hadoop fs -put zr.mp4 / 。 在它执行的过程中,在active的namenode节点上执行:kill -9 7048 (这里的7048就是namenode进程ID)。在执行过程中,我们可以看到如下图所示: 通过上图可以看出,最终zr.mp4上传成功了。至此HDFS的HA方案已经完成。完全可以投入使用。