标签:datanode 比较 ret htm async already org ogg war
hadoop集群搭建了ha,初次启动正常,最近几天启动时偶尔发现,namenode1节点启动后一段时间(大约10几秒-半分钟左右),namenode1上namenode进程停掉,查看日志:
1 2017-08-28 21:54:37,617 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: slave1/192.168.128.136:8485. Already tried 9 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS) 2 2017-08-28 21:54:37,629 WARN org.apache.hadoop.hdfs.server.namenode.FSEditLog: Unable to determine input streams from QJM to [192.168.128.135:8485, 192.168.128.136:8485, 192.168.128.137:8485]. Skipping. 3 org.apache.hadoop.hdfs.qjournal.client.QuorumException: Got too many exceptions to achieve quorum size 2/3. 1 successful responses: 4 192.168.128.137:8485: [[1557,1558], [1559,1560], [1561,1562], [1563,1564], [1565,1566], [1567,1568], [1569,1570], [1571,1572], [1573,1574], [1575,1576], [1577,1578], [1579,1580], [1581,1582], [1583,1584], [1585,1586], [1587,1588], [1589,1589], [1590,1591]] 5 2 exceptions thrown: 6 192.168.128.136:8485: Call From master/192.168.128.135 to slave1:8485 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused 7 192.168.128.135:8485: Call From master/192.168.128.135 to master:8485 failed on connection exception: java.net.ConnectException: Connection refused; For more details see: http://wiki.apache.org/hadoop/ConnectionRefused 8 at org.apache.hadoop.hdfs.qjournal.client.QuorumException.create(QuorumException.java:81) 9 at org.apache.hadoop.hdfs.qjournal.client.QuorumCall.rethrowException(QuorumCall.java:223) 10 at org.apache.hadoop.hdfs.qjournal.client.AsyncLoggerSet.waitForWriteQuorum(AsyncLoggerSet.java:142) 11 at org.apache.hadoop.hdfs.qjournal.client.QuorumJournalManager.selectInputStreams(QuorumJournalManager.java:471)
一、查阅资料后找到原因:
我是用start-al.sh启动的集群,journalnode(端口8485)是在namenode后启动的。默认情况下namenode启动10s(maxRetries=10, sleepTime=1000)后journalnode还没有启动,就会报上述错误。
二、解决方案:
1. 修改core-site.xml中的ipc参数
1 <!--修改core-site.xml中的ipc参数,防止出现连接journalnode服务ConnectException--> 2 <property> 3 <name>ipc.client.connect.max.retries</name> 4 <value>100</value> 5 <description>Indicates the number of retries a client will make to establish a server connection.</description> 6 </property> 7 <property> 8 <name>ipc.client.connect.retry.interval</name> 9 <value>10000</value> 10 <description>Indicates the number of milliseconds a client will wait for before retrying to establish a server connection.</description> 11 </property>
注意:仅对于这种由于服务没有启动完成造成连接超时的问题,都可以调整core-site.xml中的ipc参数来解决。如果目标服务本身没有启动成功,这边调整ipc参数是无效的。
2. 手动分步启动
1 #启动hadfs,注意有的是在多个节点执行的。 2 hadoop-daemons.sh start journalnode 3 hadoop-daemon.sh start namenode #每个namenode都要执行 4 hadoop-daemon.sh start zkfc #每个namenode都要执行 5 hadoop-daemons.sh start datanode 6 #启动yarn 7 start-yarn.sh
分步启动集群的方式,因为journalnode是在namenode之前启动的,所以正常情况下一次就会连接成功,不会重试多次。
三、错误再次分析
由于部署好ha后,首次启动我是分步启动的,没有遇到该问题。之后都是start-all.sh启动,大约70%情况下会有该问题,30%左右的启动是正常的,究其原因,我想70%的时候journalnode启动比较慢,另有个别时候是启动比较快。实测中确实发现集群主机刚刚开机,就启动hadoop,会比较慢;等一段时间或者首次启动hadoop后停止,然后再重新启动,这两种情况下hadoop启动会比较快。我是虚拟机,通常会第一时间启动hadoop,所以遇到这个坑的时候比较多。当然,找到了根本原因,无论hadoop启动快慢namenode都不会挂掉了。
另外namenode启动后有退出有多种原因,本文只针对启动的一种,具体情况需要查看日志并寻找合适解决方案。
致谢:http://www.linuxidc.com/Linux/2016-03/129437.htm
标签:datanode 比较 ret htm async already org ogg war
原文地址:http://www.cnblogs.com/tibit/p/7447190.html