• Hadoop常见问题

     报错1 ：is group-writable, and the group is not root.  Its permissions are 0775,

    datanode启动时，日志报错

    1.“xxxx” is group-writable, and the group is not root.  Its permissions are 0775, and it is owned by gid 3245.  Please fix this or select a different socket path.

    从报错可以看出，hadoop的目录结构中，权限不对，重置下目录权限

    chown 755 hadoop/    再启动datanode，查查日志，又是

    xxx  is group-writable, and the group is not root.  Its permissions are 0775, and it is owned by gid 3245.  Please fix this or select a different socket path

    没办法，换了个目录生成socket文件，并设置权限为755，datanode启动正常

    报错2：we cannot start a localDataXceiverServer because libhadoop cannot be loaded.

    java.lang.RuntimeException: Although a UNIX domain socket path is configured as /app/log4x/apps/hadoop/etc/DN_PORT, we cannot start a localDataXceiverServer because libhadoop cannot be loaded.

     1.检查主机情况

    1. # hadoop checknative -a
    2. 23/11/07 21:13:08 DEBUG util.NativeCodeLoader: Trying to load the custom-built native-hadoop library...
    3. 23/11/07 21:13:08 DEBUG util.NativeCodeLoader: Failed to load native-hadoop with error: java.lang.UnsatisfiedLinkError: no hadoop in java.library.path
    4. 23/11/07 21:13:08 DEBUG util.NativeCodeLoader: java.library.path=:/app/log4x/apps/hadoop/lib/native/Linux-amd64-64/:/usr/java/packages/lib/amd64:/usr/lib64:/lib64:/lib:/usr/lib
    5. 23/11/07 21:13:08 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
    6. 23/11/07 21:13:08 DEBUG util.Shell: setsid exited with exit code 0
    7. Native library checking:
    8. hadoop:  false 
    9. zlib:    false 
    10. snappy:  false 
    11. lz4:     false 
    12. bzip2:   false 
    13. openssl: false 
    14. 23/11/07 21:13:08 INFO util.ExitUtil: Exiting with status 1

    2.新加配置

    在~/.bash_profile文件中，新增配置,指定nvtive的路径

    1. export JAVA_LIBRARY_PATH=/app/log4x/apps/hadoop/lib/native
    2.  
    3. source ~/.bash_porfile
    4.  
    5. # hadoop checknative -a
    6. 23/11/08 16:06:00 WARN bzip2.Bzip2Factory: Failed to load/initialize native-bzip2 library system-native, will use pure-Java version
    7. 23/11/08 16:06:00 INFO zlib.ZlibFactory: Successfully loaded & initialized native-zlib library
    8. Native library checking:
    9. hadoop:  true /app/log4x/apps/hadoop/lib/native/libhadoop.so
    10. zlib:    true /lib64/libz.so.1
    11. snappy:  true /lib64/libsnappy.so.1
    12. lz4:     true revision:99
    13. bzip2:   false 
    14. openssl: false Cannot load libcrypto.so (libcrypto.so: cannot open shared object file: No such file or directory)!
    15. 23/11/08 16:06:00 INFO util.ExitUtil: Exiting with status 1

    报错3：IPC's epoch 1 is not the current writer epoch  0

    org.apache.hadoop.hdfs.qjournal.client.QuorumException: Got too many exceptions to achieve quorum size 3/5. 4 exceptions thrown:
    10.255.33.120:43001: IPC's epoch 1 is not the current writer epoch  0

    奇怪了，我也报出这样的问题，看了别人的解决思路

     １,先把报错关键信息　"IPC's epoch  is less than the last promised epoch"　贴到google上查了一下,大部分外国人的回答都是因为网络原因引起的.
        ２,据上,经过看日志,每次启动另一个namenode的时候都会去探测三个 journalnode服务的8485端口,提示是faild的,
            说明最有可能是网络问题,排查如下:
            ifconfig -a看网卡是否有丢包,
            查看/etc/sysconfig/selinux　配置 SELINUX=disabled　是否是对的,
            /etc/init.d/iptables status　　查看防火墙是否运行,因为我们hadoop是运行内网环境,记得之前部署的时候,防火墙是关闭的,　看来问题找到了
            /etc/init.d/iptables stop
            先后检查了,三个 journalnode服务器的防火墙,都莫名其妙的启着的,马上关闭
            再重新启动两个namenode,查看日志,正常了,

    我检查了下我的这边：

    1.SELINUX都是disabled

    1. # ansible -i hosts tt -m shell -a "sudo cat /etc/sysconfig/selinux  | grep SELINUX"
    2. [WARNING]: Consider using 'become', 'become_method', and 'become_user' rather than running sudo
    3. host_121 | CHANGED | rc=0 >>
    4. # SELINUX= can take one of these three values:
    5. SELINUX=disabled
    6. # SELINUXTYPE= can take one of three two values:
    7. SELINUXTYPE=targeted 
    8. host_118 | CHANGED | rc=0 >>
    9. # SELINUX= can take one of these three values:
    10. SELINUX=disabled
    11. # SELINUXTYPE= can take one of three two values:
    12. SELINUXTYPE=targeted 
    13. host_119 | CHANGED | rc=0 >>
    14. # SELINUX= can take one of these three values:
    15. SELINUX=disabled
    16. # SELINUXTYPE= can take one of three two values:
    17. SELINUXTYPE=targeted 
    18. host_122 | CHANGED | rc=0 >>
    19. # SELINUX= can take one of these three values:
    20. SELINUX=disabled
    21. # SELINUXTYPE= can take one of three two values:
    22. SELINUXTYPE=targeted 
    23. host_120 | CHANGED | rc=0 >>
    24. # SELINUX= can take one of these three values:
    25. SELINUX=disabled
    26. # SELINUXTYPE= can take one of three two values:
    27. SELINUXTYPE=targeted 
    28. host_126 | CHANGED | rc=0 >>
    29. # SELINUX= can take one of these three values:
    30. SELINUX=disabled
    31. # SELINUXTYPE= can take one of three two values:
    32. SELINUXTYPE=targeted

    2.查看防火墙状态

            有关着的也有起着的。我以为是防火墙的问题，但看了另一套hadoop的环境信息，都是在相通主机上，用户名不通而已，不可能只对当前用户不起作用。 

    1. # ansible -i hosts tt -m shell -a "systemctl status iptables |grep  Active:"
    2. host_118 | CHANGED | rc=0 >>
    3.    Active: failed (Result: exit-code) since ? 2022-04-26 11:04:51 CST; 1 years 6 months ago
    4. host_121 | CHANGED | rc=0 >>
    5.    Active: active (exited) since ? 2022-04-26 11:05:40 CST; 1 years 6 months ago
    6. host_119 | CHANGED | rc=0 >>
    7.    Active: active (exited) since ? 2022-04-26 10:59:41 CST; 1 years 6 months ago
    8. host_122 | CHANGED | rc=0 >>
    9.    Active: active (exited) since ? 2021-03-18 17:13:00 CST; 2 years 7 months ago
    10. host_120 | CHANGED | rc=0 >>
    11.    Active: failed (Result: exit-code) since ? 2022-04-26 11:05:06 CST; 1 years 6 months ago
    12. host_126 | CHANGED | rc=0 >>
    13.    Active: active (exited) since ? 2021-03-18 15:58:10 CST; 2 years 7 months ago

    3.查看 journalnode日志

    2023-11-08 11:46:08,640 INFO org.apache.hadoop.hdfs.server.common.Storage: Formatting journal Storage Directory /app/log4x/apps/hadoop/jn/log4x-hcluster with nsid: 1531841238
    2023-11-08 11:46:08,642 INFO org.apache.hadoop.hdfs.server.common.Storage: Lock on /app/log4x/apps/hadoop/jn/log4x-hcluster/in_use.lock acquired by nodename 3995@hkcrmlog04
    2023-11-08 11:49:37,343 INFO org.apache.hadoop.hdfs.qjournal.server.Journal: Updating lastPromisedEpoch from 0 to 1 for client /10.255.33.121
    2023-11-08 11:49:37,345 INFO org.apache.hadoop.ipc.Server: IPC Server handler 2 on 43001, call org.apache.hadoop.hdfs.qjournal.protocol.QJournalProtocol.journal from 10.255.33.121:51086 Call#65 Retry#0
    java.io.IOException: IPC's epoch 1 is not the current writer epoch  0
            at org.apache.hadoop.hdfs.qjournal.server.Journal.checkWriteRequest(Journal.java:445)
            at org.apache.hadoop.hdfs.qjournal.server.Journal.journal(Journal.java:342)
            at org.apache.hadoop.hdfs.qjournal.server.JournalNodeRpcServer.journal(JournalNodeRpcServer.java:148)
            at org.apache.hadoop.hdfs.qjournal.protocolPB.QJournalProtocolServerSideTranslatorPB.journal(QJournalProtocolServerSideTranslatorPB.java:158)
            at org.apache.hadoop.hdfs.qjournal.protocol.QJournalProtocolProtos$QJournalProtocolService$2.callBlockingMethod(QJournalProtocolProtos.java:25421)
            at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:619)
            at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:975)
            at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2040)
            at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2036)
            at java.security.AccessController.doPrivileged(Native Method)
            at javax.security.auth.Subject.doAs(Subject.java:422)
            at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1692)
            at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2034)

    确实更防火墙没关系，format了namenode节点后，根据hdfs-site.xml文件中配置的sockt路径，创建了sock。

    1. # ll
    2. total 4
    3. srwxr-xr-x 1 log4x log4x    0 Nov  8 15:05 DN_PORT
    4. drwxr-xr-x 2 log4x log4x 4096 Nov  8 14:47 hadoop
    5. [log4x@hkcrmlog04 etc]$ pwd
    6. /app/log4x/apps/hadoop/etc
    7. [log4x@hkcrmlog04 etc]$ grep -ir 'DN_PORT' hadoop/
    8. hadoop/hdfs-site.xml.bajk:              /app/ailog4x/hadoop/etc/DN_PORT
    9. hadoop/hdfs-site.xml.1107bak:           /app/log4x/apps/hadoop/etc/DN_PORT
    10. hadoop/hdfs-site.xml:           /app/log4x/apps/hadoop/etc/DN_PORT

    4.说明

    4.1 nc -Ul DN_PORT  创建为socket 。若命令不存在，yum install -y nc安装即可

    4.2 chmod 666 DN_PORT  权限给到666

    4.2 chmod -R 755 hadoop/   整个hadoop的权限给到755

    报错4：Operation category JOURNAL is not supported in state standby

     Operation category JOURNAL is not supported in state standby。。。

     Call From hkcrmlog03/10.255.33.120 to hkcrmlog04:40101 failed on connection exception: java.net.ConnectException: Connection refused; For more details see:  http://wiki.apache.org/hadoop/ConnectionRefused
            at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
            at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62)
            at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
            at java.lang.reflect.Constructor.newInstance(Constructor.java:423)
            at org.apache.hadoop.net.NetUtils.wrapWithMessage(NetUtils.java:791)
            at org.apache.hadoop.net.NetUtils.wrapException(NetUtils.java:731)
            at org.apache.hadoop.ipc.Client.call(Client.java:1474)
            at org.apache.hadoop.ipc.Client.call(Client.java:1401)
            at org.apache.hadoop.ipc.ProtobufRpcEngine$Invoker.invoke(ProtobufRpcEngine.java:232)
            at com.sun.proxy.$Proxy15.rollEditLog(Unknown Source)
            at org.apache.hadoop.hdfs.protocolPB.NamenodeProtocolTranslatorPB.rollEditLog(NamenodeProtocolTranslatorPB.java:148)
            at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer.triggerActiveLogRoll(EditLogTailer.java:271)
            at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer.access$600(EditLogTailer.java:61)
            at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread.doWork(EditLogTailer.java:313)
            at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread.access$200(EditLogTailer.java:282)
            at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread$1.run(EditLogTailer.java:299)
            at org.apache.hadoop.security.SecurityUtil.doAsLoginUserOrFatal(SecurityUtil.java:412)
            at org.apache.hadoop.hdfs.server.namenode.ha.EditLogTailer$EditLogTailerThread.run(EditLogTailer.java:295)
    Caused by: java.net.ConnectException: Connection refused
            at sun.nio.ch.SocketChannelImpl.checkConnect(Native Method)
            at sun.nio.ch.SocketChannelImpl.finishConnect(SocketChannelImpl.java:715)
            at org.apache.hadoop.net.SocketIOWithTimeout.connect(SocketIOWithTimeout.java:206)
            at org.apache.hadoop.net.NetUtils.connect(NetUtils.java:530)
            at org.apache.hadoop.net.NetUtils.connect(NetUtils.java:494)
            at org.apache.hadoop.ipc.Client$Connection.setupConnection(Client.java:609)
            at org.apache.hadoop.ipc.Client$Connection.setupIOstreams(Client.java:707)
            at org.apache.hadoop.ipc.Client$Connection.access$2800(Client.java:370)
            at org.apache.hadoop.ipc.Client.getConnection(Client.java:1523)
            at org.apache.hadoop.ipc.Client.call(Client.java:1440)

    1.查看状态

    1. # bin/hdfs haadmin -getServiceState l4xnn2
    2. standby
    3. [log4x@hkcrmlog03 hadoop]$ 
    4. [log4x@hkcrmlog03 hadoop]$ bin/hdfs haadmin -getServiceState l4xnn1
    5. standby
    6.  
    7. # bin/hdfs haadmin -transitionToActive --forcemanual l4xnn1
    8. You have specified the forcemanual flag. This flag is dangerous, as it can induce a split-brain scenario that WILL CORRUPT your HDFS namespace, possibly irrecoverably.
    9.  
    10. It is recommended not to use this flag, but instead to shut down the cluster and disable automatic failover if you prefer to manually manage your HA state.
    11.  
    12. You may abort safely by answering 'n' or hitting ^C now.
    13.  
    14. Are you sure you want to continue? (Y or N) y
    15. 23/11/08 15:46:00 WARN ha.HAAdmin: Proceeding with manual HA state management even though
    16. automatic failover is enabled for NameNode at /10.255.33.121:40101
    17. 23/11/08 15:46:00 WARN ha.HAAdmin: Proceeding with manual HA state management even though
    18. automatic failover is enabled for NameNode at /10.255.33.120:40101
    19. [log4x@hkcrmlog03 hadoop]$ 
    20. [log4x@hkcrmlog03 hadoop]$ 
    21. [log4x@hkcrmlog03 hadoop]$ 
    22. [log4x@hkcrmlog03 hadoop]$ bin/hdfs haadmin -getServiceState l4xnn1             
    23. active
    24. [log4x@hkcrmlog03 hadoop]$ bin/hdfs haadmin -getServiceState l4xnn2             
    25. standby

    namenode的日志已正常。

    报错5：Error replaying edit log at offset 0.  Expected transaction ID was 1

    org.apache.hadoop.hdfs.server.namenode.EditLogInputException: Error replaying edit log at offset 0.  Expected transaction ID was 1

    在节点上从新执行了

    # $HADOOP_PREFIX/bin/hdfs namenode -format

    在启动namenode

    # $HADOOP_PREFIX/sbin/hadoop-daemon.sh --script hdfs start namenode
    1. $ bin/hdfs haadmin -getServiceState l4xnn1
    2. active
    3. [log4x@hkcrmlog03 hadoop]$ 
    4. [log4x@hkcrmlog03 hadoop]$ bin/hdfs haadmin -getServiceState l4xnn2
    5. standby

    namenode已正常。

    报错6：journal Storage Directory

    journal Storage Directory /app/log4x/apps/hadoop/jn/log4x-hcluster: NameNode

    大概为journalnode保存的元数据和namenode的不一致，导致，3台机器中有2台报了这个错误。

    在nn1上启动journalnode，再执行hdfs namenode -initializeSharedEdits，使得journalnode与namenode保持一致。再重新启动namenode就没有问题了。

    在查看2个namenode的状态。

    报错7：Decided to synchronize log to startTxId: 1

     Decided to synchronize log to startTxId: 1  namenode.RedundantEditLogInputStream$PrematureEOFException: got premature end-of-file at txid 0; expected file to go up to 2

     namenode元数据被破坏，需要修复

    解决恢复一下namenode
      

    hadoop namenode –recover

  • 相关阅读:
    （设计模式）工厂模式
    Opencv算术操作
    python、ruby、go、java写的端口扫描工具
    密码学系列之八：密码协议
    记录一次dma_alloc_coherent失败的解决
    Java架构师缓存通用设计方案
    Haproxy搭建Web群集
    202435读书笔记|《半小时漫画中国史》——读点经济学与历史，生活更美好，趣味烧脑土地制度、商鞅变法、华丽丽的丝绸之路这里都有
    计算机组成原理习题课第一章-1（唐朔飞）
    PYTHON
  • 原文地址：https://blog.csdn.net/red_sky_blue/article/details/134276082