如何在CDH集群启用Kerberos

1.文档编写目的

---

本文档讲述如何在CDH集群启用及配置Kerberos，您将学习到以下知识：

1.如何安装及配置KDC服务

2.如何通过CDH启用Kerberos

3.如何登录Kerberos并访问Hadoop相关服务

文档主要分为以下几步：

1.安装及配置KDC服务

2.CDH集群启用Kerberos

3.Kerberos使用

这篇文档将重点介绍如何在CDH集群启用及配置Kerberos，并基于以下假设：

1.CDH集群运行正常

2.集群未启用Kerberos

3.MySQL 5.1.73

以下是本次测试环境，但不是本操作手册的必需环境：

1.操作系统：CentOS 6.5

2.CDH和CM版本为5.12.0

3.采用root用户进行操作

2.KDC服务安装及配置

---

本文档中将KDC服务安装在Cloudera Manager Server所在服务器上（KDC服务可根据自己需要安装在其他服务器）

1.在Cloudera Manager服务器上安装KDC服务

[root@ip-172-31-6-148~]# yum -y install krb5-serverkrb5-libs krb5-auth-dialog krb5-workstation

2.修改/etc/krb5.conf配置

[root@ip-172-31-6-148 fayson_r]# vim /etc/krb5.conf 
[logging]
 default = FILE:/var/log/krb5libs.log
 kdc = FILE:/var/log/krb5kdc.log
 admin_server = FILE:/var/log/kadmind.log

[libdefaults]
 default_realm = FAYSON.COM
 dns_lookup_realm = false
 dns_lookup_kdc = false
 ticket_lifetime = 24h
 renew_lifetime = 7d
 forwardable = true

[realms]
 FAYSON.COM = {
  kdc = ip-172-31-6-148.fayson.com
  admin_server = ip-172-31-6-148.fayson.com
 }

[domain_realm]
 .ip-172-31-6-148.fayson.com = FAYSON.COM
 ip-172-31-6-148.fayson.com = FAYSON.COM

标红部分为需要修改的信息。

3.修改/var/kerberos/krb5kdc/kadm5.acl配置

[root@ip-172-31-6-148~]# vim /var/kerberos/krb5kdc/kadm5.acl 
*/[email protected]    *

4.修改/var/kerberos/krb5kdc/kdc.conf配置

[root@ip-172-31-6-148 ~]# vim /var/kerberos/krb5kdc/kdc.conf 
[kdcdefaults]
 kdc_ports = 88
 kdc_tcp_ports = 88

[realms]
 FAYSON.COM= {
  #master_key_type = aes256-cts
  max_renewable_life= 7d 0h 0m 0s
  acl_file = /var/kerberos/krb5kdc/kadm5.acl
  dict_file = /usr/share/dict/words
  admin_keytab = /var/kerberos/krb5kdc/kadm5.keytab
  supported_enctypes = aes256-cts:normal aes128-cts:normal des3-hmac-sha1:normal arcfour-hmac:normal des-hmac-sha1:n
ormal des-cbc-md5:normal des-cbc-crc:normal
 }

标红部分为需要修改的配置。

5.创建Kerberos数据库

[root@ip-172-31-6-148 ~]# kdb5_util create –r FAYSON.COM -s
Loading random data
Initializing database '/var/kerberos/krb5kdc/principal' for realm 'FAYSON.COM',
master key name 'K/[email protected]'
You will be prompted for the database Master Password.
It is important that you NOT FORGET this password.
Enter KDC database master key: 
Re-enter KDC database master key to verify: 

此处需要输入Kerberos数据库的密码。

6.创建Kerberos的管理账号

[root@ip-172-31-6-148 ~]# kadmin.local
Authenticating as principal fayson/[email protected] with password.
kadmin.local:  addprinc admin/[email protected]
WARNING: no policy specified for admin/[email protected]; defaulting to no policy
Enter password for principal "admin/[email protected]": 
Re-enter password for principal "admin/[email protected]": 
Principal "admin/[email protected]" created.
kadmin.local:  exit
[root@ip-172-31-6-148 ~]# 

标红部分为Kerberos管理员账号，需要输入管理员密码。

7.将Kerberos服务添加到自启动服务，并启动krb5kdc和kadmin服务

[root@ip-172-31-6-148~]# chkconfig krb5kdc on
[root@ip-172-31-6-148 ~]# chkconfig kadmin on
[root@ip-172-31-6-148 ~]# service krb5kdc start
Starting Kerberos 5 KDC:                                   [  OK  ]
[root@ip-172-31-6-148 ~]# service kadmin start
Starting Kerberos 5 Admin Server:                          [  OK  ]
[root@ip-172-31-6-148 ~]# 

8.测试Kerberos的管理员账号

[root@ip-172-31-6-148 ~]# kinit admin/[email protected]
Password for admin/[email protected]: 
[root@ip-172-31-6-148 ~]# klist
Ticket cache: FILE:/tmp/krb5cc_0
Default principal: admin/[email protected]

Valid starting     Expires            Service principal
09/05/17 16:39:17  09/06/17 16:39:17  krbtgt/[email protected]
        renew until 09/12/17 16:39:17
[root@ip-172-31-6-148 ~]# 

9.为集群安装所有Kerberos客户端，包括Cloudera Manager

[root@ip-172-31-6-148 cdh-shell-master]# yum -y install krb5-libs krb5-workstation

10.在Cloudera Manager Server服务器上安装额外的包

[root@ip-172-31-6-148cdh-shell-master]# yum -y install openldap-clients

11.将KDC Server上的krb5.conf文件拷贝到所有Kerberos客户端

[root@ip-172-31-6-148cdh-shell-master]# scp -r /etc/krb5.conf [email protected]:/etc/

此处使用脚本进行拷贝

[root@ip-172-31-6-148cdh-shell-master]# sh b.sh node.list /etc/krb5.conf  /etc/
krb5.conf                                                                         100% 451     0.4KB/s   00:00   
krb5.conf                                                                         100% 451     0.4KB/s   00:00   
krb5.conf                                                                         100% 451     0.4KB/s   00:00   
krb5.conf                                                                          100%  451     0.4KB/s  00:00    
[root@ip-172-31-6-148 cdh-shell-master]# 

3.CDH集群启用Kerberos

---

1.在KDC中给Cloudera Manager添加管理员账号

[root@ip-172-31-6-148 cdh-shell-bak]# kadmin.local
Authenticating as principal admin/[email protected] with password.
kadmin.local:  addprinc cloudera-scm/[email protected]
WARNING: no policy specified for cloudera-scm/[email protected]; defaulting to no policy
Enter password for principal "cloudera-scm/[email protected]": 
Re-enter password for principal "cloudera-scm/[email protected]": 
Principal "cloudera-scm/[email protected]" created.
kadmin.local:  exit
[root@ip-172-31-6-148 cdh-shell-bak]# 

2.进入Cloudera Manager的“管理”-> “安全”界面

3.选择“启用Kerberos”，进入如下界面

确保如下列出的所有检查项都已完成

4.点击“继续”，配置相关的KDC信息，包括类型、KDC服务器、KDC Realm、加密类型以及待创建的Service Principal（hdfs，yarn,，hbase，hive等）的更新生命期等

5.点击“继续”

6.不建议让Cloudera Manager来管理krb5.conf, 点击“继续”

7.输入Cloudera Manager的Kerbers管理员账号，必须和之前创建的账号一致，点击“继续”

8.等待启用Kerberos完成，点击“继续”

9.点击“继续”

10.勾选重启集群，点击“继续”

11.等待集群重启成功，点击“继续”

至此已成功启用Kerberos。

4.Kerberos使用

---

使用fayson用户运行MapReduce任务及操作Hive，需要在集群所有节点创建fayson用户。

1.使用kadmin创建一个fayson的principal

[root@ip-172-31-6-148 cdh-shell-bak]# kadmin.local
Authenticating as principal admin/[email protected] with password.
kadmin.local:  addprinc [email protected]
WARNING: no policy specified for [email protected]; defaulting to no policy
Enter password for principal "[email protected]":
Re-enter password for principal "[email protected]":
Principal "[email protected]" created.
kadmin.local:  exit
[root@ip-172-31-6-148 cdh-shell-bak]#

2.使用fayson用户登录Kerberos

[root@ip-172-31-6-148 cdh-shell-bak]# kdestroy
[root@ip-172-31-6-148 cdh-shell-bak]# kinit fayson
Password for [email protected]: 
[root@ip-172-31-6-148 cdh-shell-bak]# klist
Ticket cache: FILE:/tmp/krb5cc_0
Default principal: [email protected]

Valid starting     Expires            Service principal
09/05/17 17:19:08  09/06/17 17:19:08  krbtgt/[email protected]
        renew until 09/12/17 17:19:08
[root@ip-172-31-6-148 cdh-shell-bak]# 

3.运行MapReduce作业

[root@ip-172-31-6-148~]# hadoop jar /opt/cloudera/parcels/CDH/lib/hadoop-0.20-mapreduce/hadoop-examples.jar pi 10 1
...
Starting Job
17/09/02 20:10:43 INFO mapreduce.Job: Running job: job_1504383005209_0001
17/09/02 20:10:56 INFO mapreduce.Job: Job job_1504383005209_0001 running in ubermode : false
17/09/02 20:10:56 INFO mapreduce.Job:  map0% reduce 0%
17/09/02 20:11:09 INFO mapreduce.Job:  map20% reduce 0%
17/09/02 20:11:12 INFO mapreduce.Job:  map40% reduce 0%
17/09/02 20:11:13 INFO mapreduce.Job:  map50% reduce 0%
17/09/02 20:11:15 INFO mapreduce.Job:  map60% reduce 0%
17/09/02 20:11:16 INFO mapreduce.Job:  map70% reduce 0%
17/09/02 20:11:19 INFO mapreduce.Job:  map80% reduce 0%
17/09/02 20:11:21 INFO mapreduce.Job:  map100% reduce 0%
17/09/02 20:11:26 INFO mapreduce.Job:  map100% reduce 100%
17/09/02 20:11:26 INFO mapreduce.Job: Job job_1504383005209_0001 completedsuccessfully

4.使用beeline连接hive进行测试

[root@ip-172-31-6-148 cdh-shell-bak]# beeline 
Beeline version 1.1.0-cdh5.12.1 by Apache Hive
beeline> !connect jdbc:hive2://localhost:10000/;principal=hive/[email protected]
...
Transaction isolation: TRANSACTION_REPEATABLE_READ
0: jdbc:hive2://localhost:10000/> show tables;
...
INFO  : OK
+-------------+--+
|  tab_name   |
+-------------+--+
| test_table  |
+-------------+--+
1 row selected (0.194 seconds)
0: jdbc:hive2://localhost:10000/> select * from test_table;
...
INFO  : OK
+----------------+----------------+--+
| test_table.s1  | test_table.s2  |
+----------------+----------------+--+
| 4              | lisi           |
| 1              | test           |
| 2              | fayson         |
| 3              | zhangsan       |
+----------------+----------------+--+
4 rows selected (0.144 seconds)
0: jdbc:hive2://localhost:10000/> 

运行Hive MapReduce作业

0: jdbc:hive2://localhost:10000/> select count(*) from test_table;
...
INFO  : OK
+------+--+
| _c0  |
+------+--+
| 4    |
+------+--+
1 row selected (35.779 seconds)
0: jdbc:hive2://localhost:10000/> 

5.常见问题

---

1.使用Kerberos用户身份运行MapReduce作业报错

main : run as user is fayson
main : requested yarn user is fayson
Requested user fayson is not whitelisted and has id 501,whichis below the minimum allowed 1000

Failing this attempt. Failing the application.
17/09/02 20:05:04 INFO mapreduce.Job: Counters: 0
Job Finished in 6.184 seconds
java.io.FileNotFoundException: File does not exist:hdfs://ip-172-31-6-148:8020/user/fayson/QuasiMonteCarlo_1504382696029_1308422444/out/reduce-out
        at org.apache.hadoop.hdfs.DistributedFileSystem$20.doCall(DistributedFileSystem.java:1266)
        at org.apache.hadoop.hdfs.DistributedFileSystem$20.doCall(DistributedFileSystem.java:1258)
        at org.apache.hadoop.fs.FileSystemLinkResolver.resolve(FileSystemLinkResolver.java:81)
        at org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:1258)
        at org.apache.hadoop.io.SequenceFile$Reader.(SequenceFile.java:1820)
        at org.apache.hadoop.io.SequenceFile$Reader.(SequenceFile.java:1844)
        at org.apache.hadoop.examples.QuasiMonteCarlo.estimatePi(QuasiMonteCarlo.java:314)
        at org.apache.hadoop.examples.QuasiMonteCarlo.run(QuasiMonteCarlo.java:354)
        at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70)
        at org.apache.hadoop.examples.QuasiMonteCarlo.main(QuasiMonteCarlo.java:363)
        at sun.reflect.NativeMethodAccessorImpl.invoke0(NativeMethod)
        at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:606)
        at org.apache.hadoop.util.ProgramDriver$ProgramDescription.invoke(ProgramDriver.java:71)
        at org.apache.hadoop.util.ProgramDriver.run(ProgramDriver.java:144)
        at org.apache.hadoop.examples.ExampleDriver.main(ExampleDriver.java:74)
        at sun.reflect.NativeMethodAccessorImpl.invoke0(NativeMethod)
        at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:57)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:606)
        atorg.apache.hadoop.util.RunJar.run(RunJar.java:221)
        at org.apache.hadoop.util.RunJar.main(RunJar.java:136)

问题原因：是由于Yarn限制了用户id小于10000的用户提交作业；

解决方法：修改Yarn的min.user.id来解决