配置两个不同kerberos认证中心的集群间的互信

两个Hadoop集群开启Kerberos验证后，集群间不能够相互访问，需要实现Kerberos之间的互信，使用Hadoop集群A的客户端访问Hadoop集群B的服务（实质上是使用Kerberos Realm A上的Ticket实现访问Realm B的服务）。
先决条件：
1）两个集群(XDF.COM和HADOOP.COM)均开启Kerberos认证
2）Kerberos的REALM分别设置为XDF.COM和HADOOP.COM
步骤如下：

1 配置KDC之间的信任ticket

实现DXDF.COM和HADOOP.COM之间的跨域互信，例如使用XDF.COM的客户端访问HADOOP.COM中的服务，两个REALM需要共同拥有名为krbtgt/HADOOP.COM@XDF.COM的principal，两个Keys需要保证密码，version number和加密方式一致。默认情况下互信是单向的， HADOOP.COM的客户端访问XDF.COM的服务，两个REALM需要有krbtgt/XDF.COM@HADOOP.COM的principal。
向两个集群中添加krbtgt principal

  #XDF CLUSTER
  kadmin.local: addprinc –e “aes128-cts:normal des3-hmac-sha1:normal arcfour-hmac:normal camellia256-cts:normal camellia128-cts:normal des-hmac-sha1:normal des-cbc-md5:normal ” krbtgt/HADOOP.COM@XDF.COM
  kadmin.local: addprinc –e “aes128-cts:normal des3-hmac-sha1:normal arcfour-hmac:normal camellia256-cts:normal camellia128-cts:normal des-hmac-sha1:normal des-cbc-md5:normal ”  krbtgt/XDF.COM@HADOOP.COM

  #HADOOP CLUSTER
   kadmin.local: addprinc –e “aes128-cts:normal des3-hmac-sha1:normal arcfour-hmac:normal camellia256-cts:normal camellia128-cts:normal des-hmac-sha1:normal des-cbc-md5:normal ” krbtgt/HADOOP.COM@XDF.COM
   kadmin.local: addprinc –e “aes128-cts:normal des3-hmac-sha1:normal arcfour-hmac:normal camellia256-cts:normal camellia128-cts:normal des-hmac-sha1:normal des-cbc-md5:normal ”  krbtgt/XDF.COM@HADOOP.COM

要验证两个entries具有匹配的kvno和加密type，查看命令使用getprinc <principal_name>

kadmin.local:  getprinc  krbtgt/XDF.COM@HADOOP.COM
Principal: krbtgt/XDF.COM@HADOOP.COM
Expiration date: [never]
Last password change: Wed Jul 05 14:18:11 CST 2017
Password expiration date: [none]
Maximum ticket life: 1 day 00:00:00
Maximum renewable life: 30 days 00:00:00
Last modified: Wed Jul 05 14:18:11 CST 2017 (admin/admin@XDF.COM)
Last successful authentication: [never]
Last failed authentication: [never]
Failed password attempts: 0
Number of keys: 7
Key: vno 1, aes128-cts-hmac-sha1-96
Key: vno 1, des3-cbc-sha1
Key: vno 1, arcfour-hmac
Key: vno 1, camellia256-cts-cmac
Key: vno 1, camellia128-cts-cmac
Key: vno 1, des-hmac-sha1
Key: vno 1, des-cbc-md5
MKey: vno 1
Attributes:
Policy: [none]
kadmin.local:  getprinc  addprinc krbtgt/HADOOP.COM@XDF.COM
usage: get_principal [-terse] principal
kadmin.local:  getprinc  krbtgt/HADOOP.COM@XDF.COM
Principal: krbtgt/HADOOP.COM@XDF.COM
Expiration date: [never]
Last password change: Wed Jul 05 14:17:47 CST 2017
Password expiration date: [none]
Maximum ticket life: 1 day 00:00:00
Maximum renewable life: 30 days 00:00:00
Last modified: Wed Jul 05 14:17:47 CST 2017 (admin/admin@XDF.COM)
Last successful authentication: [never]
Last failed authentication: [never]
Failed password attempts: 0
Number of keys: 7
Key: vno 1, aes128-cts-hmac-sha1-96
Key: vno 1, des3-cbc-sha1
Key: vno 1, arcfour-hmac
Key: vno 1, camellia256-cts-cmac
Key: vno 1, camellia128-cts-cmac
Key: vno 1, des-hmac-sha1
Key: vno 1, des-cbc-md5
MKey: vno 1
Attributes:
Policy: [none]

2 在core-site中配置principal和user的映射RULES

 

Paste_Image.png

设置hadoop.security.auth_to_local参数，该参数用于将principal转变为user，一个需要注意的问题是SASL RPC客户端需要远程Server的Kerberos principal在本身的配置中匹配该principal。相同的pricipal name需要分配给源和目标cluster的服务，例如Source Cluster中的NameNode的kerbeors principal name为nn/h@XDF.COM，在Destination cluster中NameNode的pricipal设置为nn/h@HADOOP.COM（不能设置为nn2/h***@HADOOP.COM），例如：
在XDF Cluster和 HADOOP Cluster的core-site中增加：

<property>
<name>hadoop.security.auth_to_local</name>
<value>
RULE:[1:$1@$0](^.*@HADOOP\.COM$)s/^(.*)@HADOOP\.COM$/$1/g
RULE:[2:$1@$0](^.*@HADOOP\.COM$)s/^(.*)@HADOOP\.COM$/$1/g
RULE:[1:$1@$0](^.*@XDF\.COM$)s/^(.*)@XDF\.COM$/$1/g
RULE:[2:$1@$0](^.*@XDF\.COM$)s/^(.*)@XDF\.COM$/$1/g
DEFAULT
</value>
</property>

使用hadoop org.apache.hadoop.security.HadoopKerberosName <principal-name>来实现验证，例如:

[root@node1a141 ~]#  hadoop org.apache.hadoop.security.HadoopKerberosName hdfs/nodea1a141@XDF.COM
Name: hdfs/nodea1a141@XDF.COM to hdfs

3 在krb5.conf中配置信任关系

3.1 配置capaths

第一种方式是配置shared hierarchy of names，这个是默认及比较简单的方式，第二种方式是在krb5.conf文件中改变capaths，复杂但是比较灵活，这里采用第二种方式。
在两个集群的节点的/etc/krb5.conf文件配置domain和realm的映射关系，例如：在XDF cluster中配置：

[capaths]
       XDF.COM = {
              HADOOP.COM = .
       }

在HADOOP Cluster中配置：

 [capaths]
       HADOOP.COM = {
              XDF.COM = .
       }

配置成'.'是表示没有intermediate realms

3.2 配置realms

为了是XDF 可以访问HADOOP的KDC，需要将HADOOP的KDC Server配置到XDF cluster中，如下，反之相同：

 [realms]
  XDF.COM = {
    kdc = {host}.XDF.COM:88
    admin_server = {host}.XDF.COM:749
    default_domain = XDF.COM
  }
  HADOOP.COM = {
    kdc = {host}.HADOOP.COM:88
    admin_server = {host}.HADOOP.COM:749
    default_domain = HADOOP.COM
  }

3.3 配置domain_realm

在domain_realm中，一般配置成'.XDF.COM'和'XDF.COM'的格式，'.'前缀保证kerberos将所有的XDF.COM的主机均映射到XDF.COM realm。但是如果集群中的主机名不是以XDF.COM为后缀的格式，那么需要在domain_realm中配置主机与realm的映射关系，例XDF.nn.local映射为XDF.COM，需要增加XDF.nn.local = XDF.COM。

[domain_realm]
.hadoop.com=HADOOP.COM
 hadoop.com=HADOOP.COM
 .xdf.com=XDF.COM
 xdf.com=XDF.COM
 node1a141 = XDF.COM
 node1a143 = XDF.COM
 node1a210 = HADOOP.COM
 node1a202 = HADOOP.COM
 node1a203 = HADOOP.COM

重启kerberos服务

3.4 配置hdfs-site.xml

在hdfs-site.xml，设置允许的realms
在hdfs-site.xml中设置dfs.namenode.kerberos.principal.pattern为"*"

 

Paste_Image.png

这个是客户端的匹配规则用于控制允许的认证realms，如果该参数不配置，会有下面的异常：

java.io.IOException: Failed on local exception: java.io.IOException:
java.lang.IllegalArgumentException:
       Server has invalid Kerberosprincipal:nn/ HADOOP.COM@ XDF.COM;
       Host Details : local host is: "host1.XDF.COM/10.181.22.130";
                        destination host is: "host2.HADOOP.COM":8020;

4 测试

1)使用hdfs命令测试XDF 和HADOOP 集群间的数据访问，例如在XDF Cluster中kinit admin@XDF.CON，然后运行hdfs命令：

[root@node1a141 ~]# kdestroy
[root@node1a141 ~]# kinit admin
Password for admin@XDF.COM:
[root@node1a141 ~]# hdfs dfs -ls /
Found 3 items
drwxrwxrwx+  - hdfs supergroup          0 2017-06-13 15:13 /tmp
drwxrwxr-x+  - hdfs supergroup          0 2017-06-22 15:55 /user
drwxrwxr-x+  - hdfs supergroup          0 2017-06-14 14:11 /wa
[root@node1a141 ~]# hdfs dfs -ls hdfs://node1a202:8020/
Found 9 items
drwxr-xr-x   - root  supergroup          0 2017-05-27 18:55 hdfs://node1a202:8020/cdtest
drwx------   - hbase hbase               0 2017-05-22 18:51 hdfs://node1a202:8020/hbase
drwx------   - hbase hbase               0 2017-07-05 19:16 hdfs://node1a202:8020/hbase1
drwxr-xr-x   - hbase hbase               0 2017-05-11 10:46 hdfs://node1a202:8020/hbase2
drwxr-xr-x   - root  supergroup          0 2016-12-01 17:30 hdfs://node1a202:8020/home
drwxr-xr-x   - mdss  supergroup          0 2016-12-13 18:30 hdfs://node1a202:8020/idfs
drwxr-xr-x   - hdfs  supergroup          0 2017-05-22 18:51 hdfs://node1a202:8020/system
drwxrwxrwt   - hdfs  supergroup          0 2017-05-31 17:37 hdfs://node1a202:8020/tmp
drwxrwxr-x+  - hdfs  supergroup          0 2017-05-04 15:48 hdfs://node1a202:8020/user

在HADOOP.COM中进行相同的操作
2）运行distcp程序将XDF的数据复制到HADOOP集群，命令如下：

[root@node1a141 ~]# hadoop distcp hdfs://node1a141:8020/tmp/test.sh  hdfs://node1a202:8020/tmp/

5 附录

两集群的/etc/krb5.conf完整文件内容如下：

[root@node1a141 xdf]# cat /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 = XDF.COM
 dns_lookup_realm = false
 dns_lookup_kdc = false
 ticket_lifetime = 7d
 renew_lifetime = 30
 forwardable = true
 renewable=true
 #default_ccache_name = KEYRING:persistent:%{uid}

[realms]
 HADOOP.COM = {
   kdc = node1a198
   admin_server = node1a198
   default_realm = HADOOP.COM
   supported_enctypes = aes128-cts:normal des3-hmac-sha1:normal arcfour-hmac:normal camellia256-cts:normal camellia128-cts:normal des-hmac-sha1:normal des-cbc-md5:normal des-cbc-crc:normal
 }
 XDF.COM = {
   kdc = node1a141
   admin_server = node1a141
   default_realm = XDF.COM
   supported_enctypes = aes128-cts:normal des3-hmac-sha1:normal arcfour-hmac:normal camellia256-cts:normal camellia128-cts:normal des-hmac-sha1:normal des-cbc-md5:normal des-cbc-crc:normal
 }

[domain_realm]
 .hadoop.com=HADOOP.COM
 hadoop.com=HADOOP.COM
 .xdf.com=XDF.COM
 xdf.com=XDF.COM
 node1a141 = XDF.COM
 node1a143 = XDF.COM
 node1a210 = HADOOP.COM
 node1a202 = HADOOP.COM
 node1a203 = HADOOP.COM

[capaths]
XDF.COM = {
 HADOOP.COM = .
}