Setup cross realm trust between two MIT KDC

[#1]

How to setup cross realm trust between two MIT KDC – In this post, we will see how to setup cross realm trust between two MIT KDC. We can access and copy data from one cluster to another if the cross realm trust is setup correctly.

.

In our example, we have 2 clusters with same HDP version(2.4.2.0) and Ambari version(2.2.2.0)

.

Cluster 1:

172.26.68.47 hwx-1.hwx.com  hwx-1
172.26.68.46 hwx-2.hwx.com  hwx-2
172.26.68.45 hwx-3.hwx.com  hwx-3

Cluster 2:

172.26.68.48 support-1.support.com   support-1
172.26.68.49 support-2.support.com   support-2
172.26.68.50 support-3.support.com   support-3

.

Below are the steps:

.

Step 1: Make sure both the clusters are kerberized with MIT KDC. You can use below automated script for configuring Kerberos on HDP.

https://community.hortonworks.com/articles/29203/automated-kerberos-installation-and-configuration.h...

.

Step 2: Please configure /etc/hosts file on both the clusters to have Ip <-> hostname mappings.

Example:

On both clusters /etc/hosts file should look like below:

172.26.68.47 hwx-1.hwx.com  hwx-1
172.26.68.46 hwx-2.hwx.com  hwx-2
172.26.68.45 hwx-3.hwx.com  hwx-3
172.26.68.48 support-1.support.com   support-1
172.26.68.49 support-2.support.com   support-2 
172.26.68.50 support-3.support.com   support-3

.

Step 3: Configure krb5.conf:

.

3.1 Configure [realm] section to add another cluster’s KDC server details – This is required to find KDC to authenticate user which belongs to another cluster.

Example on Cluster1:

[realms]
  HWX.COM = {
    admin_server = hwx-1.hwx.com
    kdc = hwx-1.hwx.com
  }
  SUPPORT.COM = {    
	admin_server = support-1.support.com
	kdc = support-1.support.com
}

.

3.2 Configure [domain_realm] section to add another cluster’s domain <-> realm mapping.

[domain_realm]
  .hwx.com = HWX.COM
  hwx.com = HWX.COM
  .support.com = SUPPORT.COM
  support.com = SUPPORT.COM

.

3.3 Configure [capaths] to add another cluster’s realm

[capaths]
    HWX.COM = {
         SUPPORT.COM = .
    }

On Cluster 1, the krb5.conf should look like below:

[libdefaults]
  renew_lifetime = 7d
  forwardable = true
  default_realm = HWX.COM
  ticket_lifetime = 24h
  dns_lookup_realm = false
  dns_lookup_kdc = false
  #default_tgs_enctypes = aes des3-cbc-sha1 rc4 des-cbc-md5
  #default_tkt_enctypes = aes des3-cbc-sha1 rc4 des-cbc-md5
[logging]
  default = FILE:/var/log/krb5kdc.log
  admin_server = FILE:/var/log/kadmind.log
  kdc = FILE:/var/log/krb5kdc.log
[realms]
  HWX.COM = {
    admin_server = hwx-1.hwx.com
    kdc = hwx-1.hwx.com
  }
  SUPPORT.COM = {
    admin_server = support-1.support.com
    kdc = support-1.support.com
  }
[domain_realm]
  .hwx.com = HWX.COM
  hwx.com = HWX.COM
  .support.com = SUPPORT.COM
  support.com = SUPPORT.COM
[capaths]
    HWX.COM = {
         SUPPORT.COM = .
    }

Note – Please copy modified /etc/krb5.conf to all the nodes in Cluster 1

.

Similarly on Cluster2, the krb5.conf should look like below:

[libdefaults]
  renew_lifetime = 7d
  forwardable = true
  default_realm = SUPPORT.COM
  ticket_lifetime = 24h
  dns_lookup_realm = false
  dns_lookup_kdc = false
  #default_tgs_enctypes = aes des3-cbc-sha1 rc4 des-cbc-md5
  #default_tkt_enctypes = aes des3-cbc-sha1 rc4 des-cbc-md5
[logging]
  default = FILE:/var/log/krb5kdc.log
  admin_server = FILE:/var/log/kadmind.log
  kdc = FILE:/var/log/krb5kdc.log
[realms]
  SUPPORT.COM = {
    admin_server = support-1.support.com
    kdc = support-1.support.com
  }
  HWX.COM = {
    admin_server = hwx-1.hwx.com
    kdc = hwx-1.hwx.com
  }
[domain_realm]
  .hwx.com = HWX.COM
  hwx.com = HWX.COM
  .support.com = SUPPORT.COM
  support.com = SUPPORT.COM
[capaths]
    SUPPORT.COM = {
        HWX.COM = .
    }

Note – Please copy modified /etc/krb5.conf to all the nodes in Cluster 2

.

Step 4: Modify below property in hdfs-site.xml on a cluster from where you want to execute distcp command ( specifically speaking - client side )

dfs.namenode.kerberos.principal.pattern=*

.

Step 5: Add a common trust principal in both the KDCs. Please keep same password for both the principals

.

On Cluster 1 and 2, execute below commands in kadmin utility:

addprinc krbtgt/HWX.COM@SUPPORT.COM 
addprinc krbtgt/SUPPORT.COM@HWX.COM

.

Step 6: Configure auth_to_local rules on both the clusters:

.

On Cluster1, append auth_to_local rules from Cluster2

Example on Cluster 1:

RULE:[1:$1@$0](ambari-qa-hadoop@HWX.COM)s/.*/ambari-qa/
RULE:[1:$1@$0](hdfs-hadoop@HWX.COM)s/.*/hdfs/
RULE:[1:$1@$0](spark-hadoop@HWX.COM)s/.*/spark/
RULE:[1:$1@$0](.*@HWX.COM)s/@.*//
RULE:[2:$1@$0](dn@HWX.COM)s/.*/hdfs/
RULE:[2:$1@$0](hive@HWX.COM)s/.*/hive/
RULE:[2:$1@$0](jhs@HWX.COM)s/.*/mapred/
RULE:[2:$1@$0](jn@HWX.COM)s/.*/hdfs/
RULE:[2:$1@$0](nm@HWX.COM)s/.*/yarn/
RULE:[2:$1@$0](nn@HWX.COM)s/.*/hdfs/
RULE:[2:$1@$0](rm@HWX.COM)s/.*/yarn/
RULE:[2:$1@$0](yarn@HWX.COM)s/.*/yarn/
DEFAULT
RULE:[1:$1@$0](ambari-qa-support@SUPPORT.COM)s/.*/ambari-qa/
RULE:[1:$1@$0](hdfs-support@SUPPORT.COM)s/.*/hdfs/
RULE:[1:$1@$0](spark-support@SUPPORT.COM)s/.*/spark/
RULE:[1:$1@$0](.*@SUPPORT.COM)s/@.*//
RULE:[2:$1@$0](dn@SUPPORT.COM)s/.*/hdfs/
RULE:[2:$1@$0](hive@SUPPORT.COM)s/.*/hive/
RULE:[2:$1@$0](jhs@SUPPORT.COM)s/.*/mapred/
RULE:[2:$1@$0](jn@SUPPORT.COM)s/.*/hdfs/
RULE:[2:$1@$0](nm@SUPPORT.COM)s/.*/yarn/
RULE:[2:$1@$0](nn@SUPPORT.COM)s/.*/hdfs/
RULE:[2:$1@$0](rm@SUPPORT.COM)s/.*/yarn/
RULE:[2:$1@$0](yarn@SUPPORT.COM)s/.*/yarn/

.

On Cluster2, append auth_to_local rules from Cluster1

Example on Cluster 2:

RULE:[1:$1@$0](ambari-qa-support@SUPPORT.COM)s/.*/ambari-qa/
RULE:[1:$1@$0](hdfs-support@SUPPORT.COM)s/.*/hdfs/
RULE:[1:$1@$0](spark-support@SUPPORT.COM)s/.*/spark/
RULE:[1:$1@$0](.*@SUPPORT.COM)s/@.*//
RULE:[2:$1@$0](dn@SUPPORT.COM)s/.*/hdfs/
RULE:[2:$1@$0](hive@SUPPORT.COM)s/.*/hive/
RULE:[2:$1@$0](jhs@SUPPORT.COM)s/.*/mapred/
RULE:[2:$1@$0](jn@SUPPORT.COM)s/.*/hdfs/
RULE:[2:$1@$0](nm@SUPPORT.COM)s/.*/yarn/
RULE:[2:$1@$0](nn@SUPPORT.COM)s/.*/hdfs/
RULE:[2:$1@$0](rm@SUPPORT.COM)s/.*/yarn/
RULE:[2:$1@$0](yarn@SUPPORT.COM)s/.*/yarn/
DEFAULT
RULE:[1:$1@$0](ambari-qa-hadoop@HWX.COM)s/.*/ambari-qa/
RULE:[1:$1@$0](hdfs-hadoop@HWX.COM)s/.*/hdfs/
RULE:[1:$1@$0](spark-hadoop@HWX.COM)s/.*/spark/
RULE:[1:$1@$0](.*@HWX.COM)s/@.*//
RULE:[2:$1@$0](dn@HWX.COM)s/.*/hdfs/
RULE:[2:$1@$0](hive@HWX.COM)s/.*/hive/
RULE:[2:$1@$0](jhs@HWX.COM)s/.*/mapred/
RULE:[2:$1@$0](jn@HWX.COM)s/.*/hdfs/
RULE:[2:$1@$0](nm@HWX.COM)s/.*/yarn/
RULE:[2:$1@$0](nn@HWX.COM)s/.*/hdfs/
RULE:[2:$1@$0](rm@HWX.COM)s/.*/yarn/
RULE:[2:$1@$0](yarn@HWX.COM)s/.*/yarn/

.

Step 7: Login to Cluster 2, do a kinit by local user and try to access hdfs files of Cluster 1

Example:

hdfs dfs -ls hdfs://hwx-2.hwx.com:8020/tmp
Found 8 items
drwx------   - ambari-qa hdfs          0 2016-07-29 23:24 hdfs://hwx-2.hwx.com:8020/tmp/ambari-qa
drwxr-xr-x   - hdfs      hdfs          0 2016-07-29 22:02 hdfs://hwx-2.hwx.com:8020/tmp/entity-file-history
drwx-wx-wx   - ambari-qa hdfs          0 2016-07-29 23:25 hdfs://hwx-2.hwx.com:8020/tmp/hive
-rwxr-xr-x   3 hdfs      hdfs       1414 2016-07-29 23:50 hdfs://hwx-2.hwx.com:8020/tmp/id1aac2d44_date502916
-rwxr-xr-x   3 ambari-qa hdfs       1414 2016-07-29 23:26 hdfs://hwx-2.hwx.com:8020/tmp/idtest.ambari-qa.1469834803.19.in
-rwxr-xr-x   3 ambari-qa hdfs        957 2016-07-29 23:26 hdfs://hwx-2.hwx.com:8020/tmp/idtest.ambari-qa.1469834803.19.pig
drwxr-xr-x   - ambari-qa hdfs          0 2016-07-29 23:53 hdfs://hwx-2.hwx.com:8020/tmp/tezsmokeinput

Note – hwx-2.hwx.com is the Active Namenode of Cluster 1.

.

You can try copying files from Cluster 2 to Cluster 1 using distcp

Example:

[kuldeepk@support-1 root]$ hadoop distcp hdfs://hwx-1.hwx.com:8020/tmp/test.txt /tmp/


16/07/30 22:03:27 INFO tools.DistCp: Input Options: DistCpOptions{atomicCommit=false, syncFolder=false, deleteMissing=false, ignoreFailures=false, maxMaps=20, sslConfigurationFile='null', copyStrategy='uniformsize', sourceFileListing=null, sourcePaths=[hdfs://hwx-1.hwx.com:8020/tmp/test.txt], targetPath=/tmp, targetPathExists=true, preserveRawXattrs=false}
16/07/30 22:03:27 INFO impl.TimelineClientImpl: Timeline service address: http://support-3.support.com:8188/ws/v1/timeline/
16/07/30 22:03:27 INFO client.RMProxy: Connecting to ResourceManager at support-3.support.com/172.26.68.50:8050
16/07/30 22:03:28 INFO hdfs.DFSClient: Created HDFS_DELEGATION_TOKEN token 20 for kuldeepk on 172.26.68.47:8020
16/07/30 22:03:28 INFO security.TokenCache: Got dt for hdfs://hwx-1.hwx.com:8020; Kind: HDFS_DELEGATION_TOKEN, Service: 172.26.68.47:8020, Ident: (HDFS_DELEGATION_TOKEN token 20 for kuldeepk)
16/07/30 22:03:29 INFO impl.TimelineClientImpl: Timeline service address: http://support-3.support.com:8188/ws/v1/timeline/
16/07/30 22:03:29 INFO client.RMProxy: Connecting to ResourceManager at support-3.support.com/172.26.68.50:8050
16/07/30 22:03:29 INFO hdfs.DFSClient: Created HDFS_DELEGATION_TOKEN token 24 for kuldeepk on ha-hdfs:support
16/07/30 22:03:29 INFO security.TokenCache: Got dt for hdfs://support; Kind: HDFS_DELEGATION_TOKEN, Service: ha-hdfs:support, Ident: (HDFS_DELEGATION_TOKEN token 24 for kuldeepk)
16/07/30 22:03:29 INFO mapreduce.JobSubmitter: number of splits:1
16/07/30 22:03:29 INFO mapreduce.JobSubmitter: Submitting tokens for job: job_1469916118318_0003
16/07/30 22:03:29 INFO mapreduce.JobSubmitter: Kind: HDFS_DELEGATION_TOKEN, Service: 172.26.68.47:8020, Ident: (HDFS_DELEGATION_TOKEN token 20 for kuldeepk)
16/07/30 22:03:29 INFO mapreduce.JobSubmitter: Kind: HDFS_DELEGATION_TOKEN, Service: ha-hdfs:support, Ident: (HDFS_DELEGATION_TOKEN token 24 for kuldeepk)
16/07/30 22:03:30 INFO impl.YarnClientImpl: Submitted application application_1469916118318_0003
16/07/30 22:03:31 INFO mapreduce.Job: The url to track the job: http://support-3.support.com:8088/proxy/application_1469916118318_0003/
16/07/30 22:03:31 INFO tools.DistCp: DistCp job-id: job_1469916118318_0003
16/07/30 22:03:31 INFO mapreduce.Job: Running job: job_1469916118318_0003
16/07/30 22:03:43 INFO mapreduce.Job: Job job_1469916118318_0003 running in uber mode : false
16/07/30 22:03:43 INFO mapreduce.Job:  map 0% reduce 0%
16/07/30 22:03:52 INFO mapreduce.Job:  map 100% reduce 0%
16/07/30 22:03:53 INFO mapreduce.Job: Job job_1469916118318_0003 completed successfully
16/07/30 22:03:53 INFO mapreduce.Job: Counters: 32
  File System Counters
    FILE: Number of bytes read=0
    FILE: Number of bytes written=142927
    FILE: Number of read operations=0
    FILE: Number of large read operations=0
    FILE: Number of write operations=0
    HDFS: Number of bytes read=346
    HDFS: Number of bytes written=45
    HDFS: Number of read operations=12
    HDFS: Number of large read operations=0
    HDFS: Number of write operations=2
  Job Counters
    Launched map tasks=1
    Other local map tasks=1
    Total time spent by all maps in occupied slots (ms)=14324
    Total time spent by all reduces in occupied slots (ms)=0
    Total time spent by all map tasks (ms)=7162
    Total vcore-seconds taken by all map tasks=7162
    Total megabyte-seconds taken by all map tasks=7333888
  Map-Reduce Framework
    Map input records=1
    Map output records=1
    Input split bytes=118
    Spilled Records=0
    Failed Shuffles=0
    Merged Map outputs=0
    GC time elapsed (ms)=77
    CPU time spent (ms)=1210
    Physical memory (bytes) snapshot=169885696
    Virtual memory (bytes) snapshot=2337554432
    Total committed heap usage (bytes)=66584576
  File Input Format Counters
    Bytes Read=228
  File Output Format Counters
    Bytes Written=45
  org.apache.hadoop.tools.mapred.CopyMapper$Counter
    BYTESSKIPPED=0
    SKIP=1

Note – hwx-1.hwx.com is the Active Namenode of Cluster 1.

.

Please comment if you have any feedback/questions/suggestions. Happy Hadooping!! 🙂

References:

http://crazyadmins.com

https://community.hortonworks.com/articles/18686/kerberos-cross-realm-trust-for-distcp.html

[#2]

@Kuldeep Kulkarni,

In the auth-to-local rule set examples, DEFAULT should be the last rule.

Also, this is a bit more than setting up the trust relationship between two MIT KDCs. It also includes some details about allowing two clusters to access each other's data. To do this, I believe that there are a few more steps. See https://community.hortonworks.com/articles/18686/kerberos-cross-realm-trust-for-distcp.html

[#3]

@Robert Levas - DEFAULT at the middle worked when I tried this setup.

I checked given article and I agree that modifying dfs.namenode.kerberos.principal.pattern was somehow missed while writing this article. I will add that missing step now.

Thank you! 🙂