Setting up Kerberos on SnapLogic groundplex for authentication to Cloudera - Hive account and snap pack

Hello @PSAmmirata -

Thank you for your question about configuration of Kerberos setup with Snaplogic Groundplex. The steps listed are sufficient to configure groundplex to work with Kerberized Hadoop clusters. Were you able to follow steps and configure the groundplex? Do you see any issues?

We’re using RHEL7. From what I’ve read the krb5-auth-diaglog package is depracated in RHEL7.

Step 5 says to “Generate the keytab file for the kerberos user.” Does anyone have details on how to do this?

Creating a keytab file is straightforward.

If you are creating a keytab file for a user with a password you should use the ‘kutil’ program.

$ ktuil
ktutil: add_entry -password -p principal -k knvo -e enctype
(enter password)
ktutil: write_kt keytabfile
ktutil quit

where

• principal is your principal, e.g., bob@example.com or bob/hdfs@example.com for a more restricted principal
• knvo is the key version number. 1 should be fine.
• enctype is the encryption type. This is typically something like aes128-cts-hmac-sha1-96, des3-cbc-sha1, or arcfour-hmac. You should check with your system administrator to get the precise encryption types required. You can call this line multiple times, once for each encryption type.
• keytabfile is your keytab file. It traditionally ends with the .keytab extension.

You can verify the new file with ‘klist -kt keytabfile’.

If you are creating a keytab file for a server you must use the ‘kadmin’ program.

If the server principal does not exist yet:

$ kadmin
kadmin: add_principal principal
kadmin: ktadd -k keytabfile principal

If the server principal already exists:

$ kadmin
kdamin: ktadd -k keytabfile -norandkey principal

where principal is something like “hdfs/172.3.1.7@MYORG.EXAMPLE.COM”.

Step 6/7 refers to a Hadoop (HDFS) configuration directory. We’re using Cloudera and our Cloudera admin said we don’t have a config directory. Can anyone provide details on what directory this is?

We did get this working. Here are some of my notes that may be useful.

1. We use EMC Isilon storage and the Hadoop configuration details were contained in the Isilon client configuration file.
2. We used Cloudera Hive JDBC driver 2.5.19 and needed to specify all of the JAR files extracted from the ZIP file archive.