Enable Kerberos Authentication

You can enable Kerberos authentication for Impala on a secure and insecure MapR cluster.

Once you have configured Impala to use Kerberos for authentication, restart Impala and then start the impala-shell with the -s mapr -k flags to enable Kerberos.

To enable Kerberos authentication for Impala, complete the following steps:
  1. Copy the following files to the $IMPALA_HOME/conf/ directory:
    • $HIVE_HOME/conf/hive-site.xml
    • $HADOOP_HOME/etc/hadoop/core-site.xml
    NOTE: Any time the hive-site.xml file is modified, copy the file to the $IMPALA_HOME/conf/ directory.
  2. Create service principals for each host that runs impalad, catalogd, or statestored and for the HTTP service. Principal names take the following form:
    
    mapr/<fully.qualified.domain.name>@<KERBEROS.REALM>
    1. Create an Impala service principal and specify the following information:
      • Name “mapr”
      • Fully qualified domain name of each node running impalad
      • Realm name

        kadmin: addprinc -requires_preauth -randkey -allow_renewable mapr/impala_host.example.com@TEST.EXAMPLE.COM
    2. Create an HTTP service principal.
      kadmin: addprinc -randkey HTTP/impala_host.example.com@TEST.EXAMPLE.COM
  3. Create, merge, and distribute keytab files for the principals.
    1. Create keytab files with both principals.
      kadmin: xst -k /opt/mapr/conf/mapr.keytab mapr/impala_host.example.com
    2. Use the keytab utility to read the content of the keytab files and then write the content to a new file.
      ktutil
      ktutil: rkt /opt/mapr/conf/mapr.keytab
      ktutil: rkt /opt/mapr/conf/http.keytab
      ktutil: wkt /opt/mapr/conf/mapr-http.keytab
      ktutil: quit
    3. Optionally, test the credentials in the merged keytab file to verify their validity and to verify that “renew until” data is set to a future time.
      klist -e -k -t /opt/mapr/conf/mapr-http.keytab
    4. Change the file owner to the mapr user to make mapr the only user authorized to read the file content.
      chmod 400 /opt/mapr/conf/mapr-http.keytab
  4. Edit /opt/mapr/impala/impala-<version>/conf/env.sh to include the fully qualified domain name for the IMPALA_STATE_STORE_HOST, IMPALA_STATE_STORE_HOST variables, and Kerberos options.
    1. Set the IMPALA_STATE_STORE_HOST and CATALOG_SERVICE_HOST variables to point to the fully qualified domain name.
      IMPALA_STATE_STORE_HOST=impala_host.example.com
      IMPALA_STATE_STORE_PORT=24000
      CATALOG_SERVICE_HOST=impala_host.example.com
    2. Add the following Kerberos options for impalad, catalogd, and statestored daemons using the IMPALA_SERVER_ARGS, IMPALA_CATALOG_ARGS, and IMPALA_STATE_STORE_ARGS variables:
      -kerberos_reinit_interval=60
      -principal=mapr/impala_host.example.com@TEST.EXAMPLE.COM
      -keytab_file=/opt/mapr/conf/mapr-http.keytab
      
      IMPALA_SERVER_ARGS=" \
           -log_dir=${IMPALA_LOG_DIR} \
           -state_store_port=${IMPALA_STATE_STORE_PORT} \
           -use_statestore \
           -authorized_proxy_user_config=mapr=* \
           -state_store_host=${IMPALA_STATE_STORE_HOST} \
           -catalog_service_host=${CATALOG_SERVICE_HOST} \
           -be_port=${IMPALA_BACKEND_PORT} \
           -disable_admission_control=true \
           -kerberos_reinit_interval=60 \
           -principal=mapr/impala_host.example.com@TEST.EXAMPLE.COM \
           -keytab_file=/opt/mapr/conf/mapr-http.keytab "
    3. Restart Impala and the catalog and statestore services. See Managing Impala.
    4. To enable Kerberos from the impala-shell, start the impala-shell with the -s mapr -k flags.
      impala-shell -s mapr -k
      For more information on changing the Impala defaults specified in env.sh, see Impala-Shell Commands.