impala-shell-action.sh

#!/bin/bash

# paranoia with shell scripts, always to be encouraged
set -e
LOG=/tmp/shell-impala-$USER-$(date +%s).log

# echo-ing params to find logs on local fs in case of error
echo "HOST: ${HOSTNAME}"
echo "LOG: ${LOG}"

if [ $# -ne 4 ];
then
  ERROR="Usage: $0 <keytab-file> <kerberos-principal> <query-file> <impalad_host:impalad_port>"
  echo "$ERROR"
  echo "$ERROR" >> ${LOG}
  exit 1
fi

KEYTAB=$1
PRINCIPAL=$2
QUERY_FILE=$3
CONN_PARAM=$4
export PYTHON_EGG_CACHE=./myeggs

if [ ! -f $KEYTAB ];
then
  ERROR="Unable to access [ keytab-file: $KEYTAB ], check existence and permissions"
  echo "$ERROR"
  echo "$ERROR" >> ${LOG}
  exit 2
fi

if [ ! -f $QUERY_FILE ];
then
  ERROR="Unable to access [ query-file: $QUERY_FILE ], check existence and permissions"
  echo "$ERROR"
  echo "$ERROR" >> ${LOG}
  exit 3
fi

# redirecting both stderror and stdout to append to the log-file
/usr/bin/kinit -kt $KEYTAB -V "$PRINCIPAL" >>${LOG} 2>&1
/usr/bin/klist -e >>${LOG} 2>&1
impala-shell -k -i $CONN_PARAM -f $QUERY_FILE >>${LOG} 2>&1
RC=$?

if [ $RC -ne 0 ];
then
  echo "Impala-shell did not finish successfully, [ rc = $RC ]" >>${LOG}
  hadoop fs -put ${LOG} /tmp
  # indicating error so shell action fails in Oozie
  exit 4
fi

hadoop fs -put ${LOG} /tmp