start-accumulo

#!/bin/bash

rewrite() {
    cat $1 | \
	sed \
	    -e "s#@ZOOKEEPERS@#${ZOOKEEPERS}#" \
	    -e "s#@HDFS_VOLUMES@#${HDFS_VOLUMES}#" \
	    -e "s#@SECRET@#${SECRET}#" \
	    -e "s#@MEMORY_MAPS_MAX@#${MEMORY_MAPS_MAX}#" \
	    -e "s#@CACHE_DATA_SIZE@#${CACHE_DATA_SIZE}#" \
	    -e "s#@CACHE_INDEX_SIZE@#${CACHE_INDEX_SIZE}#" \
	    -e "s#@SORT_BUFFER_SIZE@#${SORT_BUFFER_SIZE}#" \
	    -e "s#@WALOG_MAX_SIZE@#${WALOG_MAX_SIZE}#" \
	    > $1.tmp
    mv -f $1.tmp $1
}

# Default memory allocations
MEMORY_MAPS_MAX=${MEMORY_MAPS_MAX:-8M}
CACHE_DATA_SIZE=${CACHE_DATA_SIZE:-2M}
CACHE_INDEX_SIZE=${CACHE_INDEX_SIZE:-2M}
SORT_BUFFER_SIZE=${SORT_BUFFER_SIZE:-5M}
WALOG_MAX_SIZE=${WALOG_MAX_SIZE:-5M}

MY_IP=$(/usr/sbin/ifconfig | grep 'inet[^6]' | awk '{print $2}' |
	    sed 's/addr://' | grep -v 127.0.0.1 | grep -v 0.0.0.0 |
	    head -n 1)
echo My IP is ${MY_IP}.

MY_HOSTNAME=${MY_HOSTNAME:-${MY_IP}}
if [ "${MY_HOSTNAME}" == MY_IP ]
then
    MY_HOSTNAME=${MY_IP}
fi

ZOOKEEPERS=${ZOOKEEPERS:-zookeeper:2181}
HDFS_VOLUMES=${HDFS_VOLUMES:-hdfs://hadoop:9000/accumulo}
NAMENODE_URI=${NAMENODE_URI:-hdfs://hadoop:9000/}
INSTANCE_NAME=${INSTANCE_NAME:-accumulo}
SECRET=${SECRET:-accumulo}
ACCUMULO_INIT=${ACCUMULO_INIT:-${HDFS_VOLUMES}_init}

DAEMONS=${DAEMONS:-gc,master,tserver,monitor,tracer}

GC_HOSTS=${GC_HOSTS:-${MY_HOSTNAME}}
MASTER_HOSTS=${MASTER_HOSTS:-${MY_HOSTNAME}}
SLAVE_HOSTS=${SLAVE_HOSTS:-${MY_HOSTNAME}}
MONITOR_HOSTS=${MONITOR_HOSTS:-${MY_HOSTNAME}}
TRACER_HOSTS=${TRACER_HOSTS:-${MY_HOSTNAME}}

. /usr/local/accumulo/conf/accumulo-env.sh

rewrite ${ACCUMULO_HOME}/conf/accumulo-site.xml

(
    for v in $(echo ${GC_HOSTS} | 
		      sed -e 's/,/ /g' -e "s/MY_IP/${MY_IP}/g")
    do
	echo $v
    done
) > /usr/local/accumulo/conf/gc

(
    for v in $(echo ${MASTER_HOSTS} | 
		      sed -e 's/,/ /g' -e "s/MY_IP/${MY_IP}/g")
    do
	echo $v
    done
) > /usr/local/accumulo/conf/masters

(
    for v in $(echo ${SLAVE_HOSTS} | 
		      sed -e 's/,/ /g' -e "s/MY_IP/${MY_IP}/g")
    do
	echo $v
    done
) > /usr/local/accumulo/conf/slaves

(
    for v in $(echo ${MONITOR_HOSTS} | 
		      sed -e 's/,/ /g' -e "s/MY_IP/${MY_IP}/g")
    do
	echo $v
    done
) > /usr/local/accumulo/conf/monitor

(
    for v in $(echo ${TRACER_HOSTS} | 
		      sed -e 's/,/ /g' -e "s/MY_IP/${MY_IP}/g")
    do
	echo $v
    done
) > /usr/local/accumulo/conf/tracers

(
  echo "<configuration>"
  echo "  <property>"
  echo "    <name>fs.defaultFS</name>"
  echo "    <value>${NAMENODE_URI}</value>"
  echo "  </property>"
  echo "</configuration>"
) > /usr/local/hadoop/etc/hadoop/core-site.xml

# Loop until Accumulo is successfully initialised.
while true
do

  /usr/local/hadoop/bin/hdfs dfs -ls /
  if [ $? -ne 0 ]
  then
    echo 'Hadoop not up?'
    sleep 10  
    continue
  fi

  /usr/local/hadoop/bin/hdfs dfs -ls ${ACCUMULO_INIT}
  if [ $? -eq 0 ]
  then
    echo 'Accumulo has been initialised.'
    break
  fi

  echo 'Accumulo not initialised.'

  echo 'Initialise accumulo...'
  /usr/local/accumulo/bin/accumulo init --instance-name ${INSTANCE_NAME} \
    --password ${SECRET} --clear-instance-name
  if [ $? -ne 0 ]
  then
    echo Initialisation failed.
    sleep 1
    continue
  fi

  /usr/local/hadoop/bin/hdfs dfs -put /dev/null ${ACCUMULO_INIT}

done

/usr/local/accumulo/bin/start-here.sh

while true
do
    sleep 100000
done

#<configuration>
#  <property>
#    <name>fs.defaultFS</name>
#    <value>hdfs://hadoop:9000/</value>
#  </property>
#</configuration>