#!/bin/sh
# Create a Hadoop AMI. Runs on the EC2 instance.

# Import variables
bin=`dirname "$0"`
bin=`cd "$bin"; pwd`
. "$bin"/hadoop-ec2-env.sh

# Remove environment script since it contains sensitive information
rm -f "$bin"/hadoop-ec2-env.sh

# Install Java
cd /usr/local
wget -nv -O java.bin $JAVA_BINARY_URL
sh java.bin
rm -f java.bin

# Install tools
yum install rsync

# Install Hadoop
cd /usr/local
wget -nv http://www.apache.org/dist/lucene/hadoop/hadoop-$HADOOP_VERSION.tar.gz
tar xzf hadoop-$HADOOP_VERSION.tar.gz
rm -f hadoop-$HADOOP_VERSION.tar.gz

# Configure Hadoop
sed -i -e "s|# export JAVA_HOME=.*|export JAVA_HOME=/usr/local/jdk${JAVA_VERSION}|" \
       -e 's|# export HADOOP_LOG_DIR=.*|export HADOOP_LOG_DIR=/mnt/hadoop/logs|' \
       -e 's|# export HADOOP_SLAVE_SLEEP=.*|export HADOOP_SLAVE_SLEEP=1|' \
      /usr/local/hadoop-$HADOOP_VERSION/conf/hadoop-env.sh
mkdir -p /mnt/hadoop/logs

# Do configuration on instance startup
echo "/root/hadoop-init" >> /etc/rc.d/rc.local

# Configure networking.
# Delete SSH authorized_keys since it includes the key it was launched with. (Note that it is re-populated when an instance starts.)
rm -f /root/.ssh/authorized_keys
# Ensure logging in to new hosts is seamless.
echo '    StrictHostKeyChecking no' >> /etc/ssh/ssh_config

# Bundle and upload image
cd ~root
# Don't need to delete .bash_history since it isn't written until exit.
ec2-bundle-vol -d /mnt -k /mnt/pk-*.pem -c /mnt/cert-*.pem -u $AWS_ACCOUNT_ID -s 1536 -p hadoop-$HADOOP_VERSION
rm /mnt/pk-*.pem /mnt/cert-*.pem
ec2-upload-bundle -b $S3_BUCKET -m /mnt/hadoop-$HADOOP_VERSION.manifest.xml -a $AWS_ACCESS_KEY_ID -s $AWS_SECRET_ACCESS_KEY

# End
echo Done