#!/bin/bash

# on AWS and OpenStack, This script ends up in:
#       /var/lib/cloud/instance/scripts/part-001
# running as the 'root' user and is run during first start up of the machine
# This script attempts to keep its own log file.  If anything leaks out from
#   here, it will end up in:
#    /var/log/cloud-init-output.log

# start a log file to record information from the operation of this script:
export logFile="/tmp/startUpScript.log.$$"

# record environment to find out the operating conditions of this script

printf "#### begin start up script log %s\n" "`date '+%s %F %T'`" > "${logFile}"
printf "#### uname -a\n" >> "${logFile}"
uname -a >> "${logFile}" 2>&1
## printf "#### df -h\n" >> "${logFile}"
## df -h >> "${logFile}" 2>&1
printf "#### env\n" >> "${logFile}"
env >> "${logFile}" 2>&1
printf "#### set\n" >> "${logFile}"
set >> "${logFile}" 2>&1
printf "#### arp -a\n" >> "${logFile}"
arp -a >> "${logFile}" 2>&1
printf "#### ifconfig -a\n" >> "${logFile}"
ifconfig -a >> "${logFile}" 2>&1
printf "#### /etc/fstab\n" >> "${logFile}"
cat /etc/fstab >> "${logFile}"
printf "#### PATH\n" >> "${logFile}"
printf "${PATH}\n" >> "${logFile}" 2>&1

export homeDir="notFound"
export nativeUser="notFound"

# determine OpenStack or AWS environment
if [ -d "/home/centos" ]; then
  nativeUser="centos"
elif [ -d "/home/ec2-user" ]; then
  nativeUser="ec2-user"
fi
homeDir="/home/$nativeUser"

if [ -s "${homeDir}/.bashrc" ]; then
  export startTime=`date "+%s"`
  printf "## parasol node machine started install %s\n" "`date '+%s %F %T'`" >> /etc/motd
  # directories and symlinks to make this machine appear very much as it
  # would in the U.C. Santa Cruz Genome browser development environment
  mkdir -p /data ${homeDir}/bin /hive /cluster/bin /scratch /genomes
  chmod 777 /data /scratch /hive /cluster /cluster/bin /genomes
  chmod 755 ${homeDir}/bin
  chmod 755 ${homeDir}
  ln -s /data/bedtools /cluster/bin/bedtools
  ln -s /data /hive/data
  ln -s /dev/shm /scratch/tmp
  chown -R ${nativeUser}:${nativeUser} /data ${homeDir}/bin /genomes
  # these additions to .bashrc should protect themselves from repeating
  printf "export PATH=/data/bin:/data/bin/blat:${homeDir}/bin:\$PATH\n" >> "${homeDir}/.bashrc"
  printf "export LANG=C\n" >> "${homeDir}/.bashrc"
  printf "alias og='ls -ogrt'\n" >> "${homeDir}/.bashrc"
  printf "alias plb='parasol list batches'\n" >> "${homeDir}/.bashrc"
  printf "alias vi='vim'\n" >> "${homeDir}/.bashrc"
  printf "set -o vi\n" >> "${homeDir}/.bashrc"
  printf "set background=dark\n" >> "${homeDir}/.vimrc"
  printf "#### before yum install %s\n" "`date '+%s %F %T'`" >> "${logFile}"
  # useful to have the 'host' command, 'traceroute' and nmap ('nc')
  # to investigate the network, and wget for fetching, tcsh for csh shell
  #  bc for math calcs, perl-core installs 112 packages, vim for editing
  # convenience
  yum -y update >> "${logFile}" 2>&1
  yum repolist >> "${logFile}" 2>&1
  yum -y install bind-utils traceroute nmap-ncat tcsh screen vim-X11 vim-common vim-enhanced vim-minimal git-all bc >> "${logFile}" 2>&1
  yum -y install epel-release >> "${logFile}" 2>&1
  yum -y install gcc gcc-c++ zlib-devel tkinter libpng12 strace >> "${logFile}" 2>&1
  # this python business allows the crispr calculation pipeline to function
  # need newer python on the OpenStack
  if [ "${nativeUser}" = "centos" ]; then
     yum -y install centos-release-scl >> "${logFile}"
     yum -y install python36 python36-devel >> "${logFile}"
     python3.6 -m ensurepip --default-pip >> "${logFile}"
     python3.6 -m pip install --upgrade pip >> "${logFile}"
     python3.6 -m pip install --ignore-installed pytabix pandas twobitreader scipy matplotlib numpy >> "${logFile}"
     python3.6 -m pip install scikit-learn==0.16.1 Biopython xlwt >> "${logFile}" 2>&1
  else
     yum -y install python-pip python-devel >> "${logFile}" 2>&1
     pip install --ignore-installed pytabix pandas twobitreader scipy matplotlib numpy >> "${logFile}" 2>&1
     pip install scikit-learn==0.16.1 Biopython xlwt >> "${logFile}" 2>&1
  fi
  printf "#### after yum install %s\n" "`date '+%s %F %T'`" >> "${logFile}"

  export privateIp=`ifconfig -a | grep broadcast | head -1 | awk '{print $2}'`

  # this P A R A H U B will be sed changed to have the IP address of
  # the parasol hub machine
  printf "PARAHUB:/data /data nfs4 auto 0 0\n" >> /etc/fstab
  printf "PARAHUB:/genomes /genomes nfs4 auto 0 0\n" >> /etc/fstab
  mount /data >> "${logFile}" 2>&1
  mount /genomes >> "${logFile}" 2>&1

  # wait for NFS to work
  sleep 15
  # wake up NFS, ensure the files are here
  touch /data/parasol/nodeInfo >> "${logFile}" 2>&1
  # and wake it up again in a different way:
  stat /data/parasol/nodeInfo/id_rsa >> "${logFile}" 2>&1
  cat /data/parasol/nodeInfo/id_rsa.pub > "${homeDir}/.ssh/id_rsa.pub"
  # adding id_rsa.pub to authorized_keys will allow all machines in this
  # parasol system to ssh to each other without prompts, including localhost
  cat "${homeDir}/.ssh/id_rsa.pub" >> "${homeDir}/.ssh/authorized_keys"
  chmod 644 "${homeDir}/.ssh/id_rsa.pub"
  chown -R ${nativeUser}:${nativeUser} "${homeDir}/.ssh" "${homeDir}/.vimrc"
  export endTime=`date "+%s"`
  export et=`echo $endTime $startTime | awk '{printf "%d", $1-$2}'`
  printf "## parasol node machine finished install %s\n" "`date '+%s %F %T'`" >> /etc/motd
  printf "## elapsed time: %d seconds\n" "${et}" >> /etc/motd
  printf "## elapsed time: %d seconds\n## %s\n" "${et}" "`date '+%s %F %T'`" >> /tmp/node.machine.ready.signal
  printf "## this parasol node privateIp: %s\n" "${privateIp}" >> /tmp/node.machine.ready.signal
fi
printf "#### end start up script log %s\n" "`date '+%s %F %T'`" >> "${logFile}"