#!/bin/bash

# rburkholder@quovadis.bm

# raymond@burkholder.net
# http://blog.raymond.burkholder.net

# requires use of:
# https://github.com/juanpabloaj/slacker-cli but install with pip
# https://api.slack.com/bot-users
# https://my.slack.com/apps/build/custom-integration, choose bots, create new one, and add token
# https://my.slack.com/admin#disabled)  to delete test bots
# apt-get install python-pip
# pip install slacker-cli
# #git clone https://pypi.python.org/pypi/slacker-cli/
while true
do
  # user defined settings
  txtEmail="name@example.com"
  txtSubject="Path Loss Report"
  txtSeparator="==============================\n"
  nTrigger=4
  nAttempts=5
  txtPdServiceKey="000000000000000"
  txtSlackBotToken="xoxb-250697658007-4127858946773-gyM0PMGA5XPbh3ZEYot4Yy5Z"
  txtSlackChannel="general"

  declare -A nodes
  nodes=( \
    [217.17.208.20]="NG-firsthop" \
    [149.36.0.254]="NG-router" \
    )

  # local variables
  status="|"
  cntNotify=0
  tmpLog=$(mktemp)
  tmpPing=$(mktemp)
  cntNodes=0
  cntNodesDown=0
  cntMembers=0
  cntMembersDown=0
  declare -a items
  echo "tmplog @ ${tmpLog}"
  # preload output
  date > ${tmpLog}

  # loop through nodes and test
  for node in  ${!nodes[*]}; do

    ((cntNodes++))

    # split out node details
    # 0: alias/name
    # 1: optional 'member' for determining edge outage
    info=${nodes[${node}]}
    items[1]="none"
    ix=0
    for arg in ${info}; do
      items[ix]=${arg}
      ((ix++))
      done

    name="${items[0]}"

    echo -e ${txtSeparator} >> ${tmpLog};
    echo "checking node ${name}:" >> ${tmpLog};
    echo "" >> ${tmpLog}

    ping -W 1 -c ${nAttempts} ${node} > ${tmpPing}
    cat ${tmpPing} >> ${tmpLog}

    value=$(grep transmitted ${tmpPing} | cut -d ' ' -f 4)
    if [[ nTrigger -ge value ]]; then
      flagNxt="dn"
      ((cntNodesDown++))
      echo "node ${name} DOWN" >> ${tmpLog};

    else
      echo "node ${name} UP" >> ${tmpLog};  
      flagNxt="up"
      fi

    if test "member" == "${items[1]}"; then
      ((cntMembers++))
      if test "dn" = "${flagNxt}"; then ((cntMembersDown++)); fi
      fi

    flagPrv="na"

    if [[ -f /tmp/pl.dn.${node} ]]; then
      flagPrv="dn"
      if test "up" = "${flagNxt}"; then
        rm /tmp/pl.dn.${node}
        fi
      fi

    if [[ -f /tmp/pl.up.${node} ]]; then
      flagPrv="up"
      if test "dn" = "${flagNxt}"; then
        rm /tmp/pl.up.${node}
      fi
    fi

    if test "${flagPrv}" != "${flagNxt}"; then
      echo "Above State Change: ${flagPrv}>${flagNxt}" >> ${tmpLog}
      touch /tmp/pl.${flagNxt}.${node}
      ((cntNotify++));
      mtr -w -b --report -n ${node} >> ${tmpLog};
      status="${status} ${name} ${flagPrv}>${flagNxt} |"
      echo "" >> ${tmpLog}
    fi

  done

  # footer
  echo -e ${txtSeparator} >> ${tmpLog}
  date >> ${tmpLog}
  echo "cntNodes ${cntNodes}" >> ${tmpLog};
  echo "cntNodesDown ${cntNodesDown}" >> ${tmpLog};
  echo "cntNotify ${cntNotify}" >> ${tmpLog};
  # notify on failure
  if [[ cntNodes -eq cntNodesDown ]]; then
    echo "path-loss - all nodes unreachable" >> ${tmpLog}
    logger "path-loss - all nodes unreachable"
  else
    # if something to notify
    if [[ cntNotify -gt 0 ]]; then
      echo "Sending an alert" >> ${tmpLog}
      # need a pagerduty alert if all important members are down
      if [[ cntMembersDown >1 ]]; then
        #response=$(cat ${tmpLog} | ./pd-trigger.sh -L -s "${txtPdServiceKey}" -d "${status}")
        #fi
      # attempt an email
      #cat ${tmpLog} | mail -s "${txtSubject}:${status}" ${txtEmail};
      echo "All members down sending slack alert with  /usr/local/bin/slacker -c ${txtSlackChannel} -t ${txtSlackBotToken}" >> ${tmpLog}
      cat ${tmpLog} | /home/ubuntu/.local/bin/slacker -c ${txtSlackChannel} -t ${txtSlackBotToken}
     fi
    fi
  fi

  cp ${tmpLog} ./log.log
  cp ${tmpPing} ./ping.log

  # clean up
  rm ${tmpLog}
  rm ${tmpPing}
done
exit 0