#!/usr/bin/env bash
# This program is part of Aspersa (http://code.google.com/p/aspersa/)

# ########################################################################
# A script to watch MySQL and run the 'collect' program when some condition
# becomes true.  By default, it watches for a too-many-connections condition.
# This is a good script to run in a screen session.  It's separate from the
# 'collect' script because that lets you change 'collect' without stopping
# and restarting this one.
#
# The name 'stalk' is because 'watch' is already taken, and 'stalk' is fun.
#
# Author: Baron Schwartz
# ########################################################################

# ########################################################################
# Check for the existence of a config file and source it if it exists 
# ########################################################################
if [ -f "${0}.conf" ]; then
   . "${0}.conf"
fi

# ########################################################################
# Configuration settings.
# ########################################################################
# This is the max number of <whatever> we want to tolerate.
THRESHOLD=${THRESHOLD:-100}

# This is the thing to check for.
VARIABLE=${VARIABLE:-Threads_connected}

# How many times must the condition be met before the script will fire?
CYCLES=${CYCLES:-1}

# Collect GDB stacktraces?
GDB=${GDB:-no}

# Collect oprofile data?
OPROFILE=${OPROFILE:-yes}

# Collect strace data?
STRACE=${STRACE:-no}

# Collect tcpdump data?
TCPDUMP=${TCPDUMP:-yes}

# Send mail to this list of addresses when the script triggers.
# EMAIL=

# Any options to pass to mysql/mysqladmin, such as -u, -p, etc
# MYSQLOPTIONS=""

# This is the interval between checks.
INTERVAL=${INTERVAL:-30}

# If the command you're running to detect the condition is allowed to return
# nothing (e.g. a grep line that might not even exist if there's no problem),
# then set this to "yes".
MAYBE_EMPTY=${MAYBE_EMPTY:-no}

# This is the location of the 'collect' script.
if [ -z "${COLLECT}" ]; then
   COLLECT="${HOME}/bin/collect";
fi

# This is where to store the collected data.
if [ -z "${DEST}" ]; then
   DEST="${HOME}/collected/"
fi

# How long to collect statistics data for?  Make sure that this isn't longer
# than SLEEP.
DURATION=${DURATION:-30}

# How long to sleep after collecting?
if [ -z "${SLEEP}" ]; then
   SLEEP=$(($DURATION * 10))
fi

# Bail out if the disk is more than this %full.
PCT_THRESHOLD=${PCT_THRESHOLD:-95}

# Bail out if the disk has less than this many MB free.
MB_THRESHOLD=${MB_THRESHOLD:-100}

# Remove samples after this many days.
PURGE=${PURGE:-30}

# ########################################################################
# End configuration
# ########################################################################

# ########################################################################
# Echo to STDERR and exit false.
# ########################################################################
die() {
   echo "${1}" >&2
   exit 1
}

# ########################################################################
# Echo to STDERR and possibly email.
# ########################################################################
log() {
   if [ "${EMAIL}" ]; then
      echo "${1} on $(hostname)" | mail -s "${2} on $(hostname)" ${EMAIL}
   fi
   echo "${1}" >&2
}

# Make the collection location
mkdir -p "${DEST}" || die "Can't make the destination directory"
test -d "${DEST}"  || die "${DEST} isn't a directory"
test -w "${DEST}"  || die "${DEST} isn't writable"

# Test if we have root; warn if not, but it isn't critical.
if [ "$(id -u)" != "0" ]; then
   echo 'Not running with root privileges!';
fi

# We increment this variable every time that the check is true, and set it to 0
# if it's false.
cycles_true=0;

while true; do
   d=$(date +%F-%T | tr :- _);

   # XXX This is where we decide whether to execute 'collect'.
   # XXX Customize this if needed.  The idea is to generate a number and store
   # XXX it into $detected, and if $detected > $THRESHOLD, then we'll execute
   # XXX the collection process.
   detected=$(mysqladmin ext ${MYSQLOPTIONS} | grep ${VARIABLE} | awk '{print $4}');
   if [ -z "${detected}" -a ${MAYBE_EMPTY} = "no" ]; then
      # Oops, couldn't connect, maybe max_connections problem?
      echo "$d The detected value is empty; something failed?  Exit status is $?"
      matched="yes"
      cycles_true=$(($cycles_true + 1))
   elif [ "${detected:-0}" -gt ${THRESHOLD} ]; then
      matched="yes"
      cycles_true=$(($cycles_true + 1))
   else
      matched="no"
      cycles_true=0
   fi

   # XXX Stop customizing here; everything above should be what you need.

   NOTE="$d check results: ${VARIABLE} = ${detected}, matched = ${matched}, cycles_true = ${cycles_true}"
   # Actually execute the collection script.
   if [ "${matched:-no}" = "yes" -a ${cycles_true} -ge ${CYCLES} ]; then 

      log "${NOTE}" "${COLLECT} triggered"
      PREFIX="$(date +%F-%T | tr :- _)"
      echo "${NOTE}" > "${DEST}/${PREFIX}-trigger"
      ${COLLECT} -d "${DEST}" -i "${DURATION}" -g "${GDB}" -o "${OPROFILE}" -p "${PREFIX}" -s "${STRACE}" -t "${TCPDUMP}" ${MYSQLOPTIONS} -f "${PCT_THRESHOLD}" -m "${MB_THRESHOLD}"
      echo "$d sleeping ${SLEEP} seconds to avoid DOS attack"
      sleep ${SLEEP}
   else
      echo ${NOTE}
      sleep ${INTERVAL}
   fi


   # Delete things more than $PURGE days old
   find "${DEST}" -type f -mtime +${PURGE} -exec rm -f '{}' \;
   find "/var/lib/oprofile/samples" -type d -name 'aspersa_collect_*' \
      -depth -mtime +${PURGE} -exec rm -f '{}' \;

done