#!/bin/bash # ######################################################################## # This program is part of $PROJECT_NAME$ # License: GPL License (see COPYING) # Authors: # Baron Schwartz, Roman Vynar # ######################################################################## # ######################################################################## # Redirect STDERR to STDOUT; Nagios doesn't handle STDERR. # ######################################################################## exec 2>&1 # ######################################################################## # Set up constants, etc. # ######################################################################## STATE_OK=0 STATE_WARNING=1 STATE_CRITICAL=2 STATE_UNKNOWN=3 STATE_DEPENDENT=4 # ######################################################################## # Run the program. # ######################################################################## main() { # Get options for o; do case "${o}" in -c) shift; OPT_CRIT="${1}"; shift; ;; --defaults-file) shift; OPT_DEFT="${1}"; shift; ;; -H) shift; OPT_HOST="${1}"; shift; ;; -i) shift; OPT_INTERVAL="${1}"; shift; ;; -l) shift; OPT_USER="${1}"; shift; ;; -L) shift; OPT_LOPA="${1}"; shift; ;; -p) shift; OPT_PASS="${1}"; shift; ;; -P) shift; OPT_PORT="${1}"; shift; ;; -S) shift; OPT_SOCK="${1}"; shift; ;; -T) shift; OPT_TABLE="${1}"; shift; ;; -w) shift; OPT_WARN="${1}"; shift; ;; --version) grep -A2 '^=head1 VERSION' "$0" | tail -n1; exit 0 ;; --help) perl -00 -ne 'm/^ Usage:/ && print' "$0"; exit 0 ;; -*) echo "Unknown option ${o}. Try --help."; exit 1; ;; esac done OPT_INTERVAL=${OPT_INTERVAL:-0} OPT_TABLE="${OPT_TABLE:-percona.checksums}" if [ -e '/etc/nagios/mysql.cnf' ]; then OPT_DEFT="${OPT_DEFT:-/etc/nagios/mysql.cnf}" fi if is_not_sourced; then if [ -n "$1" ]; then echo "WARN spurious command-line options: $@" exit 1 fi fi NOTE="UNK couldn't query the checksum table" # Set up a temp file to hold error messages from MySQL. TEMP=$(mktemp -t "${0##*/}.XXXXXX") || exit $? trap "rm -f '${TEMP}' >/dev/null 2>&1" EXIT # Get the query from the documentation and execute it. SQL=$(get_magic_query "${0}" checksum_diff_query) PROBLEMS=$(mysql_exec "${SQL/CHECKSUM_TABLE/${OPT_TABLE}}" 2>"${TEMP}") if [ $? = 0 ]; then if [ "${PROBLEMS}" ]; then NOTE="pt-table-checksum found ${PROBLEMS}" if [ "${OPT_CRIT}" ]; then NOTE="CRIT $NOTE" else NOTE="WARN $NOTE" fi else NOTE="OK pt-table-checksum found no out-of-sync tables" if [ "${OPT_INTERVAL}" -gt 0 ]; then RECENT_CHUNKS=$(mysql_exec "SELECT IF(COALESCE(MAX(ts), NOW()) > NOW() - INTERVAL ${OPT_INTERVAL} DAY, 1, 0) FROM ${OPT_TABLE}") if [ "${RECENT_CHUNKS}" = 0 ]; then NOTE="pt-table-checksum was not run over last ${OPT_INTERVAL} days" if [ "${OPT_CRIT}" ]; then NOTE="CRIT $NOTE" else NOTE="WARN $NOTE" fi fi fi fi else if grep "Table '${OPT_TABLE}' doesn't exist" "${TEMP}" >/dev/null 2>&1; then NOTE="UNK table '${OPT_TABLE}' doesn't exist" fi fi echo $NOTE } # ######################################################################## # Execute a MySQL command. # ######################################################################## mysql_exec() { mysql ${OPT_DEFT:+--defaults-file="${OPT_DEFT}"} \ ${OPT_LOPA:+--login-path="${OPT_LOPA}"} \ ${OPT_HOST:+-h"${OPT_HOST}"} ${OPT_PORT:+-P"${OPT_PORT}"} \ ${OPT_USER:+-u"${OPT_USER}"} ${OPT_PASS:+-p"${OPT_PASS}"} \ ${OPT_SOCK:+-S"${OPT_SOCK}"} -ss -e "$1" } # ######################################################################## # Retrieve a paragraph from the given file, which includes MAGIC_$2 as a # pattern. # ######################################################################## get_magic_query() { perl -00 -ne"m/MAGIC_$2/ && print" "$1" } # ######################################################################## # Determine whether this program is being executed directly, or sourced/included # from another file. # ######################################################################## is_not_sourced() { [ "${0##*/}" = "pmp-check-pt-table-checksum" ] || [ "${0##*/}" = "bash" -a "$_" = "$0" ] } # ######################################################################## # Execute the program if it was not included from another file. # This makes it possible to include without executing, and thus test. # ######################################################################## if is_not_sourced; then OUTPUT=$(main "$@") EXITSTATUS=$STATE_UNKNOWN case "${OUTPUT}" in UNK*) EXITSTATUS=$STATE_UNKNOWN; ;; OK*) EXITSTATUS=$STATE_OK; ;; WARN*) EXITSTATUS=$STATE_WARNING; ;; CRIT*) EXITSTATUS=$STATE_CRITICAL; ;; esac echo "${OUTPUT}" exit $EXITSTATUS fi # ############################################################################ # Documentation # ############################################################################ : <<'DOCUMENTATION' =pod =head1 NAME pmp-check-pt-table-checksum - Alert when pt-table-checksum finds data differences on a replica. =head1 SYNOPSIS Usage: pmp-check-pt-table-checksum [OPTIONS] Options: -c CRIT Raise a critical error instead of a warning. --defaults-file FILE Only read mysql options from the given file. Defaults to /etc/nagios/mysql.cnf if it exists. -H HOST MySQL hostname. -l USER MySQL username. -L LOGIN-PATH Use login-path to access MySQL (with MySQL client 5.6). -p PASS MySQL password. -P PORT MySQL port. -S SOCKET MySQL socket file. -i INTERVAL Interval over which to ensure pt-table-checksum was run, in days; default - not to check. -T TABLE The checksum table; default percona.checksums -w WARN Warning threshold; ignored. --help Print help and exit. --version Print version and exit. Options must be given as --option value, not --option=value or -Ovalue. Use perldoc to read embedded documentation with more details. =head1 DESCRIPTION This Nagios plugin examines whether MySQL replication has drifted out of sync with the master's data, according to checks performed by the pt-table-checksum tool in Percona Toolkit. It uses the following query to determine whether the server's data matches its master's: SELECT /* MAGIC_checksum_diff_query */ CONCAT( COUNT(*), ' chunks differ in ', COUNT(DISTINCT CONCAT(db, tbl)), ' tables, including ', MIN(CONCAT(db, '.', tbl))) FROM CHECKSUM_TABLE WHERE master_cnt <> this_cnt OR master_crc <> this_crc OR ISNULL(master_crc) <> ISNULL(this_crc) HAVING COUNT(*) > 0 The word CHECKSUM_TABLE is replaced by the value of the -T option. If the table specified by -T does not exist, unknown is raised. Optionally, you can specify an interval in days over which to ensure pt-table-checksum was run. It is useful in cases when the cron job doing the checksumming suddenly stopped working. This option will have an effect when no diffs are found and the checksum table is not empty. Alerts are raised at a WARNING level by default, but specifying the -c option with any value will change this to CRITICAL instead. =head1 PRIVILEGES This plugin executes the following commands against MySQL: =over =item * C