#!/usr/bin/env bash
-# (C) Costa Tsaousis
-# pushover support by Jan Arnold
+# netdata
+# real-time performance and health monitoring, done right!
+# (C) 2016 Costa Tsaousis <costa@tsaousis.gr>
+# GPL v3+
+#
+# Script to send alarm notifications for netdata
+#
+# Features:
+# - multiple notification methods
+# - multiple roles per alarm
+# - multiple recipients per role
+# - severity filtering per recipient
+#
+# Supported notification methods:
+# - emails
+# - pushover.net notifications
+# - slack.com notifications
+# - telegram.org notifications
+#
+
+export PATH="${PATH}:/sbin:/usr/sbin:/usr/local/sbin"
+export LC_ALL=C
-me="${0}"
+# -----------------------------------------------------------------------------
-if [ $(( ${BASH_VERSINFO[0]} )) -lt 4 ]
-then
- echo >&2
- echo >&2 "$me: ERROR"
- echo >&2 "BASH version 4 or later is required."
- echo >&2 "You are running version: ${BASH_VERSION}"
- echo >&2 "Please upgrade."
- echo >&2
+PROGRAM_NAME="$(basename "${0}")"
+
+logdate() {
+ date "+%Y-%m-%d %H:%M:%S"
+}
+
+log() {
+ local status="${1}"
+ shift
+
+ echo >&2 "$(logdate): ${PROGRAM_NAME}: ${status}: ${*}"
+
+}
+
+warning() {
+ log WARNING "${@}"
+}
+
+error() {
+ log ERROR "${@}"
+}
+
+info() {
+ log INFO "${@}"
+}
+
+fatal() {
+ log FATAL "${@}"
exit 1
-fi
+}
+
+debug=0
+debug() {
+ [ $debug -eq 1 ] && log DEBUG "${@}"
+}
+
+# -----------------------------------------------------------------------------
+# check for BASH v4+ (required for associative arrays)
+[ $(( ${BASH_VERSINFO[0]} )) -lt 4 ] && \
+ fatal "BASH version 4 or later is required (this is ${BASH_VERSION})."
+
+# -----------------------------------------------------------------------------
# defaults to allow running this script by hand
+
NETDATA_CONFIG_DIR="${NETDATA_CONFIG_DIR-/etc/netdata}"
NETDATA_CACHE_DIR="${NETDATA_CACHE_DIR-/var/cache/netdata}"
[ -z "${NETDATA_REGISTRY_URL}" ] && NETDATA_REGISTRY_URL="https://registry.my-netdata.io"
# -----------------------------------------------------------------------------
# parse command line parameters
-recipient="${1}" # the recepient of the email
-host="${2}" # the host this event refers to
+roles="${1}" # the roles that should be notified for this event
+host="${2}" # the host generated this event
unique_id="${3}" # the unique id of this event
alarm_id="${4}" # the unique id of the alarm that generated this event
-event_id="${5}" # the incremental id of the event, for this alarm
-when="${6}" # the timestamp this event occured
+event_id="${5}" # the incremental id of the event, for this alarm id
+when="${6}" # the timestamp this event occurred
name="${7}" # the name of the alarm, as given in netdata health.d entries
chart="${8}" # the name of the chart (type.id)
family="${9}" # the family of the chart
status="${10}" # the current status : REMOVED, UNITIALIZED, UNDEFINED, CLEAR, WARNING, CRITICAL
old_status="${11}" # the previous status: REMOVED, UNITIALIZED, UNDEFINED, CLEAR, WARNING, CRITICAL
-value="${12}" # the current value
-old_value="${13}" # the previous value
+value="${12}" # the current value of the alarm
+old_value="${13}" # the previous value of the alarm
src="${14}" # the line number and file the alarm has been configured
-duration="${15}" # the duration in seconds the previous state took
-non_clear_duration="${16}" # the total duration in seconds this is non-clear
+duration="${15}" # the duration in seconds of the previous alarm state
+non_clear_duration="${16}" # the total duration in seconds this is/was non-clear
units="${17}" # the units of the value
info="${18}" # a short description of the alarm
# don't do anything if this is not WARNING, CRITICAL or CLEAR
if [ "${status}" != "WARNING" -a "${status}" != "CRITICAL" -a "${status}" != "CLEAR" ]
then
- echo >&2 "${me}: not sending notification for ${status} on '${chart}.${name}'"
+ info "not sending notification for ${status} on '${chart}.${name}'"
exit 1
fi
# don't do anything if this is CLEAR, but it was not WARNING or CRITICAL
if [ "${old_status}" != "WARNING" -a "${old_status}" != "CRITICAL" -a "${status}" = "CLEAR" ]
then
- echo >&2 "${me}: not sending notification for ${status} on '${chart}.${name}' (last status was ${old_status})"
+ info "not sending notification for ${status} on '${chart}.${name}' (last status was ${old_status})"
exit 1
fi
# -----------------------------------------------------------------------------
# load configuration
-# this is defined here so that private registries
-# can setup their own
-# images_base_url="${NETDATA_REGISTRY_URL}"
+# By default fetch images from the global public registry.
+# This is required by default, since all notification methods need to download
+# images via the Internet, and private registries might not be reachable.
+# This can be overwritten at the configuration file.
images_base_url="https://registry.my-netdata.io"
# needed commands
fi
# -----------------------------------------------------------------------------
-# filter recipients based on the criticality of each
+# filter a recipient based on alarm event severity
filter_recipient_by_criticality() {
local method="${1}" x="${2}" r s
shift
- r="${x/|*/}"
- s="${x/*|/}"
+ r="${x/|*/}" # the recipient
+ s="${x/*|/}" # the severity required for notifying this recipient
# no severity filtering for this person
[ "${r}" = "${s}" ] && return 0
}
# -----------------------------------------------------------------------------
-# find the recipient's addresses per method
+# find the recipients' addresses per method
declare -A arr_slack=()
declare -A arr_pushover=()
declare -A arr_telegram=()
declare -A arr_email=()
-# netdata may call us with multiple recipients
-# so, here we find the unique ones
-for x in ${recipient//,/ }
+# netdata may call us with multiple roles, and roles may have multiple but
+# overlapping recipients - so, here we find the unique recipients.
+for x in ${roles//,/ }
do
+ # the roles 'silent' and 'disabled' mean:
+ # don't send a notification for this role
+ [ "${x}" = "silent" -o "${x}" = "disabled" ] && continue
+
# email
- a="${role_recipients_email[${recipient}]}"
+ a="${role_recipients_email[${x}]}"
[ -z "${a}" ] && a="${DEFAULT_RECIPIENT_EMAIL}"
for r in ${a//,/ }
do
done
# pushover
- a="${role_recipients_pushover[${recipient}]}"
+ a="${role_recipients_pushover[${x}]}"
[ -z "${a}" ] && a="${DEFAULT_RECIPIENT_PUSHOVER}"
for r in ${a//,/ }
do
done
# telegram
- a="${role_recipients_telegram[${recipient}]}"
+ a="${role_recipients_telegram[${x}]}"
[ -z "${a}" ] && a="${DEFAULT_RECIPIENT_TELEGRAM}"
for r in ${a//,/ }
do
done
# slack
- a="${role_recipients_slack[${recipient}]}"
+ a="${role_recipients_slack[${x}]}"
[ -z "${a}" ] && a="${DEFAULT_RECIPIENT_SLACK}"
for r in ${a//,/ }
do
# check that we have at least a method enabled
if [ "${SEND_EMAIL}" != "YES" -a "${SEND_PUSHOVER}" != "YES" -a "${SEND_TELEGRAM}" != "YES" -a "${SEND_SLACK}" != "YES" ]
then
- echo >&2 "All notification methods are disabled. Not sending a notification."
- exit 1
+ fatal "All notification methods are disabled. Not sending a notification."
fi
# -----------------------------------------------------------------------------
if [ $ret -eq 0 ]
then
- echo >&2 "${me}: Sent email notification for: ${host} ${chart}.${name} is ${status} to '${to_email}'"
+ info "sent email notification for: ${host} ${chart}.${name} is ${status} to '${to_email}'"
return 0
else
- echo >&2 "${me}: Failed to send email notification for: ${host} ${chart}.${name} is ${status} to '${to_email}' with error code ${ret}."
+ error "failed to send email notification for: ${host} ${chart}.${name} is ${status} to '${to_email}' with error code ${ret}."
return 1
fi
fi
if [ "${httpcode}" == "200" ]
then
- echo >&2 "${me}: Sent pushover notification for: ${host} ${chart}.${name} is ${status} to '${user}'"
+ info "sent pushover notification for: ${host} ${chart}.${name} is ${status} to '${user}'"
sent=$((sent + 1))
else
- echo >&2 "${me}: Failed to send pushover notification for: ${host} ${chart}.${name} is ${status} to '${user}' with HTTP error code ${httpcode}."
+ error "failed to send pushover notification for: ${host} ${chart}.${name} is ${status} to '${user}' with HTTP error code ${httpcode}."
fi
done
if [ "${httpcode}" == "200" ]
then
- echo >&2 "${me}: Sent telegram notification for: ${host} ${chart}.${name} is ${status} to '${chatid}'"
+ info "sent telegram notification for: ${host} ${chart}.${name} is ${status} to '${chatid}'"
sent=$((sent + 1))
elif [ "${httpcode}" == "401" ]
then
- echo >&2 "${me}: Failed to send telegram notification for: ${host} ${chart}.${name} is ${status} to '${chatid}': Wrong bot token."
+ error "failed to send telegram notification for: ${host} ${chart}.${name} is ${status} to '${chatid}': Wrong bot token."
else
- echo >&2 "${me}: Failed to send telegram notification for: ${host} ${chart}.${name} is ${status} to '${chatid}' with HTTP error code ${httpcode}."
+ error "failed to send telegram notification for: ${host} ${chart}.${name} is ${status} to '${chatid}' with HTTP error code ${httpcode}."
fi
done
httpcode=$(${curl} --write-out %{http_code} --silent --output /dev/null -X POST --data-urlencode "payload=${payload}" "${webhook}")
if [ "${httpcode}" == "200" ]
then
- echo >&2 "${me}: Sent slack notification for: ${host} ${chart}.${name} is ${status} to '${channel}'"
+ info "sent slack notification for: ${host} ${chart}.${name} is ${status} to '${channel}'"
sent=$((sent + 1))
else
- echo >&2 "${me}: Failed to send slack notification for: ${host} ${chart}.${name} is ${status} to '${channel}', with HTTP error code ${httpcode}."
+ error "failed to send slack notification for: ${host} ${chart}.${name} is ${status} to '${channel}', with HTTP error code ${httpcode}."
fi
done
SENT_PUSHOVER=$?
+# -----------------------------------------------------------------------------
+# send the telegram.org message
+
# https://core.telegram.org/bots/api#formatting-options
-raised_for_paranthesis=" (${raised_for})"
-[ -z "$raised_for" ] && raised_for_paranthesis=""
-send_telegram "${TELEGRAM_BOT_TOKEN}" "${to_telegram}" "<b>${severity}, ${status_message}
-${chart} (${family})</b>
-<a href=\"${goto_url}\">${alarm}${raised_for_paranthesis}</a>
+send_telegram "${TELEGRAM_BOT_TOKEN}" "${to_telegram}" "${host} ${status_message} - <b>${name//_/ }</b>
+${chart} (${family})
+<a href=\"${goto_url}\">${alarm}</a>
<i>${info}</i>"
SENT_TELEGRAM=$?
# -----------------------------------------------------------------------------
# let netdata know
-# we did send somehting
+# we did send something
[ ${SENT_EMAIL} -eq 0 -o ${SENT_PUSHOVER} -eq 0 -o ${SENT_TELEGRAM} -eq 0 -o ${SENT_SLACK} -eq 0 ] && exit 0
# we did not send anything
projects / netdata.git / blobdiff