mailcow/data/Dockerfiles/watchdog/watchdog.sh

361 lines
13 KiB
Bash
Raw Normal View History

2017-09-20 16:56:49 +08:00
#!/bin/bash
trap "exit" INT TERM
trap "kill 0" EXIT
# Prepare
BACKGROUND_TASKS=()
2017-10-03 18:05:38 +08:00
if [[ "${USE_WATCHDOG}" =~ ^([nN][oO]|[nN])+$ ]]; then
2017-10-03 18:07:48 +08:00
echo -e "$(date) - USE_WATCHDOG=n, skipping watchdog..."
2017-10-03 18:05:38 +08:00
sleep 365d
exec $(readlink -f "$0")
fi
2017-09-20 16:56:49 +08:00
# Checks pipe their corresponding container name in this pipe
if [[ ! -p /tmp/com_pipe ]]; then
mkfifo /tmp/com_pipe
fi
# Common functions
progress() {
SERVICE=${1}
TOTAL=${2}
CURRENT=${3}
DIFF=${4}
[[ -z ${DIFF} ]] && DIFF=0
[[ -z ${TOTAL} || -z ${CURRENT} ]] && return
[[ ${CURRENT} -gt ${TOTAL} ]] && return
[[ ${CURRENT} -lt 0 ]] && CURRENT=0
2017-09-21 05:24:56 +08:00
PERCENT=$(( 200 * ${CURRENT} / ${TOTAL} % 2 + 100 * ${CURRENT} / ${TOTAL} ))
2017-12-09 20:15:24 +08:00
redis-cli -h redis LPUSH WATCHDOG_LOG "{\"time\":\"$(date +%s)\",\"service\":\"${SERVICE}\",\"lvl\":\"${PERCENT}\",\"hpnow\":\"${CURRENT}\",\"hptotal\":\"${TOTAL}\",\"hpdiff\":\"${DIFF}\"}" > /dev/null
log_msg "${SERVICE} health level: ${PERCENT}% (${CURRENT}/${TOTAL}), health trend: ${DIFF}" no_redis
2017-09-20 16:56:49 +08:00
}
log_msg() {
2017-12-09 20:15:24 +08:00
if [[ ${2} != "no_redis" ]]; then
redis-cli -h redis LPUSH WATCHDOG_LOG "{\"time\":\"$(date +%s)\",\"message\":\"$(printf '%s' "${1}" | \
tr '%&;$"_[]{}-\r\n' ' ')\"}" > /dev/null
fi
redis-cli -h redis LTRIM WATCHDOG_LOG 0 ${LOG_LINES} > /dev/null
echo $(date) $(printf '%s\n' "${1}")
}
function mail_error() {
[[ -z ${1} ]] && return 1
[[ -z ${2} ]] && return 2
RCPT_DOMAIN=$(echo ${1} | awk -F @ {'print $NF'})
RCPT_MX=$(dig +short ${RCPT_DOMAIN} mx | sort -n | awk '{print $2; exit}')
if [[ -z ${RCPT_MX} ]]; then
log_msg "Cannot determine MX for ${1}, skipping email notification..."
return 1
fi
./smtp-cli --missing-modules-ok \
--subject="Watchdog: ${2} service hit the error rate limit" \
--body-plain="Service was restarted, please check your mailcow installation." \
--to=${1} \
--from="watchdog@${MAILCOW_HOSTNAME}" \
--server="${RCPT_MX}" \
--hello-host=${MAILCOW_HOSTNAME}
log_msg "Sent notification email to ${1}"
}
2017-09-20 18:27:24 +08:00
get_container_ip() {
# ${1} is container
CONTAINER_ID=
CONTAINER_IP=
2017-09-21 05:24:56 +08:00
LOOP_C=1
until [[ ${CONTAINER_IP} =~ ^[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}$ ]] || [[ ${LOOP_C} -gt 5 ]]; do
2017-09-20 18:27:24 +08:00
sleep 1
CONTAINER_ID=$(curl --silent http://dockerapi:8080/containers/json | jq -r ".[] | {name: .Config.Labels[\"com.docker.compose.service\"], id: .Id}" | jq -rc "select( .name | tostring | contains(\"${1}\")) | .id")
if [[ ! -z ${CONTAINER_ID} ]]; then
CONTAINER_IP=$(curl --silent http://dockerapi:8080/containers/${CONTAINER_ID}/json | jq -r '.NetworkSettings.Networks[].IPAddress')
fi
2017-09-21 05:24:56 +08:00
LOOP_C=$((LOOP_C + 1))
2017-09-20 18:27:24 +08:00
done
2017-09-21 05:24:56 +08:00
[[ ${LOOP_C} -gt 5 ]] && echo 240.0.0.0 || echo ${CONTAINER_IP}
2017-09-20 18:27:24 +08:00
}
2017-09-20 16:56:49 +08:00
# Check functions
nginx_checks() {
err_count=0
diff_c=0
THRESHOLD=16
# Reduce error count by 2 after restarting an unhealthy container
trap "[ ${err_count} -gt 1 ] && err_count=$(( ${err_count} - 2 ))" USR1
while [ ${err_count} -lt ${THRESHOLD} ]; do
2017-09-21 05:24:56 +08:00
host_ip=$(get_container_ip nginx-mailcow)
2017-09-20 16:56:49 +08:00
err_c_cur=${err_count}
2017-09-21 05:24:56 +08:00
/usr/lib/nagios/plugins/check_ping -4 -H ${host_ip} -w 2000,10% -c 4000,100% -p2 1>&2; err_count=$(( ${err_count} + $? ))
/usr/lib/nagios/plugins/check_http -4 -H ${host_ip} -u / -p 8081 1>&2; err_count=$(( ${err_count} + $? ))
2017-09-20 16:56:49 +08:00
[ ${err_c_cur} -eq ${err_count} ] && [ ! $((${err_count} - 1)) -lt 0 ] && err_count=$((${err_count} - 1)) diff_c=1
[ ${err_c_cur} -ne ${err_count} ] && diff_c=$(( ${err_c_cur} - ${err_count} ))
progress "Nginx" ${THRESHOLD} $(( ${THRESHOLD} - ${err_count} )) ${diff_c}
2017-09-22 22:40:02 +08:00
diff_c=0
2017-09-21 05:24:56 +08:00
sleep $(( ( RANDOM % 30 ) + 10 ))
2017-09-20 16:56:49 +08:00
done
return 1
}
mysql_checks() {
err_count=0
diff_c=0
THRESHOLD=12
# Reduce error count by 2 after restarting an unhealthy container
trap "[ ${err_count} -gt 1 ] && err_count=$(( ${err_count} - 2 ))" USR1
while [ ${err_count} -lt ${THRESHOLD} ]; do
2017-09-21 05:24:56 +08:00
host_ip=$(get_container_ip mysql-mailcow)
2017-09-20 16:56:49 +08:00
err_c_cur=${err_count}
2017-09-21 05:24:56 +08:00
/usr/lib/nagios/plugins/check_mysql -H ${host_ip} -P 3306 -u ${DBUSER} -p ${DBPASS} -d ${DBNAME} 1>&2; err_count=$(( ${err_count} + $? ))
/usr/lib/nagios/plugins/check_mysql_query -H ${host_ip} -P 3306 -u ${DBUSER} -p ${DBPASS} -d ${DBNAME} -q "SELECT COUNT(*) FROM information_schema.tables" 1>&2; err_count=$(( ${err_count} + $? ))
2017-09-20 16:56:49 +08:00
[ ${err_c_cur} -eq ${err_count} ] && [ ! $((${err_count} - 1)) -lt 0 ] && err_count=$((${err_count} - 1)) diff_c=1
[ ${err_c_cur} -ne ${err_count} ] && diff_c=$(( ${err_c_cur} - ${err_count} ))
progress "MySQL/MariaDB" ${THRESHOLD} $(( ${THRESHOLD} - ${err_count} )) ${diff_c}
2017-09-22 22:40:02 +08:00
diff_c=0
2017-09-21 05:24:56 +08:00
sleep $(( ( RANDOM % 30 ) + 10 ))
2017-09-20 16:56:49 +08:00
done
return 1
}
sogo_checks() {
err_count=0
diff_c=0
THRESHOLD=20
# Reduce error count by 2 after restarting an unhealthy container
trap "[ ${err_count} -gt 1 ] && err_count=$(( ${err_count} - 2 ))" USR1
while [ ${err_count} -lt ${THRESHOLD} ]; do
2017-09-21 05:24:56 +08:00
host_ip=$(get_container_ip sogo-mailcow)
2017-09-20 16:56:49 +08:00
err_c_cur=${err_count}
2017-09-21 05:24:56 +08:00
/usr/lib/nagios/plugins/check_http -4 -H ${host_ip} -u /WebServerResources/css/theme-default.css -p 9192 -R md-default-theme 1>&2; err_count=$(( ${err_count} + $? ))
/usr/lib/nagios/plugins/check_http -4 -H ${host_ip} -u /SOGo.index/ -p 20000 -R "SOGo\.MainUI" 1>&2; err_count=$(( ${err_count} + $? ))
2017-09-20 16:56:49 +08:00
[ ${err_c_cur} -eq ${err_count} ] && [ ! $((${err_count} - 1)) -lt 0 ] && err_count=$((${err_count} - 1)) diff_c=1
[ ${err_c_cur} -ne ${err_count} ] && diff_c=$(( ${err_c_cur} - ${err_count} ))
progress "SOGo" ${THRESHOLD} $(( ${THRESHOLD} - ${err_count} )) ${diff_c}
2017-09-22 22:40:02 +08:00
diff_c=0
2017-09-21 05:24:56 +08:00
sleep $(( ( RANDOM % 30 ) + 10 ))
2017-09-20 16:56:49 +08:00
done
return 1
}
postfix_checks() {
err_count=0
diff_c=0
THRESHOLD=16
# Reduce error count by 2 after restarting an unhealthy container
trap "[ ${err_count} -gt 1 ] && err_count=$(( ${err_count} - 2 ))" USR1
while [ ${err_count} -lt ${THRESHOLD} ]; do
host_ip=$(get_container_ip postfix-mailcow)
2017-09-20 16:56:49 +08:00
err_c_cur=${err_count}
/usr/lib/nagios/plugins/check_smtp -4 -H ${host_ip} -p 589 -f "watchdog@invalid" -C "RCPT TO:null@localhost" -C DATA -C . -R 250 1>&2; err_count=$(( ${err_count} + $? ))
/usr/lib/nagios/plugins/check_smtp -4 -H ${host_ip} -p 589 -S 1>&2; err_count=$(( ${err_count} + $? ))
2017-09-20 16:56:49 +08:00
[ ${err_c_cur} -eq ${err_count} ] && [ ! $((${err_count} - 1)) -lt 0 ] && err_count=$((${err_count} - 1)) diff_c=1
[ ${err_c_cur} -ne ${err_count} ] && diff_c=$(( ${err_c_cur} - ${err_count} ))
progress "Postfix" ${THRESHOLD} $(( ${THRESHOLD} - ${err_count} )) ${diff_c}
2017-09-22 22:40:02 +08:00
diff_c=0
2017-09-21 05:24:56 +08:00
sleep $(( ( RANDOM % 30 ) + 10 ))
2017-09-20 16:56:49 +08:00
done
return 1
}
dovecot_checks() {
err_count=0
diff_c=0
THRESHOLD=24
# Reduce error count by 2 after restarting an unhealthy container
trap "[ ${err_count} -gt 1 ] && err_count=$(( ${err_count} - 2 ))" USR1
while [ ${err_count} -lt ${THRESHOLD} ]; do
2017-09-21 05:24:56 +08:00
host_ip=$(get_container_ip dovecot-mailcow)
2017-09-20 16:56:49 +08:00
err_c_cur=${err_count}
/usr/lib/nagios/plugins/check_smtp -4 -H ${host_ip} -p 24 -f "watchdog@invalid" -C "RCPT TO:<watchdog@invalid>" -L -R "User doesn't exist" 1>&2; err_count=$(( ${err_count} + $? ))
2017-09-21 05:24:56 +08:00
/usr/lib/nagios/plugins/check_imap -4 -H ${host_ip} -p 993 -S -e "OK " 1>&2; err_count=$(( ${err_count} + $? ))
/usr/lib/nagios/plugins/check_imap -4 -H ${host_ip} -p 143 -e "OK " 1>&2; err_count=$(( ${err_count} + $? ))
/usr/lib/nagios/plugins/check_tcp -4 -H ${host_ip} -p 10001 -e "VERSION" 1>&2; err_count=$(( ${err_count} + $? ))
/usr/lib/nagios/plugins/check_tcp -4 -H ${host_ip} -p 4190 -e "Dovecot ready" 1>&2; err_count=$(( ${err_count} + $? ))
2017-09-20 16:56:49 +08:00
[ ${err_c_cur} -eq ${err_count} ] && [ ! $((${err_count} - 1)) -lt 0 ] && err_count=$((${err_count} - 1)) diff_c=1
[ ${err_c_cur} -ne ${err_count} ] && diff_c=$(( ${err_c_cur} - ${err_count} ))
progress "Dovecot" ${THRESHOLD} $(( ${THRESHOLD} - ${err_count} )) ${diff_c}
2017-09-22 22:40:02 +08:00
diff_c=0
2017-09-21 05:24:56 +08:00
sleep $(( ( RANDOM % 30 ) + 10 ))
2017-09-20 16:56:49 +08:00
done
return 1
}
phpfpm_checks() {
err_count=0
diff_c=0
2017-09-21 05:24:56 +08:00
THRESHOLD=10
2017-09-20 16:56:49 +08:00
# Reduce error count by 2 after restarting an unhealthy container
trap "[ ${err_count} -gt 1 ] && err_count=$(( ${err_count} - 2 ))" USR1
while [ ${err_count} -lt ${THRESHOLD} ]; do
2017-09-21 05:24:56 +08:00
host_ip=$(get_container_ip php-fpm-mailcow)
2017-09-20 16:56:49 +08:00
err_c_cur=${err_count}
nc -z ${host_ip} 9001 ; err_count=$(( ${err_count} + ($? * 2)))
nc -z ${host_ip} 9002 ; err_count=$(( ${err_count} + ($? * 2)))
2017-09-21 05:24:56 +08:00
/usr/lib/nagios/plugins/check_ping -4 -H ${host_ip} -w 2000,10% -c 4000,100% -p2 1>&2; err_count=$(( ${err_count} + $? ))
2017-09-20 16:56:49 +08:00
[ ${err_c_cur} -eq ${err_count} ] && [ ! $((${err_count} - 1)) -lt 0 ] && err_count=$((${err_count} - 1)) diff_c=1
[ ${err_c_cur} -ne ${err_count} ] && diff_c=$(( ${err_c_cur} - ${err_count} ))
progress "PHP-FPM" ${THRESHOLD} $(( ${THRESHOLD} - ${err_count} )) ${diff_c}
2017-09-22 22:40:02 +08:00
diff_c=0
2017-09-21 05:24:56 +08:00
sleep $(( ( RANDOM % 30 ) + 10 ))
2017-09-20 16:56:49 +08:00
done
return 1
}
rspamd_checks() {
err_count=0
diff_c=0
THRESHOLD=10
# Reduce error count by 2 after restarting an unhealthy container
trap "[ ${err_count} -gt 1 ] && err_count=$(( ${err_count} - 2 ))" USR1
while [ ${err_count} -lt ${THRESHOLD} ]; do
host_ip=$(get_container_ip rspamd-mailcow)
err_c_cur=${err_count}
SCORE=$(/usr/bin/curl -s --data-binary @- --unix-socket /rspamd-sock/rspamd.sock http://rspamd/scan -d '
To: null@localhost
From: watchdog@localhost
Empty
' | jq -rc .required_score)
if [[ ${SCORE} != "9999" ]]; then
echo "Rspamd settings check failed" 1>&2
err_count=$(( ${err_count} + 1))
else
echo "Rspamd settings check succeeded" 1>&2
fi
/usr/lib/nagios/plugins/check_ping -4 -H ${host_ip} -w 2000,10% -c 4000,100% -p2 1>&2; err_count=$(( ${err_count} + $? ))
[ ${err_c_cur} -eq ${err_count} ] && [ ! $((${err_count} - 1)) -lt 0 ] && err_count=$((${err_count} - 1)) diff_c=1
[ ${err_c_cur} -ne ${err_count} ] && diff_c=$(( ${err_c_cur} - ${err_count} ))
progress "Rspamd" ${THRESHOLD} $(( ${THRESHOLD} - ${err_count} )) ${diff_c}
2017-09-22 22:40:02 +08:00
diff_c=0
sleep $(( ( RANDOM % 30 ) + 10 ))
done
return 1
}
2017-09-20 16:56:49 +08:00
# Create watchdog agents
(
while true; do
if ! nginx_checks; then
log_msg "Nginx hit error limit"
[[ ! -z ${WATCHDOG_NOTIFY_EMAIL} ]] && mail_error "${WATCHDOG_NOTIFY_EMAIL}" "nginx-mailcow"
2017-09-20 16:56:49 +08:00
echo nginx-mailcow > /tmp/com_pipe
fi
done
) &
BACKGROUND_TASKS+=($!)
(
while true; do
if ! mysql_checks; then
log_msg "MySQL hit error limit"
[[ ! -z ${WATCHDOG_NOTIFY_EMAIL} ]] && mail_error "${WATCHDOG_NOTIFY_EMAIL}" "mysql-mailcow"
2017-09-20 16:56:49 +08:00
echo mysql-mailcow > /tmp/com_pipe
fi
done
) &
BACKGROUND_TASKS+=($!)
(
while true; do
if ! phpfpm_checks; then
log_msg "PHP-FPM hit error limit"
[[ ! -z ${WATCHDOG_NOTIFY_EMAIL} ]] && mail_error "${WATCHDOG_NOTIFY_EMAIL}" "php-fpm-mailcow"
2017-09-20 16:56:49 +08:00
echo php-fpm-mailcow > /tmp/com_pipe
fi
done
) &
BACKGROUND_TASKS+=($!)
(
while true; do
if ! sogo_checks; then
log_msg "SOGo hit error limit"
[[ ! -z ${WATCHDOG_NOTIFY_EMAIL} ]] && mail_error "${WATCHDOG_NOTIFY_EMAIL}" "sogo-mailcow"
2017-09-20 16:56:49 +08:00
echo sogo-mailcow > /tmp/com_pipe
fi
done
) &
BACKGROUND_TASKS+=($!)
(
while true; do
if ! postfix_checks; then
log_msg "Postfix hit error limit"
[[ ! -z ${WATCHDOG_NOTIFY_EMAIL} ]] && mail_error "${WATCHDOG_NOTIFY_EMAIL}" "postfix-mailcow"
2017-09-20 16:56:49 +08:00
echo postfix-mailcow > /tmp/com_pipe
fi
done
) &
BACKGROUND_TASKS+=($!)
(
while true; do
if ! dovecot_checks; then
log_msg "Dovecot hit error limit"
[[ ! -z ${WATCHDOG_NOTIFY_EMAIL} ]] && mail_error "${WATCHDOG_NOTIFY_EMAIL}" "dovecot-mailcow"
2017-09-20 16:56:49 +08:00
echo dovecot-mailcow > /tmp/com_pipe
fi
done
) &
BACKGROUND_TASKS+=($!)
(
while true; do
if ! rspamd_checks; then
log_msg "Rspamd hit error limit"
[[ ! -z ${WATCHDOG_NOTIFY_EMAIL} ]] && mail_error "${WATCHDOG_NOTIFY_EMAIL}" "rspamd-mailcow"
echo rspamd-mailcow > /tmp/com_pipe
fi
done
) &
BACKGROUND_TASKS+=($!)
2017-09-20 16:56:49 +08:00
# Monitor watchdog agents, stop script when agents fails and wait for respawn by Docker (restart:always:n)
(
while true; do
for bg_task in ${BACKGROUND_TASKS[*]}; do
if ! kill -0 ${bg_task} 1>&2; then
log_msg "Worker ${bg_task} died, stopping watchdog and waiting for respawn..."
kill -TERM 1
2017-09-20 16:56:49 +08:00
fi
sleep 10
2017-09-20 16:56:49 +08:00
done
done
) &
# Monitor dockerapi
(
while true; do
while nc -z dockerapi 8080; do
sleep 3
done
log_msg "Cannot find dockerapi-mailcow, waiting to recover..."
kill -STOP ${BACKGROUND_TASKS[*]}
until nc -z dockerapi 8080; do
sleep 3
done
kill -CONT ${BACKGROUND_TASKS[*]}
kill -USR1 ${BACKGROUND_TASKS[*]}
done
) &
2017-09-20 16:56:49 +08:00
# Restart container when threshold limit reached
while true; do
CONTAINER_ID=
read com_pipe_answer </tmp/com_pipe
if [[ ${com_pipe_answer} =~ .+-mailcow ]]; then
kill -STOP ${BACKGROUND_TASKS[*]}
sleep 3
CONTAINER_ID=$(curl --silent http://dockerapi:8080/containers/json | jq -r ".[] | {name: .Config.Labels[\"com.docker.compose.service\"], id: .Id}" | jq -rc "select( .name | tostring | contains(\"${com_pipe_answer}\")) | .id")
2017-09-20 16:56:49 +08:00
if [[ ! -z ${CONTAINER_ID} ]]; then
log_msg "Sending restart command to ${CONTAINER_ID}..."
curl --silent -XPOST http://dockerapi:8080/containers/${CONTAINER_ID}/restart
2017-09-20 16:56:49 +08:00
fi
log_msg "Wait for restarted container to settle and continue watching..."
2017-09-20 16:56:49 +08:00
sleep 30s
kill -CONT ${BACKGROUND_TASKS[*]}
kill -USR1 ${BACKGROUND_TASKS[*]}
fi
done