redhat init script for consul

gistfile1.sh

#!/bin/bash
#
# consul        Manage the consul agent
#       
# chkconfig:   2345 95 95
# description: Consul is a tool for service discovery and configuration
# processname: consul
# config: /etc/consul.conf
# pidfile: /var/run/consul.pid

### BEGIN INIT INFO
# Provides:       consul
# Required-Start: $local_fs $network
# Required-Stop:
# Should-Start:
# Should-Stop:
# Default-Start: 2 3 4 5
# Default-Stop:  0 1 6
# Short-Description: Manage the consul agent
# Description: Consul is a tool for service discovery and configuration
### END INIT INFO

# source function library
. /etc/rc.d/init.d/functions

prog="consul"
user="consul"
exec="/usr/local/bin/$prog"
pidfile="/var/run/$prog.pid"
lockfile="/var/lock/subsys/$prog"
logfile="/var/log/$prog"
conffile="/etc/consul.conf"
confdir="/etc/consul.d"

# pull in sysconfig settings
[ -e /etc/sysconfig/$prog ] && . /etc/sysconfig/$prog

export GOMAXPROCS=${GOMAXPROCS:-2}

start() {
    [ -x $exec ] || exit 5
    
    [ -f $conffile ] || exit 6
    [ -d $confdir ] || exit 6

    umask 077

    touch $logfile $pidfile
    chown $user:$user $logfile $pidfile

    echo -n $"Starting $prog: "
    
    ## holy shell shenanigans, batman!
    ## daemon can't be backgrounded.  we need the pid of the spawned process,
    ## which is actually done via runuser thanks to --user.  you can't do "cmd
    ## &; action" but you can do "{cmd &}; action".
    daemon \
        --pidfile=$pidfile \
        --user=consul \
        " { $exec agent -config-file=$conffile -config-dir=$confdir &>> $logfile & } ; echo \$! >| $pidfile "
    
    RETVAL=$?
    echo
    
    [ $RETVAL -eq 0 ] && touch $lockfile
    
    return $RETVAL
}

stop() {
    echo -n $"Shutting down $prog: "
    ## graceful shutdown with SIGINT
    killproc -p $pidfile $exec -INT
    RETVAL=$?
    echo
    [ $RETVAL -eq 0 ] && rm -f $lockfile
    return $RETVAL
}

restart() {
    stop
    start
}

reload() {
    echo -n $"Reloading $prog: "
    killproc -p $pidfile $exec -HUP
    echo
}

force_reload() {
    restart
}

rh_status() {
    status -p "$pidfile" -l $prog $exec
}

rh_status_q() {
    rh_status >/dev/null 2>&1
}

case "$1" in
    start)
        rh_status_q && exit 0
        $1
        ;;
    stop)
        rh_status_q || exit 0
        $1
        ;;
    restart)
        $1
        ;;
    reload)
        rh_status_q || exit 7
        $1
        ;;
    force-reload)
        force_reload
        ;;
    status)
        rh_status
        ;;
    condrestart|try-restart)
        rh_status_q || exit 0
        restart
        ;;
    *)
        echo $"Usage: $0 {start|stop|status|restart|condrestart|try-restart|reload|force-reload}"
        exit 2
esac

exit $?

When I use this init script the pidfile gets left behind when the stop function is called and consul is stopped. Should removal of the pidfile be handled in the script, or should killproc function be handling this?

Oh man, that daemon line in the start function is a life saver. I've already spent at least 5 hours trying to figure out how to run a program that does not background properly as a daemon. This is such a pain in the ass.

Line 5 should be # chkconfig: 2345 95 01. In this init script the consul service gets shutdown after networking, resulting in failed nodes. If consul is shutdown first then it will gracefully leave the cluster.

From /usr/share/doc/initscripts-*/sysvinitfiles:

# chkconfig: <startlevellist> <startpriority> <endpriority>

        (...)

        Unless there is a VERY GOOD, EXPLICIT reason to the
        contrary, the <endpriority> should be equal to
        100 - <startpriority>

So, line 5 should probably # chkconfig: 2345 95 05 instead?

@blalor

holy shell shenanigans, batman!

## daemon can't be backgrounded.  we need the pid of the spawned process,
## which is actually done via runuser thanks to --user.  you can't do "cmd
## &; action" but you can do "{cmd &}; action".

Sorry, but it doesn't work anyway (on CentOS 6.7):

# . /etc/init.d/functions
# daemon \
    --pidfile=/var/run/consul.pid \
    --user=consul \
    " { /usr/local/bin/consul agent -config-file=/etc/consul/consul.json -config-dir=/etc/consul/conf.d &>> /dev/null & } ; echo \$! >| /var/run/consul.pid"

bash: /var/run/consul.pid: Permission denied

Seems like the whole command between double quotes "" is running by the specified user

Thanks !
FYI: using your init script in https://github.com/torian/ansible-role-consul

Let me know if you have any issues with it.

Does the config file must named with .json end ?
If I use "-config-file=/etc/consul.conf" start consul, it stdout "data_dir cannot be empty" error, but then I change "consul.conf" to "consul.json ", it start success.

the content of config file like this:

{
"data_dir": "/consul/data",
"log_level": "INFO",
"bind_addr": "127.0.0.1",
"client_addr": "0.0.0.0",
"ui": true,
"server": true,
"bootstrap_expect": 1,
"acl_datacenter": "dc1",
"acl_default_policy": "deny"
}

Does the config file must named with .json end ?
If I use "-config-file=/etc/consul.conf" start consul, it stdout "data_dir cannot be empty" error, but then I change "consul.conf" to "consul.json ", it start success.

Thank you. Renaming to .json helped.

This script works fine while starting consul as sudo su on running instance.

But i got a problem on rebooting EC2 ECS instance - consul agent starting but then shutting down with error : Error starting agent: agent: timeout starting DNS servers.
I believe that it because only root user can listen 53 port, which I specified for consul agent in config:
...
"ports": {"dns": 53}
...

More over, 'sudo service consul status' gives 'consul dead but pid file exists'
Manual start 'sudo service consul start' helps, but I want an stable automatic start on reboot or stop/start instance.

Is their a right solution or what should I check to resolve this problem?

Thanks!

Your return value just after daemon in the start function will end up being the return code of echo \$! >| $pidfile and will report that your application started correctly when it actually fails