Run a Java Application as a Service on Linux

问题:

I have written a Java server application that runs on a standard virtual hosted Linux solution. The application runs all the time listening for socket connections and creating new handlers for them. It is a server side implementation to a client-server application.

The way I start it is by including it in the start up rc.local script of the server. However once started I do not know how to access it to stop it and if I want to install an update, so I have to restart the server in order to restart the application.

On a windows PC, for this type of application I might create a windows service and then I can stop and start it as I want. Is there anything like that on a Linux box so that if I start this application I can stop it and restart it without doing a complete restart of the server.

My application is called WebServer.exe. It is started on server startup by including it in my rc.local as such:

java -jar /var/www/vhosts/myweb.com/phpserv/WebServer.jar &

I am a bit of a noob at Linux so any example would be appreciated with any posts. However I do have SSH, and full FTP access to the box to install any updates as well as access to a Plesk panel.

回答1:

I wrote another simple wrapper here:

#!/bin/sh
SERVICE_NAME=MyService
PATH_TO_JAR=/usr/local/MyProject/MyJar.jar
PID_PATH_NAME=/tmp/MyService-pid
case $1 in
    start)
        echo \"Starting $SERVICE_NAME ...\"
        if [ ! -f $PID_PATH_NAME ]; then
            nohup java -jar $PATH_TO_JAR /tmp 2>> /dev/null >> /dev/null &
            echo $! > $PID_PATH_NAME
            echo \"$SERVICE_NAME started ...\"
        else
            echo \"$SERVICE_NAME is already running ...\"
        fi
    ;;
    stop)
        if [ -f $PID_PATH_NAME ]; then
            PID=$(cat $PID_PATH_NAME);
            echo \"$SERVICE_NAME stoping ...\"
            kill $PID;
            echo \"$SERVICE_NAME stopped ...\"
            rm $PID_PATH_NAME
        else
            echo \"$SERVICE_NAME is not running ...\"
        fi
    ;;
    restart)
        if [ -f $PID_PATH_NAME ]; then
            PID=$(cat $PID_PATH_NAME);
            echo \"$SERVICE_NAME stopping ...\";
            kill $PID;
            echo \"$SERVICE_NAME stopped ...\";
            rm $PID_PATH_NAME
            echo \"$SERVICE_NAME starting ...\"
            nohup java -jar $PATH_TO_JAR /tmp 2>> /dev/null >> /dev/null &
            echo $! > $PID_PATH_NAME
            echo \"$SERVICE_NAME started ...\"
        else
            echo \"$SERVICE_NAME is not running ...\"
        fi
    ;;
esac 

You can follow a full tutorial for init.d here and for systemd (ubuntu 16+) here

If you need the output log replace the 2

nohup java -jar $PATH_TO_JAR /tmp 2>> /dev/null >> /dev/null &

lines for

nohup java -jar $PATH_TO_JAR >> myService.out 2>&1&

回答2:

A simple solution is to create a script start.sh that runs Java through nohup and then stores the PID to a file:

nohup java -jar myapplication.jar > log.txt 2> errors.txt < /dev/null &
PID=$!
echo $PID > pid.txt

Then your stop script stop.sh would read the PID from the file and kill the application:

PID=$(cat pid.txt)
kill $PID

Of course I\'ve left out some details, like checking whether the process exists and removing pid.txt if you\'re done.

回答3:

Linux service init script are stored into /etc/init.d. You can copy and customize /etc/init.d/skeleton file, and then call

service [yourservice] start|stop|restart

see http://www.ralfebert.de/blog/java/debian_daemon/. Its for Debian (so, Ubuntu as well) but fit more distribution.

回答4:

Maybe not the best dev-ops solution, but good for the general use of a server for a lan party or similar.

Use screen to run your server in and then detach before logging out, this will keep the process running, you can then re-attach at any point.

Workflow:

Start a screen: screen

Start your server: java -jar minecraft-server.jar

Detach by pressing: Ctl-a, d

Re-attach: screen -r

More info here: https://www.gnu.org/software/screen/manual/screen.html

回答5:

Another alternative, which is also quite popular is the Java Service Wrapper. This is also quite popular around the OSS community.

回答6:

Referring to Spring Boot application as a Service as well, I would go for the systemd version, since it\'s the easiest, least verbose, and best integrated into modern distros (and even the not-so-modern ones like CentOS 7.x).

回答7:

The easiest way is to use supervisord. Please see full details here: http://supervisord.org/

More info:

https://askubuntu.com/questions/779830/running-an-executable-jar-file-when-the-system-starts/852485#852485

https://www.digitalocean.com/community/tutorials/how-to-install-and-manage-supervisor-on-ubuntu-and-debian-vps

回答8:

Here is a sample shell script (make sure you replace the MATH name with the name of the your application):

#!/bin/bash

### BEGIN INIT INFO
# Provides:                 MATH
# Required-Start:           $java
# Required-Stop:            $java
# Short-Description:        Start and stop MATH service.
# Description:              -
# Date-Creation:            -
# Date-Last-Modification:   -
# Author:                   -
### END INIT INFO

# Variables
PGREP=/usr/bin/pgrep
JAVA=/usr/bin/java
ZERO=0

# Start the MATH
start() {
    echo \"Starting MATH...\"
    #Verify if the service is running
    $PGREP -f MATH > /dev/null
    VERIFIER=$?
    if [ $ZERO = $VERIFIER ]
    then
        echo \"The service is already running\"
    else
        #Run the jar file MATH service
        $JAVA -jar /opt/MATH/MATH.jar > /dev/null 2>&1 &
        #sleep time before the service verification
        sleep 10
        #Verify if the service is running
        $PGREP -f MATH  > /dev/null
        VERIFIER=$?
        if [ $ZERO = $VERIFIER ]
        then
            echo \"Service was successfully started\"
        else
            echo \"Failed to start service\"
        fi
    fi
    echo
}

# Stop the MATH
stop() {
    echo \"Stopping MATH...\"
    #Verify if the service is running
    $PGREP -f MATH > /dev/null
    VERIFIER=$?
    if [ $ZERO = $VERIFIER ]
    then
        #Kill the pid of java with the service name
        kill -9 $($PGREP -f MATH)
        #Sleep time before the service verification
        sleep 10
        #Verify if the service is running
        $PGREP -f MATH  > /dev/null
        VERIFIER=$?
        if [ $ZERO = $VERIFIER ]
        then
            echo \"Failed to stop service\"
        else
            echo \"Service was successfully stopped\"
        fi
    else
        echo \"The service is already stopped\"
    fi
    echo
}

# Verify the status of MATH
status() {
    echo \"Checking status of MATH...\"
    #Verify if the service is running
    $PGREP -f MATH > /dev/null
    VERIFIER=$?
    if [ $ZERO = $VERIFIER ]
    then
        echo \"Service is running\"
    else
        echo \"Service is stopped\"
    fi
    echo
}

# Main logic
case \"$1\" in
    start)
        start
        ;;
    stop)
        stop
        ;;
    status)
        status
        ;;
    restart|reload)
        stop
        start
        ;;
  *)
    echo $\"Usage: $0 {start|stop|status|restart|reload}\"
    exit 1
esac
exit 0

回答9:

From Spring Boot application as a Service, I can recommend the Python-based supervisord application. See that stack overflow question for more information. It\'s really straightforward to set up.

回答10:

You can use Thrift server or JMX to communicate with your Java service.

回答11:

Other answers do a good job giving custom scripts and setups depending on your platform. In addition to those, here are the mature, special purpose programs that I know of:

• JSW from TanukiSoftware
• YAJSW is an open source clone from the above. It is written in Java, and it is a nanny process that manages the child process (your code) according to configurations. Works on windows / linux.
• JSVC is a native application. Its also a nanny process, but it invokes your child application through the JNI, rather than as a subprocess.

回答12:

To run Java code as daemon (service) you can write JNI based stub.

http://jnicookbook.owsiak.org/recipe-no-022/

for a sample code that is based on JNI. In this case you daemonize the code that was started as Java and main loop is executed in C. But it is also possible to put main, daemon\'s, service loop inside Java.

https://github.com/mkowsiak/jnicookbook/tree/master/recipeNo029

Have fun with JNI!

回答13:

However once started I don\'t know how to access it to stop it

You can write a simple stop script that greps for your java process, extracts the PID and calls kill on it. It\'s not fancy, but it\'s straight forward. Something like that may be of help as a start:

#!/bin/bash
PID = ps ax | grep \"name of your app\" | cut -d \' \' -f 1
kill $PID

回答14:

From Spring Boot Reference Guide

Installation as an init.d service (System V)

Simply symlink the jar to init.d to support the standard start, stop, restart and status commands. Assuming that you have a Spring Boot application installed in /var/myapp, to install a Spring Boot application as an init.d service simply create a symlink:

$ sudo ln -s /var/myapp/myapp.jar /etc/init.d/myapp

Once installed, you can start and stop the service in the usual way. For example, on a Debian based system:

$ service myapp start

If your application fails to start, check the log file written to /var/log/<appname>.log for errors.

Continue reading to know how to secure a deployed service.

After doing as written I\'ve discovered that my service fails to start with this error message in logs: start-stop-daemon: unrecognized option --no-close. And I\'ve managed to fix it by creating a config file /var/myapp/myapp.conf with the following content

USE_START_STOP_DAEMON=false