Bash – Find the process id of a java application in a bash script (to see if the target application is already running)

bashbash-scriptjavaprocess-managementshell-script

I know there are a million questions on getting the process ID, but this one seems to be unique. Google has not given me the answer, so I hope stackexhange will help rather than close this question.

When Java is involved it seems trickier to find a process ID (pgrep doesn't work afaik).

Furthermore, I need to automate this in a bash script. One issue I've encountered is that when I use ps aux | grep the grep process itself always shows up, so handling the results in a simple bash script is not trivial enough for me to figure out a good solution on my own (with my limited bash skills).

Some things I have tried:

Example 1 – this returns a process even though there is no application by that name:

$ ps aux | grep -i anythingnotreal
user2   3040  0.0  0.0   4640   856 pts/3    S+   18:17   0:00 grep --color=auto -i anythingnotreal

Example 2 – this returns nothing even though "java_app" is currently running:

$ pgrep java_app

It returns nothing. However, here's proof that "java_app" is running:

$ ps aux | grep java_app
tester2   2880  0.7  2.8 733196 58444 ?        Sl   18:02   0:07 java -jar /opt/java_app2/my_java_app.jar
tester2   3058  0.0  0.0   4644   844 pts/3    S+   18:19   0:00 grep --color=auto java_app

What I need is a solution I can plug into a bash script that will tell me if the java application of interest (for which I know the jar file name and path) is currently running. (If it is running, I need to ask the user to close it before my script continues.)

Best Answer

By default, pgrep only matches the command, not the arguements. To match the full command line, you need the -f option.

$ pgrep -f java_app

From the pgrep manpage:

-f The pattern is normally only matched against the process name. When -f is set, the full command line is used

Related Solutions

Bash – How to check with a Bash script if an application is running

This shell script should handle the starting and stopping of any program:

#!/bin/bash

BASECMD=${1%%\ *}
PID=$(pgrep "$BASECMD")
if [ "$?" -eq "0" ]; then
    echo "at least one instance of "$BASECMD" found, killing all instances"
    kill $PID
else
    echo "no running instances of "$BASECMD" found, starting one"
    $1
 fi

let's say you saved it under ~/mystarter, you can run any command with it using ~/mystarter <name>, eg in your case, bind Meta+R to:

~/mystarter gnome-run

and make sure the script is executable: chmod u+x ~/mystarter. Also it's probably best to put it somewhere in your PATH, so you don't have to type it's full location every time.

As for the fact that gnome-run doesn't show up in ps -A, make sure that gnome run itself isn't a script that launches the actual process. Check if there is a difference between ps -A | wc -l before and after launching it (this counts all running processes).

Edit:

Since you've accepted the answer, I thought I'd add support for running commands that have commandline arguments, so that this might become a place of reference. Run a command like so:

 ./mystarter 'cmd args'

eg:

./mystarter 'ncmpcpp -c ~/.ncmpcpp'

The command just looks up ncmpcpp to see if it's running already, but executes the full command (with arguments) when ncmpcpp wasn't running.

Shell script hangs on mail command

The mail program expects the user to type the message on its standard input (terminated by EOF (Ctrl-D)). You are not redirecting stdin, the program is waiting for input. You can either redirect from /dev/null (for an empty message), from a canned file, or from a pipe, for example:

echo Hi, just sending you a message | mail -s "pdfbackup" "$EMAIL_TO"

mail -s "pdfbackup" "$EMAIL_TO" < /dev/null

Both of these should clear up problem.

Best Answer

Related Solutions

Bash – How to check with a Bash script if an application is running

Edit:

Shell script hangs on mail command

Related Question