How do I start/stop failed upstart job?

Question

I have an upstart script in /etc/init/myexample.conf. This job basically runs a java jar file. Normally, I can start and stop this using sudo start myexample and sudo stop myexample.

I changed the jar file and introduced an error that caused the jar to crash. After starting the jar with sudo start myexample, I see the errors in my log file.

Now I would like to stop the upstart job and replace the jar file with a working one again. I have seen two scenarios:

1. sudo stop myexample hangs indefinitely (at least 10 minutes). I also cannot start the job again with sudo start myexample, I receive the error start: Job is already running: myexample.

2. sudo start myexample hangs indefinitely, and sudo stop myexample gives stop: Job has already been stopped: myexample.

I can also run the jar manually from the command line.

pgrep -lf myexample and pgrep -l myexample return nothing. pgrep -l java shows other java processes that are running, but not the failed one corresponding to myexample. Upstart thinks the process is running, but it is not.

How can I get the upstart job running again, without a reboot?

---

/etc/init/myexample.conf:

description "myexample"  

start on runlevel [2345]  
stop on shutdown  

respawn
respawn limit 3 120

expect daemon

setuid exampleuser

script     
    chdir /home/exampleuser/ 
    java -jar /home/exampleuser/example.jar
end script

Answer 1

Kill the process.

You can kill the process (if running) by:

pkill <process_name>

Or

killall <process_name>

This will match the process name only, if you want to match the whole command line, use (with extreme care):

pkill -f <part_of_the_command>

---

So in you case:

sudo pkill myexample

should do.

You can be sure beforehand, by running:

pgrep -l myexample

to check the process ID to make sure no other process coincides with the name.

On a different note, after finding the PID with pgrep -l myexample, you can kill the process with kill too:

kill PID