Deployment of a Spring Boot App with Jenkins in a Unix System using Gradle
One thing that I found very hard to do was to integrate a spring boot project into a CI environment using jenkins. As a default behavior, the Jenkins process tree killer always kills the process started by a job which stops the execution of a Spring Boot App after the jenkins job finishes. In addition of that, I wanted to see the server log on the jenkyns windows until it finishes loading. This article will try to help us solving this problems.
But first I would like to discuss what I consider a good practice to a Spring Boot App CI environment. I find very useful to first copy the artifacts to a specified area on the server to keep track of the artifacts deployed and deploy the artifact from that location. Also, I create a server log file there and start to listening on the jenkins window until the server started.
So the script below does that. With some minor improvements self explained on the comments, but in summary it does this:
- stop whatever process running on the deployed port.
- delete the files of the previous deploy
- copy the files to deploy location
- start application with nohup command, java – jar
- start listening to the server log until it reaches an specific instruction.
Finally you have to do some adjustments to your job on Jenkins to avoid the default tree killing process. Just add this instruction before calling the sh: BUILD_ID=dontKillMe /path/to/my/script.sh (FIGURE 3)
You can see the jenkins job configuration window on FIGURES 1, 2, and 3 and the log result window on FIGURES 4 and 5.
Go to my github repo to check the project, but it recommend to extract the shell script to another repo to keep it lifecycle independent of your app.