How to deploy a quarkus application to a openshift namespace using the rolling deployment strategy?
In my case always the the previous pod will be tear down first.
I would expect some property to define the used deployment strategy.
Currently I use maven to deploy the application
mvn clean package -Dquarkus.kubernetes.deploy=true -Dquarkus.openshift.namespace=my_target_namespace
Related
I prefer to use gradle with Spring Boot. I have found resources about deploying to Azure with the Maven Plugin for Azure App Service Web Apps. How would one take an approach with gradle as a build system?
Well you need to separate deployment and build.
You can use Azure build pipeline task to build with gradle.
Part of that build you can create artifacts which contains your app
Then create release which will deploy jars to azure app service
I know we can use the maven plugin to deploy the application to GCP. E.G
mvn appengine:deploy
But for certain reason, I would like to deploy the application not from the source code, but from artifacts repository (like Nexus).
What I want to do is to fetch the war from Nexus and use some command line to deploy the war to GCP. Anyone know how to?
I didn't find a way to do it but if you are interested in such functionality, you can use Cloud Source Repositories, which provides private Git repositories hosted on GCP.
You can use them to deploy in App Engine and even automate the process.
If you need to specifically use Nexus, you could file a feature request for App Engine.
Is it a good idea or practice to start a Spring Boot application at production with a following command of Spring Boot Maven plugin ?
mvn spring-boot:run
No, this is a bad idea.
You would re-build your project on every run which means that you would pull all needed dependencies on each new VM / container.
Also using the spring-boot-maven-plugin in conjunction with the dev-tools for example would lead to options that you don't want in production.
This ranges from using other database settings to disabled caching mechanisms for your templating engine.
Use the executable jar instead.
If you want to run the application with the Maven JVM this is fine.
It is just an alternative way to run your application instead of using the executable jar.
As an alternative you could also start your application with gradle
gradle bootRun
Which is best depends on your circumstance. For live production code I would use a versioned executable jar always.
I can't find any example or article how can the continuous delivery pipeline look like when we are using Spring Boot + Jenkins.
In Java EE I usually do it like this:
Push changes to repository
Jenkins checks for changes every 5 minutes
if there was a change, Jenkins pulls the sources and run maven build
using wildfly maven plugin I run redeploy on server
And generally I wonder what to do in last point when I'm using Spring Boot. Application is packaged into single JAR and run in separate process so in Spring Boot there is actually no such thing like redeploy. Do I have to write some script to kill old process first and then run the new artifact? Or maybe there is something like "spring boot cli" where I could manage all running spring boot apps?
You need to kill old process and run new process as a service.
It is all very well explained here Spring Boot application as a Service.
There is nice ssh plugin for jenkins that we use : https://wiki.jenkins-ci.org/display/JENKINS/Publish+Over+SSH+Plugin
Copy jar to the server
Stop old service
Run new service
EDIT : Added Spring boot reference for running spring boot as a service - http://docs.spring.io/spring-boot/docs/current/reference/html/deployment-install.html #Vaelyr
I have created two separate maven modules (let's call them MODULE1 and MODULE2) which are submodules of a third integration module (SUPERMODULE).
MODULE1 and MODULE2 are both Spring Boot Web Applications. What I'm trying to achieve is to start (not build) both projects / web apps by means of SUPERMODULE.
As I see it, there are two options:
Deploy them both to the same tomcat server (probably the better & more interesting solution)
Deploy them to different tomcat servers with different ports
I found no viable example to achieve either one of these options (... by means of a single maven integration project). Hence, I would be glad if someone could point me into the right direction - or are both possibilities bad practice?
You said :
Deploy them both to the same tomcat server
(1) Build automation software
Any Build Automation tool (Jenkins, bamboo...) would allow you to create a job that deploys both your wars to tomcat (the same server or different server, you can setup your job as you wish).
Do you use an automation software ? I believe that would be the best the solution / best practices.
(2) Build an EAR - Deploy to Tomee
You said:
What I'm trying to achieve is to start (not build) both projects / web
apps by means of SUPERMODULE.
What you are really describing is an EAR!
I'll describe the idea, however, it seems Spring boot does not play well with EAR: Spring Boot EAR Packaging and https://github.com/purple52/spring-boot-ear-skinny-war
Since your 2 submodules are spring boot app, you could:
build the submodules as WAR
have your Supermodule build an EAR
include both WARs in the EAR (maven dependencies)
this however implies that you use (for instance) Tomee instead of tomcat (is that an option for you?).
as mentioned above, after some research it seems a spring-boot war does not work when packaged inside an EAR. The SpringServletContainerInitializer isn't called. So this would not be an option at the moment.