Hi, I would like to know more on how people are approaching building and testing of custom modules / installations. In our current setup we have a repo where we develop all our custom code. We use gradle and the 'com.github.zhurlik.jbossmodules' plugin to build wildfly modules from that code. Then we create a new custom docker image from the keycloak base image and those built modules. After we've built our custom image, a separate repo with integration tests / security tests / etc. is built, targeting the newly created image. If everything checks out, the image is deployed in our kubernetes cluster. Every step of the process is automated and works kind of ok. What I really don't like is the separation of our "module/SPI repo" and our test suite. Ideally, I would like to write all my integration tests in the same repo as the code that I'm testing and be able to fire them against a running keycloak server (with my code deployed) from within my IDE. Does this make sense? Has anyone done something like this? Is there an alternative way to build our custom images that is better suited? Looking forward to a discussion on this. Regards, Martin _______________________________________________ keycloak-user mailing list keycloak-user(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/keycloak-user

Is this documented somewhere (I can't seem to find it)? How would that work for completely custom SPIs with their own configuration? We make heavy use of standalone-ha.xml to configure providers with environment variables, which in turn are injected into the containers by Kubernetes. On Tue, 4 Apr 2017 at 15:54 Stian Thorgersen <sthorger(a)redhat.com&gt; wrote: > Use the new JEE deployer approach. Then you can deploy and re-reploy > providers live to the server with simply running "mvn wildfly:deploy". > > On 4 April 2017 at 14:57, Martin Hardselius <martin.hardselius(a)gmail.com&gt; > wrote: > > Hi, > > I would like to know more on how people are approaching building and > testing of custom modules / installations. > > In our current setup we have a repo where we develop all our custom code. > We use gradle and the 'com.github.zhurlik.jbossmodules' plugin to build > wildfly modules from that code. Then we create a new custom docker image > from the keycloak base image and those built modules. After we've built > our > custom image, a separate repo with integration tests / security tests / > etc. is built, targeting the newly created image. If everything checks > out, > the image is deployed in our kubernetes cluster. Every step of the process > is automated and works kind of ok. > > What I really don't like is the separation of our "module/SPI repo" and > our > test suite. Ideally, I would like to write all my integration tests in the > same repo as the code that I'm testing and be able to fire them against a > running keycloak server (with my code deployed) from within my IDE. Does > this make sense? Has anyone done something like this? Is there an > alternative way to build our custom images that is better suited? > > Looking forward to a discussion on this. > > Regards, > Martin > > _______________________________________________ > keycloak-user mailing list > keycloak-user(a)lists.jboss.org > https://lists.jboss.org/mailman/listinfo/keycloak-user > > >

Ok, thanks. Since we do a bit of both, I guess it's easier to continue deploying everything as modules to keep the build pipeline consistent. I guess I was looking for something in the lines of what Gabriel asked for in the "Integration Tests" thread. Maybe https://github.com/palantir/ docker-compose-rule is something worth exploring in our case. On Wed, 5 Apr 2017 at 08:42 Stian Thorgersen <sthorger(a)redhat.com&gt; wrote: The user storage example leverages this approach. It should be documented in the server developer guide. You can't deploy custom SPIs this way only providers. For a custom SPI you'd have to deploy that as a module, restart the server, then you can deploy your custom providers for your custom SPIs. On 4 April 2017 at 16:32, Martin Hardselius <martin.hardselius(a)gmail.com&gt; wrote: Is this documented somewhere (I can't seem to find it)? How would that work for completely custom SPIs with their own configuration? We make heavy use of standalone-ha.xml to configure providers with environment variables, which in turn are injected into the containers by Kubernetes. On Tue, 4 Apr 2017 at 15:54 Stian Thorgersen <sthorger(a)redhat.com&gt; wrote: Use the new JEE deployer approach. Then you can deploy and re-reploy providers live to the server with simply running "mvn wildfly:deploy". On 4 April 2017 at 14:57, Martin Hardselius <martin.hardselius(a)gmail.com&gt; wrote: Hi, I would like to know more on how people are approaching building and testing of custom modules / installations. In our current setup we have a repo where we develop all our custom code. We use gradle and the 'com.github.zhurlik.jbossmodules' plugin to build wildfly modules from that code. Then we create a new custom docker image from the keycloak base image and those built modules. After we've built our custom image, a separate repo with integration tests / security tests / etc. is built, targeting the newly created image. If everything checks out, the image is deployed in our kubernetes cluster. Every step of the process is automated and works kind of ok. What I really don't like is the separation of our "module/SPI repo" and our test suite. Ideally, I would like to write all my integration tests in the same repo as the code that I'm testing and be able to fire them against a running keycloak server (with my code deployed) from within my IDE. Does this make sense? Has anyone done something like this? Is there an alternative way to build our custom images that is better suited? Looking forward to a discussion on this. Regards, Martin _______________________________________________ keycloak-user mailing list keycloak-user(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/keycloak-user