Sebastian Łaskawiec commented on ISPN-8233:
Quote from Infinispan Internal mailing list:
I've been talking with Marek Goldmann about the repository and CI for JDG Online
Services. It turned out that the Cloud Enablement CI is tied to jboss-container-images 
organization. It scans repositories and branches and creates jobs based on that (actually,
it's pretty clever I must admit).
Unfortunately the CE Team is migrating images to Dogen 3. Having this in mind, Marek
suggested that we should probably start building new images based on it (here's an
example ). The biggest concern at this point is the name of the image. Our current JDG
7 image is called "jboss-datagrid-7/datagrid71-openshift" . I'm not sure
if this is the naming scheme we'd like to continue. I was thinking about something
like this: "jboss-datagrid-services/caching-service". Once we have a name, I can
start looking into repository structure.
Another obstacle is the way the CI works in the CE Team. They execute Behave tests and
validate container itself. After all images are built, they trigger Arquillian Cube tests,
which are responsible for checking templates etc. Again, I don't think this is the
right approach. So IMO we should create a small script which would start OpenShift, push
the image into the local registry and run Arquillian Cube tests against it. The advantage
of this approach is that we could run those tests on all machines using well known JUnit
framework. The drawback is that this approach is not standard for CE - this is something
that only our image will have.
Given all that, my plan is the following:
I'll start implementing JDG Caching Service in my own repository. This will buy us
some time for discussion and I won't be blocked with my work.
Once I have something that works, I will adjust it to CE requirements and migrate it to
I will create scripts (probably based on Makefile or Bash) which should be invoked by CI.
Later on we can use some JUnit reporting plugin but checking script's exit code will
give us enough information - green or not.
The new Caching Service will be based on our OpenShift image . This way I won't
duplicate Jolokia integration, KUBE_PING module installation etc. The main difference is
that we will bootstrap the image using new bootstrap script.
For configuration management I plan to use CLI in offline mode. Based on provided
configuration, the CLI will adjust cloud.xml file (remove unnecessary caches, endpoints
and so on).
Please let me know if you have any questions or comments.
Setup Jenkins for building Online images
Issue Type: Task
Components: Cloud Integrations
Reporter: Sebastian Łaskawiec
Assignee: Sebastian Łaskawiec
This message was sent by Atlassian JIRA