rht-labs / enablement-ci-cd Goto Github PK

For advanced users it might be obvious but for fresh users it might be worth including in the documentation that inventory/group_vars/all.yml -> namespace_prefix must be configured before playbooks are executed.

We have seen during different DO500 sessions that https://updates.jenkins.io/download/plugins is not very responsive. There were some cases where students could not go ahead on the exercises because of this (not able to build their own Jenkins image).

Could we consider building our own custom Jenkins container image and store it in https://quay.io/? We could create an organization for Labs or DO500 if this doesn't exist?

We should evaluate on having dedicated Nodes just for containers builds. Builds are intensive during the enablement sessions so it would probably make sense to get some dedicated Nodes just for this purpose.

We have katacoda scenarios for: static params files, dynamic params files, and secrets! we should utilize them in a manner similar to what @oybed and I have used for pushing code across into another cluster!

https://docs.ansible.com/ansible/2.5/user_guide/playbooks_reuse_roles.html

We are using persistent storage for every single component of the enablement (Nexus, Jenkins, MongoDB...). Apart from the shared GitLab no persistent storage should be required

The work for OCP v4.x is in progress, and currently we are keeping the work in -v4.1 branches to avoid breaking existing material. As soon as we think the new content is ready and we are wanting to move to use OCP v4.x, we need to merge these -v4.1 branches into the main ones.

In addition, we need to resolve the discussion in issue rht-labs/enablement-docs#346 to figure out what direction we want to go with branches (including tags/releases, etc. etc.). This may need to happen before we merge the v4.1 branches into the main one(s).

This would remove the need for the -i inventory on every ansible step.

As MongoDB template comes with default limit for memory and requests is missing from the template, the Pod will try to allocate by default 512Mi for all Mongo deployments. For the purpose of the training, this is too much, we need to add a small value 64Mi for the requests

Nexus resource limits are not properly configured. We need to configure an initial requests of 256Mi and limit up to 1Gi of memory

Jenkins resource limits are not properly configured. We need to configure an initial requests of 512Mi and limit up to 3Gi of memory