Skip to content
This repository has been archived by the owner on Jan 28, 2020. It is now read-only.

It isn't possible to customize OpenShift installer variables #340

Open
jhernand opened this issue Sep 13, 2018 · 7 comments
Open

It isn't possible to customize OpenShift installer variables #340

jhernand opened this issue Sep 13, 2018 · 7 comments

Comments

@jhernand
Copy link
Contributor

In the application that we are building we want to give users the possibility to build different kinds of clusters, which are basically defined by sets of OpenShift installer variables. Lets say, for example, that we want to have a kind of cluster that has ansible_service_broker_install set to false. Is there any way to do that without modifying the code of the operator? As far as I can see there is a template hard-coded in the source, but no mechanism to replace it.

@jhernand
Copy link
Contributor Author

CC: @kbsingh @zgalor @nimrodshn

@dgoodwin
Copy link
Contributor

Is it something you want customers to have choice over? Or do you expect to set this for all clusters?

@jhernand
Copy link
Contributor Author

We expect the customer to select a kind or flavour of cluster, that we associate with a collection of settings. For example, lets say that we define a regular kind and a wihout_service_broker kind. The customer will select one of those, and then we need to apply the corresponding settings.

@dgoodwin
Copy link
Contributor

So for 4.0 this would probably boil down to sets of optional operators to get installed. We'll want to stay in touch on how you want this to work so we can ensure Hive will meet the need.

For 3.x and current CO you could perhaps shoehorn a list of flavors into an annotation on the cluster deployment, modify CO to interpret those and inject sets of ansible variables based on flavor.

@aweiteka
Copy link

My non-exhaustive list includes:

  • override default VPC CIDR block, e.g. 172.31.48.0/20
  • authentication config (github, google, oidc), includes secrets
  • EC2 AMI (e.g. ami-12345), instance type (e.g. m5.xlarge), region (e.g. us-east-2)
  • root user public key(s) for ssh access

cc @dak1n1 @blrm

@dgoodwin
Copy link
Contributor

VPC CIDR block is not exposed in the API yet but it absolutely should be. This is one you should add to AWSClusterSpec and plumb through to the generate.go that makes the Ansible inventory.

Auth config is something that also should be in the ClusterDeployment. We planned to just directly re-use the Origin go auth struct.

AMI, instance type, and region should be controllable from the API today.

Root SSH keys should also be hooked up and working, you store them in a secret and associate it to the cluster deployment, they will be added to the AWS instance as a result.

@jhernand
Copy link
Contributor Author

The support for customizing the AMI, the region and the SSH keys we are already using via the API, and works well.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants