OCM-11842 : Improve gcp cluster creation with psc -UX

[ckandag]

https://issues.redhat.com/browse/OCM-11842

1. Make the interactive workflow for ClusterPrivacy/ BYO- VPC similar to UI
2. For WIF clusters, Ensure PSC is set when Cluster is Private

[JakobGray]

I did a dry-run flow with this branch and got the following

go run ./cmd/ocm create cluster -i --dry-run=true 
? cluster name jjjj
? Subscription type: standard (Annual: Fixed capacity subscription from Red Hat)
? Cloud provider: gcp
? CCS: Yes
? Authentication method: Workload Identity Federation (WIF)
? WIF Configuration: jg1021 (2ehvi64du4ahq77a63vf3gj70bm63bpa)
? Multiple AZ: No
? Secure boot support for Shielded VMs: No
? Region: us-east1
? OpenShift version: 4.17.0
? Compute machine type: n2-standard-4
? Enable autoscaling: No
? Compute nodes: 2
? Private cluster (optional): Yes
? Install into an existing VPC (optional): No
? Machine CIDR: 10.0.0.0/16
? Service CIDR: 172.30.0.0/16
? Pod CIDR: 10.128.0.0/14
? Host prefix: 23
? Domain Prefix: 
dry run: Would be successful.

Based on point 2 I should not be able to select No for the prompt Install into an existing VPC (optional)

Also there are typos in the PSC prompt

? PrivatSericeConnect ServiceAttachment Subnet: [? for help] 

[renan-campos]

These two const names are mixed up for their values and not used in defining the flag.

[renan-campos]

If we are adding consts for these, we should replace the hard-coded values.

[ckandag]

I just replaced in the code I was working on but yes I replace all

[JakobGray]

My comment is still mostly the same

ocm create cluster -i --dry-run=true --ccs  --provider=gcp --region=us-east1 --compute-machine-type=n2-standard-4 --psc-subnet=jdg-21-tf2d8-worker-subnet jjjj
? Subscription type: standard (Annual: Fixed capacity subscription from Red Hat)
? Authentication method: Service account
? Service account file: /Users/jdgray/gitlab/temps/sda-ccs-1/jdg-ccs-1-key.json
? Multiple AZ: No
? Secure boot support for Shielded VMs: No
? OpenShift version: 4.17.0
? Enable autoscaling: No
? Compute nodes: 2
? Private cluster (optional): Yes
? Install into an existing VPC (optional): [? for help] (y/N) 

In the interactive flow where I've already indicated that I'm using PSC it asks me if I want a private cluster and if I want to install into an existing VPC. Both of these prompts have a "right" and "wrong" answer in this case, and the default option is the incorrect one.

If managing the flow is too complex then it might not be worth pursuing, because at least if the user chooses wrong they receive quick feedback on their mistake.

[ckandag]

My comment is still mostly the same

ocm create cluster -i --dry-run=true --ccs  --provider=gcp --region=us-east1 --compute-machine-type=n2-standard-4 --psc-subnet=jdg-21-tf2d8-worker-subnet jjjj
? Subscription type: standard (Annual: Fixed capacity subscription from Red Hat)
? Authentication method: Service account
? Service account file: /Users/jdgray/gitlab/temps/sda-ccs-1/jdg-ccs-1-key.json
? Multiple AZ: No
? Secure boot support for Shielded VMs: No
? OpenShift version: 4.17.0
? Enable autoscaling: No
? Compute nodes: 2
? Private cluster (optional): Yes
? Install into an existing VPC (optional): [? for help] (y/N) 

In the interactive flow where I've already indicated that I'm using PSC it asks me if I want a private cluster and if I want to install into an existing VPC. Both of these prompts have a "right" and "wrong" answer in this case, and the default option is the incorrect one.

If managing the flow is too complex then it might not be worth pursuing, because at least if the user chooses wrong they receive quick feedback on their mistake.

The interactive flow I think has value and does work when user tries it all interactively.
An edge case with a mix of both non interactive and interactive flags which is not working.
which ideally that too should but given the main flows work in both interactive anad non-interactive mode.
I will look into fixing the above mixed case if its not too complicated.

[renan-campos]

/lgtm