Cannot create a development cluster currently. I get the following error

$ gokp create-cluster development --cluster-name=chx-$RANDOM$$ --github-token=${GH_TOKEN}  --gitops-controller=fluxcd
INFO[0000] Running checks                               
INFO[0000] Creating temporary control plane             
INFO[0020] Initializing Docker provider                 
FATA[0070] action failed after 10 attempts: failed to get cert-manager object /, Kind=, /: Object 'Kind' is missing in 'unstructured object has no kind

Currently, gokp installs an opinionated Argo CD instance. This should be extended to also provide an opinionated Flux CD instance as well. This will give the user the ability to use either ArgoCD or FluxCD.

• --gitops-controller=argocd
• --gitops-controller=fluxcd

The directory structure may or maynot stay the same. But in either case it should be opinionated for the specific controller.

There should be a flag called --skip-cloudformation. This is for users who have already ran the cloud formation prerequisite This flag should skip this step when called.

Currently create-cluster creates a cluster on AWS and development-cluster creates one on docker. create-cluster should take subcommands, and those sub commands are the hyperscalers/infrastructure provider.

Example:

• gokp create-cluster aws [options]
• gokp create-cluster development [options]
• gokp create-cluster azure [options]
• gokp create-cluster gcp [options]

After this, there should only be create-cluster and delete-cluster and development-cluster and development-delete would no longer be needed.

This update

• Development cluster now only deploys 1 master and 2 nodes. Has an optional "emulate HA" mode.
• Adds --skip-cloud-formation option for those who ran the CloudFormation already (fixes #8)
• Label Workers of the clusters as such (was previously unlabeled)
• Making create-cluster and delete-cluster provider agnostic. This also removed the need to have a separate development-create and development-delete subcomands. (fixes #3)

Using the development mode of gokp on a Mac returns the following.

INFO[0230] Control Plane Nodes are Online, saving Kubeconfig
E1014 10:30:38.616046   63630 memcache.go:179] couldn't get current server API group list: Get "https://172.18.0.3:6443/api?timeout=32s": dial tcp 172.18.0.3:6443: i/o timeout
FATA[0261] Get "https://172.18.0.3:6443/api?timeout=32s": dial tcp 172.18.0.3:6443: i/o timeout

Known issue with CAPI Docker provider (CAPD) where the kubeconfig file needs tweaking.

Potential solution: Edit the file before using it (without calling sed) in the CreateDevelK8sInstance function ~> https://github.com/christianh814/gokp/blob/main/cmd/capi/capi.go#L428-L431

After deleting a cluster on Docker, the delete command doesn't delete the bootstrapper container. This causes the next create-cluster to fail:

nillsf@SHRISU-YOGA ~ % gokp delete-cluster development \ 
--cluster-name=$MYCLUSTER --kubeconfig=/home/nillsf/.gokp/$MYCLUSTER/$MYCLUSTER.kubeconfig

INFO[0000] Deleting development cluster nfcluster       
INFO[0022] Cluster nfcluster successfully deleted       

nillsf@SHRISU-YOGA ~ % gokp create-cluster development \
--cluster-name=$MYCLUSTER --github-token=$GH_TOKEN

INFO[0000] Running checks                               
INFO[0000] Creating temporary control plane             
FATA[0001] node(s) already exist for a cluster with the name "gokp-bootstrapper" 

Currently only GitHub is supported. It would be nice to also support Bitbucket, GitLab, and GitTea

• --bitbucket-token=
• --gitlab-token=
• --gittea-token=

If the user already has a management cluster, then it should be used (instead of creating one). The default should be what it is now (create one) but a user should be given the option to use one that's already there.

• --management-cluster=/path/to/kube.config

It's not enough to just wait for the nodes to be ready. There is a taint places when they are not ready and removed when they are. gokp should ALSO wait until the taint is removed.

There was an attempt, but I never got back to it.