Unclear how to restart a fresh "bin/openshift-install install cluster"

[rwsu]

Version

$ openshift-install version
bin/openshift-install v0.5.0-master-47-gf7d6d2923a7979344fcac33293084051cecc8aab-dirty
Terraform v0.11.10
$ ~/.terraform.d/plugins/terraform-provider-libvirt -version  # only needed if you're using libvirt
Compiled against library: libvirt 4.1.0
Using library: libvirt 4.1.0
Running hypervisor: QEMU 2.11.2
Running against daemon: 4.1.0

Platform (aws|libvirt|openstack):

libvirt

What happened?

My first install failed with:

WARNING Failed to connect events watcher: Get https://demo-api.localdomain:6443/api/v1/namespaces/kube-system/events?watch=true: dial tcp: lookup demo-api.localdomain on 127.0.0.1:53: no such host 
WARNING Failed to connect events watcher: Get https://demo-api.localdomain:6443/api/v1/namespaces/kube-system/events?watch=true: dial tcp: lookup demo-api.localdomain on 127.0.0.1:53: no such host 
WARNING Failed to connect events watcher: Get https://demo-api.localdomain:6443/api/v1/namespaces/kube-system/events?watch=true: dial tcp: lookup demo-api.localdomain on 127.0.0.1:53: no such host 
WARNING Failed to connect events watcher: Get https://demo-api.localdomain:6443/api/v1/namespaces/kube-system/events?watch=true: dial tcp: lookup demo-api.localdomain on 127.0.0.1:53: no such host 
FATAL Error executing openshift-install: waiting for bootstrap-complete: timed out waiting for the condition 

Then I destroyed the cluster:

[root@puppy installer]# bin/openshift-install destroy cluster
INFO Deleted domain                                domain=demo-master-0
INFO Deleted domain                                domain=demo-bootstrap
INFO Deleted network                               network=demo
INFO Deleted volume                                volume=demo-bootstrap
INFO Deleted volume                                volume=demo-master.ign
INFO Deleted volume                                volume=demo-bootstrap.ign
INFO Deleted volume                                volume=demo-base
INFO Deleted volume                                volume=demo-master-0

When try to rerun the installer to create the cluster, it doesn't prompt me and reuses what I had entered before. It is not clear how to reset the installer so that it prompts me for domain, install type, etc..

What you expected to happen?

There should be an option or instructions on how to reset the installer or its state so it starts fresh.

How to reproduce it (as minimally and precisely as possible)?

bin/openshift-install create cluster
bin/openshift-install destroy cluster
bin/openshift-install create cluster

[ravisantoshgudimetla]

I am noticing this too on AWS. FWIW, I have created cluster with other another name and that failed too. I am trying other combinations, like removing .openshift_install_state.json. Will post the status here if anything works fine.

[ravisantoshgudimetla]

Alright deleting .openshift_install_state.json helped in recreating the cluster

/cc @wking

[akrzos]

I also had to remove .openshift_install_state.json in order to get subsequent installs on AWS to occur.

[wking]

There are some cleanup suggestions here. Sometimes recycling state from a previous cluster is a good thing, and sometimes it is confusing. Some more on this here, in #635, etc., so obviously there's still room for improving the UX.

[wgordon17]

I know there's a --dir option which installs everything into a self-contained directory. What if the "default" is to use a directory named after the cluster? You can still override with --dir, but I think this would at least make it easier to separate installs. And then commands like openshift-install destroy cluster could look in the current directory, or fail with an error message, something like cluster not found in current directory, use --dir to specify the cluster directory

[wking]

What if the "default" is to use a directory named after the cluster? You can still override with --dir, but I think this would at least make it easier to separate installs.

This has been suggested before, and I still feel like this. In fact, lets close this issue as a dup of #522. Please post a comment here if you feel like there's enough space between to two to be worth separate issues.

/close

[openshift-ci-robot]

@wking: Closing this issue.

Details

In response to this:

What if the "default" is to use a directory named after the cluster? You can still override with --dir, but I think this would at least make it easier to separate installs.

This has been suggested before, and I still feel like this. In fact, lets close this issue as a dup of #522. Please post a comment here if you feel like there's enough space between to two to be worth separate issues.

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.