Nodes are not powered down on make clean and cause side effects

[yrobla]

When i execute a make clean command, I expected that my nodes (masters/workers) are powered down. But that is not the case, while the bootstrap VM is powered down, losing all the services that were used by masters, such as coredns.
As a side effect, masters are still querying for addresses like api-int (resolved in the past by bootstrap vm), and this is causing a saturation (max queries reached) on the dns service of the installer host.

[dhellmann]

Is this problem still occurring?

[yrobla]

I think so yes, this was a problem when masters were not totally bootstrapped and vm was still there. But the issue as it was, is that i was expecting nodes to be powered down on make clean, and this is not happening.

[hardys]

This is expected as we've not got any destroy support for IPI baremetal in the installer yet ref openshift/installer#2005

We could hack around it in a dev-scripts specific way (ipmitool etc), but I'm not sure it's worth it since we expect the end-user interface to be the installer.

[dhellmann]

When I run make clean my VMs are removed completely. Are you using VMs or real hardware?

[stbenjam]

We're tracking this as a wish list item to implement proper cluster destroy in the installer

/close

[openshift-ci-robot]

@stbenjam: Closing this issue.

In response to this:

We're tracking this as a wish list item to implement proper cluster destroy in the installer

/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.