Description of problem: I registered another worker using the GUI, and it never got passed the "powering up" stage. From the CLI I could see that it was registered in the "default" namespace: $ oc get baremetalhosts --all-namespaces NAMESPACE NAME default worker-1 This problem, in turn, also makes it impossible to delete the host from the GUI (the "delete" command is not available) Version-Release number of selected component (if applicable): 4.4.0-0.ci-2019-11-28-063411 How reproducible: 100% Steps to Reproduce: 1. Register a node from the GUI
The host is registered in currently active project as expected. It seems that a host has to be created in openshift-machine-api to be acted upon by MAO or BMO (?). When user goes to Machines page, the UI switches the namespace to openshift-machine-api. We can do the same with baremetal hosts page, but it doesn't prevent user from subseguently change the namespace (project) to something different. Same applies to CLI. The exact same issue can be observed on Machines, so I guess that's expected but not very user friendly.
Allowing the user to register the hosts in the wrong namespace makes no sense. Can we just always run the registration in the right namespace? The same goes for reading the hosts - regardless of which namespace is currently selected in the UI, we should get the data from openshift-machine-api. This should be fixed in 4.3 as well.
Checked on 4.4 cluster with payload: 4.4.0-0.nightly-2020-05-18-164758 Goto "Bare Metal Hosts" page, it will switch to namespace "openshift-machine-api" by default, and adding host from console will also create host in namespace "openshift-machine-api" by default now. The bug is fixed.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:2180