-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
4.17.0
-
None
Description of problem:
Adding a node with `oc adm node-image` fails: oc adm node-image monitor --ip-addresses 192.168.250.77 time=2024-10-10T11:31:19Z level=info msg=Monitoring IPs: [192.168.250.77] time=2024-10-10T11:31:19Z level=info msg=Cannot resolve IP address 192.168.250.77 to a hostname. Skipping checks for pending CSRs. time=2024-10-10T11:31:19Z level=info msg=Node 192.168.250.77: Assisted Service API is available time=2024-10-10T11:31:19Z level=info msg=Node 192.168.250.77: Cluster is adding hosts time=2024-10-10T11:31:19Z level=warning msg=Node 192.168.250.77: Platform OpenStack Compute is allowed only for Single Node OpenShift or user-managed networking
Version-Release number of selected component (if applicable):
4.17.0
How reproducible: Always
Extra information:
The cluster is deployed using Platform:None and userManagedNetworking on an OpenStack cluster which is used as a test bed for the real hardware Agent Based Installer.
Bootstrap of the cluster itself is successfull, but adding nodes as day 2 is not working.
During the cluster bootstrap, we see the following log message:
{\"id\":\"valid-platform-network-settings\",\"status\":\"success\",\"message\":\"Platform OpenStack Compute is allowed\"}So after looking at https://github.com/openshift/assisted-service/blob/master/internal/host/validator.go#L569
we suppose that the error is related to `userManagedNetworking`
being set to true when bootstraping and false when adding a node.
A second related issue, is why the platform is seen as openstack, as neither the cluster-config-v1 configmap containing install-config or the infrastructure/cluster object mentions OpenStack.
Not sure if this is relevant but an external CNI plugin is used here, we have networkType: Calico in the install config.
- links to