You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I built a cluster in the OpenShift UI with a single worker node using IPMI (redfish-virtualmedia). It appears to have set the platform type to None. Reading the documentation it indicated I needed a load balancer like MetalLB to get access to services running on the data plane or worker node(s). If I deploy MetalLB I do in fact get access to services. My question is why isn't ingressVips used so this works out of the box?
Prior to deploying MetallLB I see the default ingress controller and I see it configured for a HostNetwork endpoint publishing strategy. If I curl the worker node IP the router is running (only 1 worker node in my case) for a known app like curl --insecure --verbose --header "Host: console-openshift-console.<cluster>.<base>" https://<worker-node-ip> I do in fact get the console back.
A side-effect of this is the OpenShift UI provides a link to OpenShift console that is dead on arrival. An administrator has to deploy and configure MetalLB in the cluster using CLI and then configure the *.apps.<cluster>.<base> wildcard after the fact. Leaving the feature feeling clunky out of the box. I'd have been less confused if the link wasn't shown.
The text was updated successfully, but these errors were encountered:
I built a cluster in the OpenShift UI with a single worker node using IPMI (
redfish-virtualmedia
). It appears to have set the platform type to None. Reading the documentation it indicated I needed a load balancer like MetalLB to get access to services running on the data plane or worker node(s). If I deploy MetalLB I do in fact get access to services. My question is why isn'tingressVips
used so this works out of the box?Prior to deploying MetallLB I see the default ingress controller and I see it configured for a
HostNetwork
endpoint publishing strategy. If Icurl
the worker node IP the router is running (only 1 worker node in my case) for a known app likecurl --insecure --verbose --header "Host: console-openshift-console.<cluster>.<base>" https://<worker-node-ip>
I do in fact get the console back.A side-effect of this is the OpenShift UI provides a link to OpenShift console that is dead on arrival. An administrator has to deploy and configure MetalLB in the cluster using CLI and then configure the
*.apps.<cluster>.<base>
wildcard after the fact. Leaving the feature feeling clunky out of the box. I'd have been less confused if the link wasn't shown.The text was updated successfully, but these errors were encountered: