-
Feature Request
-
Resolution: Unresolved
-
Major
-
None
-
None
-
None
-
Product / Portfolio Work
-
None
-
False
-
-
None
-
None
-
None
-
-
None
-
None
-
None
-
None
-
None
1. Proposed title of this feature request
Provision to use *podSelector* in Networkpolicy for specific ingresscontrollers.
2. What is the nature and description of the request?
Actual results: As per official documentation, To allow traffic from ns/openshift-ingress, a networkpolicy MUST have a *namespaceSelector* Expected results:To allow traffic from any specific ingresscontroller in ns/openshift-ingress, a networkpolicy should support a *podSelector* value.
3. Why does the customer need this? (List the business requirements here)
One of the customer has a requirement to all traffic from a SPECIFIC ingresscontroller to a workload.
4. List any affected packages or components.
Networkpolicy
As per the official documentation, to allow traffic from ns/openshift-ingress, a networkpolicy MUST have a namespaceSelector to point to the labels for ns/openshift-ingress however it does not work when there is a podSelector that points to any specific ingresscontroller among multiple ingresscontrollers.
How reproducible: Always
Steps to Reproduce:
1. Host multiple ingresscontrollers (minimum 3 including default)
1. ingresscontroller/default (HostNetwork)
2. ingresscontroller/custom1 (HostNetwork)
3. ingresscontroller/custom2 (pod network)
2. Create a networkpolicy to allow traffic ONLY from ingresscontroller/custom2 or ONLY from ingresscontroller/custom1
3. Since the `allow-from-router` or the `allow-from-openshift-ingress` networkpolicy uses the namespaceSelector and directly targets the entire namespace, it makes it difficult to isolate the traffic to any specific workload from any specific ingresscontroller and not all.