Description of problem: We are using a hardcoded namespace `kube-system` which will ignore the setting in defaultNetworks for multus. This is not a problem in kubernetes cluster or OCP 3.x cluster, since all the critical service are deployed under kube-system. But for OCP 4.0, there is a lot of system services which are running in different namespaces. Eg, the coredns in running in openshift-dns namespace We should change the hardcoded kube-system to a configurable list. Version-Release number of selected component (if applicable): 4.0 How reproducible: always Steps to Reproduce: 1. 2. 3. Actual results: The pod will ignore the defaultNetworks crd in kube-system only. Expected results: As we discussed in PR https://github.com/intel/multus-cni/pull/203 The coredns service is not running in kube-system now. We should update the kube-system namespace or even better have a configurable list. Additional info:
https://github.com/intel/multus-cni/issues/253
Is this downstreamed? can this be marked as MODIFIED?
Casay, Thank you for your reminder! Verified that the fix is in downstream, https://github.com/openshift/multus-cni/commit/73e10fb99993963aa73125aeffffa685776e3090 so changed to MODIFIED.
Tested on 4.0.0-0.nightly-2019-03-15-063749 The new added parameters systemNamespaces and multusNamespace are both working fine. Move the bug to verified.
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-2019:0758