-
Bug
-
Resolution: Done-Errata
-
Normal
-
4.12
-
None
Description of problem:
OVN br-int ovs flows do not get updated on other nodes when a nodes bond MACADDR is changed to other slave interface after reboot. This causes network traffic coming for the sdn of one node to get dropped when it hits the node that changed mac addresses on its bond interface.
Version-Release number of selected component (if applicable): 4.12+
How reproducible: 100% of the time after rebooting if the mac changes. mac does not always change.
Steps to Reproduce:
1. Capture bond0 mac before reboot 2. Reboot host 3. Confirm mac change 4. oc run --rm -it test-pod-sdn --image=registry.redhat.io/openshift4/network-tools-rhel8 --overrides='{"spec": {"tolerations": [{"operator": "Exists"}],"nodeSelector":{"kubernetes.io/hostname":"nodeb-not-rebooted"}}}' /bin/bash 5. Ping rebooted node
Actual results:
ping hits rebooted node but is dropped because the MAC address is of other slave interface and not the one bond is using.
Expected results:
OVS flows to update on all nodesafter reboot if MAC changes
Additional info:
If we restart NetworkManager a couple times this triggers OVS flows to get updated, not sure why. Possible workarounds - https://access.redhat.com/solutions/6972925 - Statically set the mac of bond0 to one of the slave interfaces.
- is duplicated by
-
OCPBUGS-20353 NetworkManager fails on RHEL 8.6 worker configuring br-ex interface post reboot
- Closed
- links to
-
RHEA-2024:0041 OpenShift Container Platform 4.16.z bug fix update