Supervisor Control Plane Management floating IP address is not persisting due to extra Ipv6 DNS in network config
search cancel

Supervisor Control Plane Management floating IP address is not persisting due to extra Ipv6 DNS in network config

book

Article ID: 386881

calendar_today

Updated On:

Products

vSphere with Tanzu

Issue/Introduction

 All 3 CPVMs are showing this repeated eth0 interface setting change. (and ending up in a systemd-networkd restart loop)

/var/log/update-controller/sync.log

2024-12-24T19:53:58.56Z INFO network_setting: Network setting changed for if eth0
2024-12-24T19:53:58.56Z DEBUG network_setting: curr if network = eth0, network={'Address': 'x.x.x.x/x', 'Gateway': 'x.x.x.x', 'DNS': 'x.x.x.x x.x.x.x x:x:x:x:x:x:x:x x:x:x:x:x:x:x:x x:x:x:x:x:x:x:x', 'Domains': 'domain.com', 'NTP': 'ntp.com'}, dhcp={}, desired if network = eth0, network={'Address': 'x.x.x.x/x', 'Gateway': 'x.x.x.x', 'DNS': 'x.x.x.x x.x.x.x', 'Domains': 'domain.com', 'NTP': 'ntp.com'}, dhcp={}
2024-12-24T19:53:58.66Z DEBUG network_setting: Ensured DHCP=ipv6 not present in eth0's network file
2024-12-24T19:53:58.478Z DEBUG network_setting: Restarted resolved and networkd.

Environment

vSphere 7.x

vSphere 8.x

Cause

An extra DNS (IPv6) server ending up in the eth0 settings that is causing update-controller to repeatedly think it needs to resync the interface settings.

Resolution

- SSH into one of the CPVMs
- cd to /etc/systemd/network/
- Backup the 10-eth0.network file somewhere in /root (_not_ in the systemd directory to be safe)
- Open the 10-eth0.network file for editing
- Add the following line in the [Network] section.

IPv6AcceptRA=no

alternate way to update the network file

echo "IPv6AcceptRA = no" >> /etc/systemd/network/10-eth0.network && systemctl restart systemd-networkd

- Restart networkd using `systemctl restart systemd-networkd` (this might already be happening every 10 seconds)

- Check/tail the output of /var/log/update-controller/sync.log on the node to see if it still logging 'Network setting changed for if eth0' with recent timestamps

- Check if the update-controller sync loop is stabilizing (no longer repeating every 10s)

- Repeat on the other two and verify if the system stabilizes.