With vSphere 8U3, when you update NTP settings of the Supervisor in the vCenter UI it triggers a rolling update on Guest Clusters.
In contrast, the change takes effect immediately on the Supervisor Cluster, without any rolling update.
In large-scale environments, administrators may prefer to update the NTP server without initiating rolling updates across Guest Clusters. This KB outlines a method to temporarily change the NTP server without triggering rolling updates on the Guest Clusters.
Note: This is a temporary workaround. A full rolling update should be carried out at a later, appropriate time to ensure persistent change.
VMware vSphere Kubernetes Service - vCenter 8U3
1. SSH login to the supervisor
2. Edit /etc/systemd/timesyncd.conf
vi /etc/systemd/timesyncd.conf
[Time]
NTP=<NTP_SERVER_IPADDRESS_1> <NTP_SERVER_IPADDRESS_2>
3. Restart systemd-timesyncd
systemctl restart systemd-timesyncd
systemctl status systemd-timesyncd
4. Check the result
timedatectl status
timedatectl timesync-status
timedatectl show-timesync --all
5. Repeat this operation for the remaining 2 Supervisors.
1. SSH login to the Guest Cluster node
2. Check the current NTP setting
sudo -i
chronyc sources -v
3. Update NTP server configuration
vi /etc/chrony.conf
server <NTP_SERVER_IPADDRESS_1> iburst
server <NTP_SERVER_IPADDRESS_2> iburst
4. Restart chronyd
systemctl restart chronyd
systemctl status chronyd
5. Check the result
chronyc sources -v
6. Repeat this operation for all remaining Kubernetes nodes.