In OpenShift 4, NTP configuration is managed through chronyd via MachineConfig. However, any changes to the NTP configuration (e.g., updating /etc/chrony.conf) currently require a node reboot, even when the changes are minimal and could be applied with a simple service reload or restart.

This request is to support non-disruptive (hot) updates to NTP configuration—on both control plane and worker nodes—without requiring full system reboots, especially in deployments using static IP configurations and non-DHCP NTP provisioning, such as Agent-based or BareMetal IPI installations.

Use Case / Business Justification:
Customers with static-network-based installations frequently change NTP servers hosted on:

• Local network switches

• Domain controllers (DCs)

• Edge time sources

The current behavior (mandatory node reboot on every NTP update) creates unacceptable disruption:

• Causes downtime and unnecessary node evictions

• Impacts HA control plane availability

• Diverges from modern OS standards (e.g., RHEL, Fedora) where time sync config can be updated via systemctl restart chronyd without reboot

The ability to hot-apply changes to NTP configuration is standard in Linux and should be expected in OCP as well.

Expected Outcome / Enhancement Request:

• Support dynamic (rebootless) updates to NTP configuration applied via MachineConfig or another supported method.

• Provide a mechanism that triggers chronyd reload or restart when a MachineConfig updates the relevant files.

• Enable this capability for both control plane and worker nodes.

Affected Components:

• machine-config-operator

• chronyd.service

• systemd

• Possibly nmstate or NetworkManager

Customer Impact:

• Enables non-disruptive operations for time source management

• Reduces operational toil in static-network installs

• Aligns OpenShift with standard RHEL/Linux system management practices

• Avoids downtime in production or edge deployments with tight maintenance windows