Skip to main content
Talos Linux uses time.cloudflare.com as the default NTP server for time synchronization. You can override this default in two ways:
  • Using kernel parameters
  • Using cluster configuration patches

​
Kernel parameters

Use kernel parameters when you need to override NTP before the machine configuration is loaded, for example, during a bare-metal ISO, PXE boot, or when a node cannot reach the default NTP server early in the boot process. Kernel parameters can be used to configure initial network settings such as interfaces, routes, DNS, and NTP servers. For more details, see the kernel documentation. The kernel parameter uses the following syntax: 
ip=<client-ip>:<server-ip>:<gw-ip>:<netmask>:<hostname>:<device>:<autoconf>:<dns0-ip>:<dns1-ip>:<ntp0-ip>
Each field corresponds to a specific network or time-synchronization setting and can be left empty if it is not being overridden. For example, to override the default NTP server (time.cloudflare.com) with 172.235.60.8, you can specify only the NTP field and leave the rest empty: 
ip=:::::::::172.235.60.8
You can then supply this ip value as a kernel parameter during boot to override the NTP server early in the startup process. To do this first download the installation media baked with the ip value:
Download the installation media with the ip kernel parameter using the Omni UI:
  1. In the Omni UI, select Download Installation Media.
  2. Choose the appropriate image for your machine.
  3. Enter the ip value in the Additional Kernel Arguments field. Additional Kernel Arguments
  4. Click Download.
Boot your machines using the installation media you just downloaded, then create your cluster with those machines. Finally verify that the NTP server has been overridden by replacing <NODE_IP> with your node’s IP address and running: 
talosctl get timeservers -n <NODE_IP>

​
Config patches

Use config patches to override NTP servers after boot on a running cluster. For example, to replace the default NTP server with 172.235.60.8, apply the following patch to your cluster using the cluster machine patch workflow. 
machine:
  time:
    servers:
      - 172.235.60.8
After the patch is applied, verify that the NTP server has been successfully overridden by replacing <NODE_IP> with the IP address of one of your nodes: 
talosctl get timeservers -n <NODE_IP>