Absolutely, you can sync two VMs to millisecond-level precision over SSH—here's a practical, step-by-step approach I’ve used repeatedly for distributed systems and latency-sensitive workloads:

Prerequisites #

• Both VMs must have passwordless SSH access set up between them (this avoids manual password prompts and streamlines automation)
• Root or sudo privileges on both machines
• A time synchronization tool like chrony (preferred for millisecond accuracy) or ntpdate installed (most Linux distros include these by default)

Step 1: Set up passwordless SSH (if not already configured) #

On the VM you want to use as the time master:

1. Generate an SSH key pair (press Enter for all prompts to use default settings):

   ssh-keygen -t ed25519
   

2. Copy the public key to the time slave VM to enable passwordless login:

   ssh-copy-id user@slave-vm-ip
   

3. Test the connection to confirm no password is required:

   ssh user@slave-vm-ip "echo Connected successfully"
   

Step 2: Configure the master VM as a local time source #

We’ll use chrony here because it’s far more reliable for millisecond-level sync than older tools like ntpd, especially over variable network links.

1. On the master VM, edit the chrony configuration file:

   sudo nano /etc/chrony/chrony.conf
   

2. Add a line to allow the slave VM to access the master’s time service:

   allow slave-vm-ip
   

3. Restart the chronyd service to apply changes:

   sudo systemctl restart chronyd
   

4. Verify the master is ready to serve time:

   chronyc tracking
   

   You should see output confirming the master’s time is stable and synchronized.

Step 3: Sync the slave VM to the master (two options) #

Option 1: Permanent, ongoing sync (recommended) #

This sets up the slave to continuously sync with the master, maintaining millisecond accuracy over time.

1. On the slave VM, edit its chrony config file:

   sudo nano /etc/chrony/chrony.conf
   

2. Comment out any existing server lines (e.g., public NTP servers) and add:

   server master-vm-ip iburst
   

   The iburst flag speeds up the initial sync process.
3. Restart chronyd on the slave to apply changes:

   sudo systemctl restart chronyd
   

4. Verify the sync status—look for a low offset (under 1ms is ideal):

   chronyc sources -v
   chronyc tracking
   

Option 2: One-time manual sync #

If you only need to align the clocks once (not ongoing sync), run this SSH-based command directly on the slave VM:

sudo date -s "$(ssh user@master-vm 'date -u +"%Y-%m-%d %H:%M:%S.%N"')"

• Using %u ensures we sync to UTC, avoiding timezone mismatches between VMs.
• The %N captures nanoseconds, which the date command will round to the nearest millisecond when setting the slave’s clock.

Step 4: Verify millisecond accuracy #

To confirm the clocks are properly aligned:

1. Run this command simultaneously on both VMs (you can script this over SSH for precise timing):

   date +%s.%N
   

2. Compare the outputs—any difference should be less than 1ms.
3. For a more formal check, use chronyc tracking on the slave: look for the System time line, which shows the exact offset from the master’s clock.

Important Notes #

• Ensure both VMs use the same timezone (UTC is strongly recommended for server environments) to avoid confusion.
• If working with virtual machines, disable host-level time sync (e.g., VMware Tools, Hyper-V time sync)—this can interfere with manual sync efforts.

内容的提问来源于stack exchange，提问作者Kostas Tsakos