The NTP server (172.22.23.1) is the machine synchronous:
Disable the monitor
Server 127.127.1.0
Fudge 127.127.1.0 stratum 10

And then the other nodes is synchronized to the NTP server:
Disable the monitor
Server 10.96.234.6

/etc/NTP. Under the conf no other configuration, if stopped the NTP service, after ntpdate, restart the NTP service, short time monitoring NTPQ - p time lag is good, but after a long period of time will find time lag is especially huge, can reach 2000 ms, ever bosses know this better?

Every 2.0 s: NTPQ -p Mon Dec 16 14:46:46 2019

Remote refid st t the when poll reach delay offset jitter
==============================================================================
59 * tdh008 LOCAL 11 u (0) 64 377 0.303 561.587 930.706