This issue does not happen if you configure the network interface with IP address, netmask and gateway. My network interfaces are configured without gateway and I'm using rule-ethx and route-ethx for each interface. Also NTP server is reachable and the firewall is off.
In the rule-eth1 for example I have
Code: Select all
from 192.168.100.10/32 table eth1table
and in route-eth1 I have
Code: Select all
192.168.100.0/24 dev eth1 table eth1table default via 192.168.100.1 dev eth1 table eth1table
I need to start the ntp service with a source interface so I am using the ntpd -I eth1 command. This makes my ntp server stuck in .INIT. state when using ntpq -pn command. Then I found out that if I add default route via the 127.0.0.1 to the main routing table the server is not anymore in .INIT. state but it is still unsynchronized. Then if I restart the service using the service ntpd restart command synchronizes successfully, but no If I use ntpd -I eth1 for the first time. After the time is synchronize I can use the ntpd -I eth1 command for restarting the service.
I have the exact steps to reproduce if needed. My question is there any other way to configure the ntp client so it can be always synchronized with the server using the ntpd -I ethx command.