This behaviour can be observed on CentOS 7/8 and derivatives such as CloudLinux/AlmaLinux.

It appears to be an issue where Directadmin is started prior to the network properly being established with DHCP.

To fix the issue we can either assign the servers ip address to the interface statically. However to maintain portability such as cloning of the server we should manipulate the DirectAdmin service.

# Alter the After statement for Directadmin service to wait for named
sudo systemctl edit directadmin

# In the override editor add the below

[Unit]
After=
After=syslog.target network-online.target named.target

Explanation of what the customization does is essentially unsetting the After statement for the original unit file and then adding on the named service with “named.target”

Once this action is completed upon reboot even if the interface address is managed by DHCP directadmin will start normally.