IPFire’s DNS seems to fairly quickly lose knowledge of which machines are connected. I have many machines defined with static addresses in the DHCP server. In the morning when the computer starts up, it gets registered and I can ping by name since the DNS has it registered. But after some time (4 or 5 hours is long enough), the DNS resolution will fail for that machine.
Perhaps it would be less of a problem if I kept my lease timeout at a small number like 6 hours. I currently have it set to 31680 minutes (22 days). (NOTE: I have now changed my lease time to 360 minutes - lets see what happens.)
I’m not sure what triggers dropping it from DNS, because some machines don’t seem to have the same problem. (For example, my server rarely reboots, and yet DNS can always resolve it, and it is not defined in HOSTS, so it should be treated just like any other entry from DHCP.)
This has been happening for a long time, and my impression is that it has generally improved in the past year. I think I have noticed it immediately when I switched to IPFire from IPCop several years ago, but I hoped it would go away in time.
In any case, I would kind-of expect that a statically defined entry in DHCP would ALWAYS be resolvable.
Currently I am on core 149.
After running /etc/init.d/unbound restart, I can ping that machine by name again.
I believe there is a script that periodically runs that transfers the DHCP info to DNS (unbound)'s /etc/unbound/dhcp-leases.conf - likely the solution is to be found there. (/usr/sbin/unbound-dhcp-leases-bridge)