DHCP is set up so that IP addresses are fixed assigned to mac addresses of allowed clients. There is no freely allocated address. In Pakon statistics, one client identified only by a MAC address is registered. I can assume that this is a client in guest wifi network. If I saw the allocated IP in the guest range (10.111.222.100-105) I would be sure. That’s how I have my doubts. How do I find the truth?
Is this solvable in Luci-Hostnames by assigning names for the specified IP range of Guest ?
the solution is to assign names even for guest IP range in Luci-DHCP and DNS- Static Leases settings. It will be reflected immediately in the Pakon output
This puts hostname IP-address tuples into the system, which will only be useful if the IP addresses are statically assigned. I believe this effectively allows to query dnsmasq for those hostnames.
This allows to identify hosts by their MAC addresses, and then assign a static DHCP lease which in turn will result in statically assigned IP addresses, making them suitable to be injected into the domain name resolution.
Pakon requires the MAC to name mapping from the latter, as far as I understand, given that otherwise pakon returns raw MAC addresses.
I use only fixed IP - these are displayed in Pakon.
Furthermore, I have set different hostnames defined in Luci - Network - Hostnames … do not manifest themselves to me anywhere (Pakon) and their Function meaning is incomprehensible to me.
So both defined names are resolved by the domain name service.
But in foris’ pakon all I see is C610A-IP, so both hostnames and static leases will be injected into the omnia’s DNS, but onlt static leases offer the required MAC to hostname mapping that pakon seems to require. Which makes sense, as it is far easier to change an IP address of an endhost (and DHCP might do that automatically anyways for non-static address assignments) than the MAC address. Sure, dedicated opponents will also be able to spoof the MAC address, if they realize that MAC addresses are monitored, but as always security is never absolute.
=== >> But it doesn’t really matter… the meaning of hostsname is not clear. If I cancel these records " LuCI->Network->Hostnames:… What’s going to happen? Something will be wrong ?
Are you sure you clicked “Save&Apply” in LuCI, otherwise you see the entries on the web page, but they are not applied…
This is benign, if you call nslookup Fujitsu_kuchyn.lan not from the turris router but from a different host it will resolve just fine, as far as I can both hostnames and static leases will be added to /tmp/kresd/hints.tmp. See cat /tmp/kresd/hints.tmp to confirm.
I keep repeating it, hostname really is just a method to inject hostnames into the routers domain name server, just note that the local network top level domain (by degault .lan) will be added.
They will not resolve any more… BUT if you also assigned static lease entries in DHCP and DNS then these are still auto added to kresd and will keep working, assuming you enabled the “Enable DCHP clients in DNS” checkbox in foris’ DNS tab.
As far as I can tell the hostnames mechanism is older, and the fancy have the DHCP server tell the DNS server option developed only later. I prefer the fancy option, but I understand why the other mechanism exists as well…
Well, nslookup returns exactly the address listed in hints.tmp. The error " Can’t find Tabl_Jirina.lan:" is related to calling nslookup on the router if you do the same on a different host in your network this error will probably not show, I believe this is benign but might be somethingvfor team turris to look into.
Yes, as I said if static leases are in use and kresd/Foris/TOS is configured to inject DHCP names into DNS the hostname mechanism is redundant. But I have the feeling I failed to convey that message, but am happy you figured it out by yourself in spite of my confusing posts.