Local DNS not reliable

I want to sum up this thread:

  • there are posted inconsistencies by @pslpsl
  • @pscar13 reports multiple IPs for the same host name. Unbound stores each (name, IP) tuple it gets by unbound-dhcp-leases-bridge. Maybe we should discuss handling of multiple names for the same IP, or multiple IPs for the same name. DHCP identifies devices by MAC, name resolution identifies by IP.
  • There was a big part of OT generated AI and unqualified posts. Therefore this thread has become big.

It is difficult to follow the discussion. If I have answered inadequate, this is caused by this fact. :frowning:

I throw in the towel :sponge:

@pscar13
do not!

maybe one is able to correctly split and clean here. :broom:

at least it should be recongnized
that you were the first one to point this flaw out :man_judge: :rocket:

the inconsistency prosted by pslpsl is an old one :mantelpiece_clock:
it was declared fixed though pslpsl shows it is not :roll_eyes:

yours however is quite nice :+1:
also one could assume that despite all these
many many many many other environments with ipf’s dhcp/dns
you have found a more than rare requirement :winking_face_with_tongue:
but it might be helpfull for future
rare requirement ipf users :person_shrugging:

My side of the argument comes from knowledge from MSDN class and what is observed when I look at other people’s programming in routers.

I’ve noticed that some VPN routers have a separate dhcp-dns bridge routine from the network when wireguard and openVPN are installed by default.

I’m sure I could find the netbios documentation from Microsoft since the subject of netbios they covered in a class on the subject. I just used AI to answer the question in 3rd party since that is documented somewhere.

If I knew if there is an environmental variable set to true when the client is a vpn, then I could pitch a code change. Because DNS-bridge should only use client-name on VPNs. So dchp bridge fallows the standard rules of hostname from ‘A’ DNS name entries or hostname advertised by client on DHCP negotiation.

The bridge generates for each (name, IP) tuple known an ‘A’ and a ‘PTR’ record. Latter for rDNS.

Knowledge is obtained from

  • host definitions
  • fixed leases
  • DHCP negotiation

As far as I know, there is no check for ambiguities etc.

Ouch! That’s not exacly sysadmin-friendly

A sysadmin should know, what he does. :wink:
Therefore he should have a documentation of the network, independant from the IPFire system. Ambiguities are contained there, also.

And thieves should not exists, according to egyptian monarchs since 5000 AC.
That should is kinda like some joke…

1 Like