Wicked and DHCPv6

I have a dual-stack openwrt router, and dhcp6 in server mode with both stateless and stateful.

In order to have the stateful address registered in the (openwrt) dns-server, I apparently have to configure


The default “yes” value does configure ipv6-address on the host, but it is not registered in dns (nslookup returned only the ipv4 address).

Any one else experienced this (not necessarily with openwrt) ?


Did you try to resolve the DNS name with “dig”?
[HR][/HR]I currently have, with “wicked”, the following DHCP Network settings:


  • On this LAN/WLAN, “nslookup
    ” reports only the IPv4 addresses of Android mobile telephones – for everything else the IPv4 and IPv6 addresses are reported. - “dig
    ” however, also reports the IPv6 addresses of the mobile telephones. - And, in the DHCP server / DSL Router, the mobile telephones are reported as having IPv6 addresses in addition to their IPv4 address on this LAN/WLAN.

BTW, my DSL Router has, setup by my ISP, a IPv6 DSL connection – IPv4 is via a “DS-Lite-Tunnel” – from what I can ascertain, the IPv4 address used by the Router is shared with other customers of my ISP …

Well, this option makes wicked send CONFIRM message which does not include FQDN option and the server is only required to check “whether the addresses in the Confirm message are appropriate for the link to which the client is attached”, there is no requirement that server had fully instantiated client configuration including DNS settings. So it seems to work as designed. That actually sounds like oversight in protocol.

only the ipv4 address

DHCPv4 send REQUEST with hostname option in this case so server can act on it.

That seems to explain it, yes. Since this was a new installation (a VM) with a new mac that hadn’t been used before, the DHCP/DNS-server had no previous registration for a possible update.

@dcurtisfra](https://forums.opensuse.org/member.php/42186-dcurtisfra): no I didn’t try dig. Because I’m not too aware of it, so I never think of it.

Now I lost you. wicked is only using Confirm if it has existing valid lease obtained from DHCP server earlier. If it is new installation then VM should not have any lease and should use normal DHCPv6 exchange to obtain one.

Ah, then I misunderstood. I thought it would only update an existing DNS record when “use last” was “on”, and not create a new record if it didn’t exist. Could be something with dnsmasq that openwrt use then, maybe?
And while it was a new VM in the sense that the hw-definition where new (i.e. a new mac), it reused the hostname of the VM it replaced.

Anyway, really not an issue for me as long as setting DHCLIENT6_USE_LAST_LEASE=“no” works, I was only curious as to anybody else also experienced it.