With the implementation of Kea DHCP for IPv4 I was wondering if there was plans to implement Kea DHCP IPv6?
Yes, maybe 24.7 if all goes well. We will discuss roadmap stuff in two weeks.
Cheers,
Franco
Quote from: franco on February 01, 2024, 05:20:22 PM
Yes, maybe 24.7 if all goes well. We will discuss roadmap stuff in two weeks.
Cheers,
Franco
Excellent, look forward to it. Thanks Franco :)
Quote from: franco on February 01, 2024, 05:20:22 PM
Yes, maybe 24.7 if all goes well. We will discuss roadmap stuff in two weeks.
Cheers,
Franco
Anywhere that I can read more about this in the meantime? Will ISC still be supported in 24.7? I'd rather wait to migrate until there is full IPv6 support.
I have been running a pair of ISC Kea DHCPv4 and DHCPv6 servers on linux for over two years since ISC stopped maintaining ISC DHCP at the end of 2022. Personally very happy to switch because of the ISC Kea ability to handle DHCP Failover for DHCPv6 which was not supported in ISC DHCP.
Quote from: franco on February 01, 2024, 05:20:22 PM
Yes, maybe 24.7 if all goes well. We will discuss roadmap stuff in two weeks.
Cheers,
Franco
I switched over to Kea DHCPv4 and it works great. I haven't run into any issues using Kea DHCPv4 and ISC DHCPv6 together, but it would be nice to have the latest and greatest running both versions. Hopefully we will see Kea DHCPv6 in 24.7. I didn't see anything in the current roadmap for 24.7, so I'm a little worried that we wont see it until 25.1
Is there any update on this?
When can we expect too have KEA DHCP functionality for IPv6?
Basically... later. There is no roadmap and no immediate need.
Cheers,
Franco
+1
To make matters better using unforeseen directions we're likely going to add DHCP to Dnsmasq instead. To do this we will first move Dnsmasq to a MVC/API implementation for 25.1.
https://github.com/opnsense/core/issues/7905
I'm sorry to say that in this case the plan changed, in part due to Kea not delivering on its promise as much as it delivered on making DHCPD EoL a long time ago already.
Cheers,
Franco
Quote from: franco on September 25, 2024, 09:34:11 PM
To make matters better using unforeseen directions we're likely going to add DHCP to Dnsmasq instead. To do this we will first move Dnsmasq to a MVC/API implementation for 25.1.
https://github.com/opnsense/core/issues/7905
I'm sorry to say that in this case the plan changed, in part due to Kea not delivering on its promise as much as it delivered on making DHCPD EoL a long time ago already.
Cheers,
Franco
Does that mean KEA DHCP will be deprecated in a future release of OPNSense? Or just for IPv6? Or not at all? I may be reading this wrong and wanted to make sure.
No. The idea is:
* DNSmasq DHCP becomes the default DHCP
* ISC DHCP moves to plugins
* Kea DHCP will gain DHCPv6 eventually
The mere fact that Kea is the only maintained effort for HA features will make it stick to the core, but honestly we do not like to see it become the default.
Cheers,
Franco
Thank you for the update
Quote from: franco on September 26, 2024, 08:46:48 PM
* Kea DHCP will gain DHCPv6 eventually
that would be nice because of the HA features ;)
Quote from: franco on September 26, 2024, 08:46:48 PM
The mere fact that Kea is the only maintained effort for HA features will make it stick to the core, but honestly we do not like to see it become the default.
can you explain the reason?
Kea doesn't appear to have a high adoption curve and the development doesn't seem to be in a rush to accommodate that either.
But truth be told our current basic gripe with Kea is that for emulating advanced features that work fine in ISC DHCP you find sparse or incomplete documentation and ending up reading the Kea source code is a good waste of valuable coding time.
Cheers,
Franco
Quote from: franco on September 26, 2024, 08:46:48 PM
No. The idea is:
* DNSmasq DHCP becomes the default DHCP
* ISC DHCP moves to plugins
* Kea DHCP will gain DHCPv6 eventually
The mere fact that Kea is the only maintained effort for HA features will make it stick to the core, but honestly we do not like to see it become the default.
Cheers,
Franco
Am I correct in assuming that DNSmasq only supports IPv4? I am currently using a mix of KEA for IPv4 and ISC DHCPv6.
I guess what I am trying to understand is what will the default for DHCPv6 be? KEA or DNSmasq?
https://kasiviswanathanblog.wordpress.com/2017/06/04/dnsmasq-a-simple-dhcpv6-server-for-embedded-devices/
FWIW, Dnsmasq rework moves further into 2025 territory. Just too much other priorities at the moment.
Cheers,
Franco
Quote from: franco on November 08, 2024, 08:01:13 AM
https://kasiviswanathanblog.wordpress.com/2017/06/04/dnsmasq-a-simple-dhcpv6-server-for-embedded-devices/
FWIW, Dnsmasq rework moves further into 2025 territory. Just too much other priorities at the moment.
Cheers,
Franco
Thanks @franco.
So given the findings would you suggest reverting back to ISC DHCPv4 and ISC DHCPv6 until further changes are made in 2025?
Yes, sure.
Cheers,
Franco
Hi Franco,
Can you give us an update on the current status of the plans for the KEA DHCPv6 implementation?
Thanks in advance,
Joerg
No plans. Dnsmasq will get DHCPv4/v6 support this year for sure.
After working on another Projects with Kea, i do agree with franco.
Most features that are mandatory are paid features, a lot is not supported and the raw and the udp socket is pure crap on linux.
The one doesnt work with vlans the other works with vlans but has issues with dhcp forwarders from the switch.
We got it working with rewriting dhcp packetheaders on our Core-Switches.
However its a shitmess with Kea, isc-dhcp worked absolutely perfect in every constellation, dhcp registration worked flawlessly and it was super easy to add dhcp options.
On Kea even dhcp registration in combination with powerdns is a mess, it works 99% but sometimes clients doesnt get registred or removed.
However we had no choice.
The ONLY nice feature of Kea is actually HA, nothing else.
Kea writes a lot of "Performance" on their Pages, but the real Performance of Kea is the worst ive seen. Even isc-dhcp is delivering clients an ip-address faster.
And dnsmasq is compared to Kea, lightspeed.
The Development of Kea is utterly slow either, nothing changes, no new features, its even impossible so search for leases with wildcards etc in stork.
Basically you have to know the exact ip or mac-address to search though leases. You cant even list all leases.
So its completely senseless.
The paid plugins are horribly expensive either and without them kea is almost useless tbh.
However, the only nice feature is HA, otherwise Kea is basically crap in my opinion that we are forced to use for the moment.
Cheers