dhcpcd-discuss

Great support, thanks!

No further questions -- just a couple notes for the record of the next 
person that finds this in the archives.

Jeff


On 10/4/17 1:17 PM, Roy Marples wrote:
> Let me start off by saying holy wall of text batman!
Apologies! I've gotten "answers" that were far from applicable as it 
isn't a "normal" home-user application to not to use the PD directly on 
their "router"

> I'll try and answer as best I can, if I miss something, poke me.
I think you hit it all pretty well. I've been poking at the IPv6 part of 
it pretty hard, and will deal with the "easier" part of the IPv4 
implementation when I'm not remoted-in over IPv4...

> > Right now, dhcpcd is being started as a systemd service:
> You might want to look into the dhcpcd-online(8) command installed by 
> the dhcpcd-ui package I also provide but don't maintain as much as I 
> would like. This provides excellent support for systemd to inform it 
> when it's up - works similar to nm-online from network manager.
Great suggestion -- I'll take a look at that.

> I do a similar thing on my NetBSD powered erlite router, also 
> connected to a DOCSIS modem. Sadly, no IPv6 support from my current 
> ISP :(
>
> Here's my current /etc/dhcpcd.exit-hook to handle this
> [...]
Again, thanks, I'll definitely look at that approach this evening.

> Remember that clientid is literally setting the ClientID option - it's 
> not a DUID as such.
> So you need to encode correctly - ff is required, the next 4 bytes are 
> the IAID and the rest is the DUID. See RFC4361 for exact details.
>
>     clientid ff:00:11:22:33:AA
>
> Of course, you could just use the duid option (enabled by default) and 
> then supply your own /var/db/dhcpcd/duid (file location varies per 
> platform) and then just set the iaid per interface definition
AH! That's the part I was missing!  That seems cleaner and more reliable 
than the overwrite of /var/db/dhcpcd/duid that I tried (successfully)

> [Not delegating the PD to an interface with the following syntax] is 
> possible, but not elegant!
>
> interface enp1s0
>     ia_pd 1//60 foobar
>
> The trick is to delegate to the foobar interface, which does not exist ;)
>
> I'm not sure why you would want to do this though.
I've got it working from the command line now, pleased to find that the 
following worked and obtained the same IA_PD I was getting before, where 
aa:bb:cc:dd is the IAID previously in use.

interface enp1s0
    ia_na 0
    ia_pd aa:bb:bb:dd/::/60
    ipv6rs
    ipv6ra_noautoconf