There's a lot of advice floating about the Internet on configuring IPv6 on Xfinity/Comcast networks, but none of it applied 100% to my local situation. I therefore wrote up what worked here with RouterOS 7.14.

While I welcome advice from the IPv6 gurus here, do keep in mind that what you see in that article was arrived at after much careful experimentation, striving for the minimum set of changes to the default configuration to get everything working.

The configuration depends on if you have mid split enabled or not. It looks like you do not have mid split enabled yet (yours looks like my old config for the default route). Once mid split was enabled in my subdivision, I had to set the default gateway to the interface name. This is most likely tied to the new vCMTS that are now in use with mid split.

I doesn't surprise me at all that my working configuration is contingent on local NAN-scale engineering decisions.

(NAN = nuke-area-network. Smaller than the Internet, bigger than a typical WAN.)

tangent, thank you for an excellent write-up. I tried following every step but in the end, it did not work. Kept telling me that there is no pool name ipv6.

Kept telling me that there is no pool name ipv6.

That pool is dynamically created by the first step in the article, "/ipv6 dhcp-client add …"

Note that the command can be given on one line without the line breaks and the backslash, or it can be given on two lines, with "/ipv6 dhcp-client" getting you into the right section of the CLI and then the other two lines being treated as one, with the backslash continuing the command from the second to the third line.

If that is not where it is failing in your configuration, say where it does fail. Paste the command you typed and the error it returned into a "code" block here.

@biomesh how would I tell if mid split is enabled? I had a working config for a while getting the default route via DHCP, then suddenly stopped working (could no longer connect to the gateway it was setting). So I just followed the advice of @tangent and turned on RA instead, but there was no default route set. If I set ::/0 to %ether1 manually, everything works fine.

The only other difference I have is that I'm using a hint to request (and get) a 60-bit prefix. Not sure if that would muck with the RA though.