Well, this is embarrassing!
We issued new FireBrick code today...
The main changes are work in PPPoE and DHCPv6 clients on PPPoE to make it all a lot slicker and handle IPv6 prefix delegations from your ISP.
As always, before a release, we did lots of testing. We even found FTTC PPPoE a tad sluggish and had to allow longer timeouts for them. We tested on 20CN, 21CN, BE, FTTC, direct connections to a FireBrick BRAS, all sorts. All worked.
So we did a beta and a few people tried it and all well - so we made a factory release - that means FireBricks updating automatically.
And then three FireBrick customers went off line and stayed off! People using PPPoE.
OK, We have a system - we withdraw the release, which we did. That stops the auto upgrades. They are spread over 24 hours for a reason! I was in Azeroth but got dragged back to reality to check this...
Found the problem - we had not tested legacy IP only connections. It seems a bug in PPP negotiation timeout caused that to fail if there is no IPv6 on the line.
The problem is that all of our test lines and equipment have IPv6. Of course it does. Why would we have a line with just the old legacy Internet protocol? The new IPv6 has been around over 15 years, so who is not using it? Really? Heck, IPv4 ran out last year.
We're sorry about that. We have issued new code that fixes it right away.
And we'll test old legacy IP protocols in future. It is really embarrassing, sorry.
P.S. In the unlikely event of an A&A broadband customer using a FireBrick with PPPoE and not having IPv6 and having been upgraded so now off line - we just have to add IPv6 to your line. I think customers can even do that on the control pages. But who is using A&A and a FireBrick and not using IPv6, seriously?