The time has come: we must enable IPv6 entirely

ipv4 servers and their A records would remain. I don’t think anyone is suggesting removing them.

Possibly after 20, 40, 60 years closing down ipv4 NTP servers could be a small part of the larger ‘closing down ipv4’ conversation???

1 Like

Sorry Steve, IPv6 will never take over IPv4 the way it’s running right now.
It’s too complicated and not everybody wants to use public IP’s all over the place.

Take my word for it, they will change it’s working in the next couple of years, being compatible with the current but make it look more like IPv4 if you wish.

They are saying for 9.5 years now the internet will be IPv6, still not happening.

Nobody can look into the future, so we’ll see. But Steve made no such predictions, he just said he is personally in favor of increased IPv6 usage. Nothing wrong with that.

Actually there is already quite some IPv6, as I have substantiated in a number of previous posts. I’ll be happy to provide the numbers again, but I trust that won’t be necessary.

There is not. Sorry there is not. Sure some ISP’s give your modem dual stack.
Most internet users don’t care what it is, as long as they have internet.

The reality is that at the server side hardly anybody supports IPv6 at all.

Also the IPv4 range hasn’t been used to the max, you can still get IPv4 static IP’s for cheap.
More and more big-tech’s releasing millions of IP’s to the public…

Yes I can predict the future. It’s easy, unless they drastically change IPv6, it will die a slow death.

You simply can not force things on people, not now, not ever.
As they still force silly rules and public-IP’s for all devices, it is not going to happen.

Let’s talk again in 10 years…when it has been abandoned if they still didn’t listen to people.

Off course they don’t care - which is why it requires techies like us to keep things running smoothly, without bothering the average user too much.

You just keep making those claims, without providing any quantification. Some 54% of all participating NTP-servers of the pool, are reachable via IPv6. That is a lot of potential that is currently not being used to its fullest extend.

This is not about squeezing the last bits out of IPv4 and whether or not that is the best way forward. This is about harnessing the potential of the many IPv6-reachable NTPpool servers that exist at this very moment, sitting there almost unused. While there is an ever increasing amount of IPv6-capable clients (millions and millions worldwide), anxious to use them via IPv6.

Why are you writing this? Nobody is forcing anything upon anybody. Think of it this way; did anyone forced you to do anything when they made this very forum-website available over IPv6? I don’t think so. It just happend, no one complained.

IPv6 adoption steadily increases in quite a few places. So it is appropriate to remind this community that there is sufficient IPv6 to justify adding a few extra AAAA’s to the pool’s DNS, just as @ask intended to do already back in 2017.

Well Marco,

To please you I have added 2 IPv6 servers to the pool.

I also tried to push my other 2 servers on IPv6 but it’s impossible to get them pass a Fritzbox.
I spend another 6 hours but I can not get them to work.

For some reason the Fritz simply doesn’t want to “forward” inbound.
I’m probably too stupid to get IPv6 past a router.

That is my problem, it’s too hard to do.

Anyway, I do want to get them all 4 working on IPv6…but it seems nobody knows how or explains it.
Sorry mate.


Non-authoritative answer:
Address: 2001:41d0:a:4558::9752:705b
Address: 2a00:7b80:477:21::9dcb:d174

Very much appreciated! :grin: The two you mention seem to be working well. [1]

Getting it to work with a Fritz!Box isn’t that hard, once you’ve figured it out. I’ve added a screenshot of my setup. Maybe it is of any use. If you are interested to find out how much of your NTP traffic is IPv6, you may like this little tool I wrote.

Although I am naturally very happy with any/your additional IPv6 capable servers and welcome them, there is this other thing that the NTPpool maintainers should do, which is even more important at this moment. And that is that they should change the DNS of the NTPpool, so that not only hands out AAAA addresses, but [013] as well.

Please note there is more ways to accomplish this. You can do it in one go (internet access via IPv4 and IPv6), or in two separate steps, depending on your specific situation.

:warning: Also, obviously a fixed address is required (just as with IPv4). Quite often ISP’s suddenly hand out a different IPv6-range and/or different IPv4-address, which can spoil the fun. :warning:

[1] 2a00:7b80:477:21::9dcb:d174 seems to have some issues from some sites, but I would have to check closer to be sure. UPDATE: tested it from the NLNOG-ring and everything looked fine!

1 Like

You have a completly different screen then I have.
I use the 7590 with 7.28 firmware.
And it doesn’t say Port Assigned IPv6.
Nor does it give an option to select sharing for IPv4 and IPv6.
How do you get that? I did delete the server in total and started again, it doesn’t do that.

That is a DNS-cache issue, due to testing on my part, it will resolve properly soon.

I can think of several possible issues. Is your Fritz!Box a router, or only a bridge? The menu can change significantly between these two. Also, do you have IPv6 enabled? Other than that there are probably more possible causes for this difference. I’m not sure about your ISP-connection. Is it DSL? Things like that. In any case I’m on 7.28 too with my 7583, so your firmware is certainly up to date.

But again, don’t bother too much. I didn’t start this discussion to persuade you to enable IPv6, but rather to ask the NTPpool admins to put more AAAA records in the DNS.


Same setting, use natiove IPv6 and enabled.

The problem is to get it passed to the servers.
As I traceroute or ping everthing works, but no inbound.


I noticed your IPv6 servers are scoring well and by now they should be receiving IPv6 traffic. If you look closely, for example with this tool I made, or just with tcpdump or so, you’ll find that the amount of IPv6 traffic is lower than could be expected - given the amount of IPv6 capable clients that exist today. The reason for this is that only 2[.x] has an AAAA record. To fix this, the pool admin(s) would have to add AAAA records to the other DNS names as well.

Again, this is not about pushing anyone towards IPv6. This is only about making it possible for those who can and want, to make use of the available IPv6 servers that exist in the pool, such as the two you kindly provided recently. We need those additional AAAA’s for that, and only @ask seems to be able to add them.


I did actually create a patch to enable ipv6 everywhere a few years ago, which was closed without further comment recently 8-(

1 Like

Hi Marco,

I deleted all my servers from the pool again.
The peer(s) from the monitor worked a bit and the monitor is going bad again.

The bad monitor has been an ongoing problem over the years.
I have quite enough of the pool.

My servers will not return until the monitor is fixed once and for all.

Couldn’t you have at least kept the IPv6 ones? :innocent:

That would have made a perhaps even stronger statement. Monitoring is (at least for the time being) working well on IPv6. :slightly_smiling_face:

Hi Marco,

You can use my servers, they are listed as:

IPv6: (2 servers stratum 2)
IPv4: (4 servers, 1x stratum 1 and 3x stratum 2)

I see no point in keeping my servers listed when a monitor can’t check them properly.

I do feel that Ask isn’t taking our complaints serious or doesn’t have time to solve the problems.

I’ll be back when the issue is solved, until then, sorry no servers to support the pool.

It’s your choice, but it’s not a technical statement. Even with a floppy monitor, your servers can still serve clients that have previously peered with it.

1 Like

It’s almost 2 years of analysis paralysis and off-topic discussion later. Or 10 years later. Or… well, it depends on when your epoch starts.

As of today, still only is handing out IPv6 addresses.

This thread diverged into discussions of the merits of IPv6, how to debug a particular IPv6 server, and more.

The main issue wasn’t resolved. Can we get the pool administrators to put AAAA records into more of (all?) the pools? There seems to be a catch-22 (circular) dependency: ntp over IPv6 isn’t used (in large part) because it’s not in the default pool; it’s not in the default pool because usage is low. (And some concerns about breaking a small number of legacy devices.)

Alternatively, can we get a new hierarchy setup that includes all servers - e.g. *
{0,1,2,3}{,.us,.ca,.europe,asia,vendor,…} (or wholepool, bigpool, bestpool,…)

Personally, I prefer the former as it’s less work for everyone and will have the fastest uptake.

A parallel structure may seem to have lower risk, but it will take years to be adopted. I run some very old equipment, and haven’t seen issues pointing it to Still, it would be better to have a slowly growing than to do nothing for another decade.

As for the name: if a parallel structure is the chosen option, let’s not spend a year choosing one. It should one that says “choose me unless you have a reason not to”. The current names should be documented as ‘legacy’, ‘limited’, ‘deprecated’ in documentation. But of course, should continue to operate, be fully supported, and share data with the new one.

This comment is not about whether you like IPv6, have IPv6, or don’t believe in it. Please don’t bring those issues up again. It’s about making the pool support the systems that use IPv6. Or can.

As a side note: How do I setup NTP to use the pool? doesn’t mention IPv6 at all…and should at least mention that (currently), if you need IPv6 service, you need to use 2.pool…


I don’t think there is much risk, so why go through this hassle? It’s probably simpler to create an IPv4-only ‘’ for the laggers, if any, that might run into trouble otherwise.


A would be fine too.

As I said, I’d prefer option 1 - populate the existing pools with an equal number of IPv6 addresses.

But it doesn’t matter to me which plan is selected. Pick one and do it. Whatever can be actually be done. If it takes some “hassle” to get off dead center, so be it.

This has been stuck in limbo for far too long, with no progress. Actually, no motion. Better to learn from a mistake than to do nothing. “The perfect is the enemy of the good (enough);” Or as Samuel Johnson said, “To do nothing is in every man’s power”. However, that’s not a productive power trip…

If actual problems arise, there are lots of solutions available. Even if and a dns server are hard-coded in some ROM, a box that policy routes the dns server address to a forwarder that strips the AAAA records could be trivially created on a SBC. If the cost of the box is less than the cost of upgrading whatever device is confused. But I’m not proposing such a box be a gate. Fixing hypothetical problems is another delay tactic.

How about turning on IPv6 equality for a week or so and see how many issues, if any develop? And if any clients switch to the legacy pool, try to find out why. If some complain that they can’t, learn why.

At least that way there’d be some data on whether there’s a real problem, exactly what it is, and the size. If it’s a disaster (which seems unlikely); roll back. If not, address the issues.


I have been supplying four ipv4 and ipv6 servers for a year. Yesterday I put the ipv4 servers in ‘monitoring only’ mode. It will be interesting to see the use decrease.

I would like to see ipv6 used on all pool addresses. I have communicated with @ask, and received a polite answer. No promise of progress.

I’m very frustrated. I will delete all the servers in another month if there is no progress… save myself $AU200 per year and the angst.

1 Like