Transcript
False meme image that says “bad news ipv4 fans. linus torvalds has announced removing ipv4 support from the linux kernel after the maintainers of the network stack got into a fight over WHAT KIND OF HRT gives the best results. this incident will impact 5 billion people and will make 95% of all network equipment on Earth binnable.” with fake screenshots of the linux kernel mailing list a girl calling another one a slur from 4chan over HRT choices and Linus Torvalds saying he will drop IPv4 support and asking the maintainers to learn to shut the fuck up.
You must log in or # to comment.
Followup:
I’m glad that Linus clarified that it was High Resolution Timers. I was honestly thinking they were arguing about Hormone Replacement Therapy.
Given the demographics of Linux devs, it probably would be the latter.
I know! It seemed totally plausible! There is a very odd discrepancy between that and the general population.
Trans people are either writing kernel code or playing hearts of iron 4. Nothing else is allowed.
THE TOASTER WILL BE REPLACED.
Fucking legendary quote from Linus.
It’s a shame it’s fake.
do you have proof that it’s fake though
There is the rule number one of the linux kernel: “We don’t break userspace.” Linux has refused fixes for buggy behavior in the past because of this rule. This would most certainly break userspace.
Also the alt text of the original image states this is fake.
Well I’m not going to switch away my perfectly functional mesh routers that uses IPv4 as using IPv6 on a local net that I may sometimes need to type in manually is rather stupid. And that would also bin my routers, so I’m not doing that either.
Oh well, I guess it’s been fun guys, no more Linux for me due to potential future security issues.
Do you even DNS?
No. But if this is true (which I do doubt completely, Linus can’t be this dumb to singlehandedly cripple his OS), this should also affect every intranet address.
The current description of IPv6 intranet is just ridiculously dumb anyway. Should I want to ssh into a local device, I’ll have to type in for example
fd9e:9aa0:c00f:1::a, with only thefdpart being the same for all intranets rather than192.168.1.10with192.168generally always being the same.Edit: wait… Are you telling me to set DNS redirects on all my local devices? Yeah, that’ll work, but why the even…
Pihole automatically adds clients that get an IP from its DHCP component. All my clients are server.local, client1.local, tv1.local, etc. So I can use their DNS name everywhere.
Even if it don’t want to use pihole(why?), you can edit the SSH config and add addresses for each host so you can just type
ssh serverOr I could also just edit
/etc/hostsif I’m just accessing stuff from my computer. I mean, I understand there are ways around this pain point. But, on IPv4, I wouldn’t have to do anything? Can’t really beat that, right?
I don’t think I’ve entered an IP address for a local device in years. Everything is accessible using
<hostname>.localthanks to mDNS. Avahi has been doing this for… 20 years I think?
I need to inflict this reply upon others. It is so gloriously cursed out of context. And in context too.
I’m kind of weirdly horny for Torvalds making unilateral decisions about long running controversies? Tell me what standards are best kernel daddy.
Maybe next he can ban tabs and ‘\t’ from Linux? Everyone indents with spaces now, debate over.
I’m sorry, WHAT? SPACES???
You ever find yourself lost for half a day indenting 1000+ lines of code in a random script you opened because the original developer was lazy?
No
Me neither. I’d use sed.
:%s/sed/vim/g
I think it is more likely that 95% of network equipment supports IPv6
Except whatever brings internet to my apartment apparently
Your ISP needs to setup IPv6 which isn’t trivial to do from scratch.
What provider is it?
Mine is Quantum Fiber, a sister of CenturyLink. CL has it, apparently QF doesn’t. Or at least not natively, rather 6rd. And then possibly not on the modem they installed? At any rate, I haven’t been able to find anything online.
It should have IPv6
Odd
🤷🏼♂️ When I had CL I could turn it on by enabling 6rd and it worked as expected. When I moved across town and got QF, their instructions didn’t account for it and following the same online instructions for CL don’t work. Others online seem to not have had any luck either, but some people’s comments make it sound like it’s the modem.
I’m ignorant enough that I didn’t realize this wasn’t actually happening until I read the comments. My networking knowledge is piss poor haha.
Maybe you shouldn’t treat random images you find on the internet as factual until you can verify the content.
I would have looked it up if the comments didn’t speak about it already. Don’t worry, I’m not a troglodyte when it comes to actually having critical thinking skills.
Can we just treat this as real? Post it everywhere. You know in a sort of “alternative facts” / flat earth sort of way. So maybe the planet finally moves on from ipv4.
Good news is that Google is reporting that >40% of their traffic is IPv6:
https://www.google.com/intl/en/ipv6/statistics.html
Should crack 50% soon. It’s been a long time coming.
I have a feeling, a lot of people willing to move to Linux are already open to trying out IPv6 when they get the time.
Pity my old router only bridges IPv4 and only does IPv6 with NAT (or maybe I just couldn’t understand the interface).
Maybe this is the push that finally makes IPv6 go mainstream!
Nah, IPX is gonna take over.
What’s IPX?
https://en.wikipedia.org/wiki/Internetwork_Packet_Exchange
Internetwork Packet Exchange (IPX) is the network-layer protocol in the IPX/SPX protocol suite. IPX is derived from Xerox Network Systems’ IDP. It also has the ability to act as a transport layer protocol.
The IPX/SPX protocol suite was very popular through the late 1980s and mid-1990s because it was used by Novell NetWare, a network operating system. Due to Novell NetWare’s popularity, IPX became a prominent protocol for internetworking.
(Just to be clear, IPX is not going to take over, and my above comment was intended to be humorous.)
It’s IPv10, which binds IPv4 and IPv6 together (4+6=10), sometimes known as Dual Stack.
A major ISP in the UK still doesn’t have any IPv6 support :( https://www.havevirginmediaenabledipv6yet.co.uk/
I still don’t have an IPv6 address over 4G with Vodafone. I want to run a web server on my phone, isn’t this a normal use case? Nat444 makes that pretty difficult, just let me use IPv6!
Cabled from Vodafone is not much better, ip6 does auto configure from the router with a local address, so it at least supports it. but no routable ips yet.
That’s impressive
Scaling NAT is complex and expensive. They are literally making it harder
They give 2 options: “Sign the petition” and “Tweet them!”, but you could also just en-masse “Switch providers” and make them rethink their strategy.
If you don’t use IPv6, you are behind. For me the transition was so hard, it’s a big step behind me, wouldn’t want to do it again.
I still remember my network lecturer telling us how IPv6 was the future and how we were running out of IPs back in… 1997.
That same year, during my work placement, I was introduced to Squid and all sorts of network fuckery to compensate for those supposedly disappearing IPs.
As long as IPv2 support remains intact, I’m good.
/s
Not today!
Tap for spoiler
Thanks, Voyager
NOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
I still miss IPv5
For real, what the hell happened to IPv5?
It was a little experiment in streaming that never went anywhere. It can be safely forgotten.
Lmao. Thankfully it’s not real and ipv6.disable=1 will stay in my kernel boot params forever. Just add more layers of NAT if you need more addresses.
There are sooooo many wrong things in such small post…
Name one?
That the internet is not longer accessible if both are behind a nat.
And we don’t start speaking about extra latency in the routers for the nat-lookup-table, the problem of the colliding ports, the mesh problem when you have bottlenecks that cannot be circumvent via bgp routing and so on…
Nat is a disgrace, an affordable one but a disgrace
ipv6 is great for public networks/ wan or extreme large nets. but for a small local net, ipv4 is so much easier if you need to access a single device. if ipv6 is parallell too, great but I hope ipv4 wont be disabled/ non existens in the future bc of this
IPv4 isn’t compatible with IPv6
I actually would argue that IPv6 is better even for small networks since it has SLAAC
Ipv4 is simpler and therefore easier for my brain to comprehend.
I deliberately disable IPv6 on all the devices on my home network because it’s really f**n annoying when some service tries to bind to localhost but picks up the IPv6 localhost instead of the IPv4 one
IPv4 is very much not simpler. You just as used to it.
Just remembering an address alone is much simpler.
4 numbers > a combination of numbers and letters in 8 groups
I’ve encountered way too many administrators and network admins who swear that “IPv6 does nothing but cause trouble” but the truth is, the trouble it’s causing is because you can’t half-implement IPv6. You either roll it out to the whole network or you don’t, and the longer you kick that can down the road the harder it’s going to be.
Basically too many professionals who haven’t learned a new technology since 2005 and refuse to try new things keep holding the world back
Basically too many professionals who haven’t learned a new technology since 2005 and refuse to try new things keep holding the world back
If it ain’t broke…
Imagine arguing that ‘solutions’ like NAT444 isn’t broke as fuck
I always bring it up when the network is experiencing problems that they wouldn’t have with IPv6. Running out of IPs in a given scope, increasing costs of public IPs, etc.
I don’t even have an ipv6 address, my ISP doesn’t provide them yet. Not much to do about it then lol.
You can maybe change ISP
Can’t even attempt to learn it if my ISP won’t provide addresses though.
Not been able to use it to even try, but doesn’t IPv6 not have subnets at all? No 192.168.1.1 on your local network with a different public facing 85.136.52.142 (and with NAT444 you also have ISP facing 10.183.23.6). So does your ISP provide you a range of IPv6 addresses?
Yes, your ISP provides you a large quantity of adresses. Not really, the adresses has several parts. Your ISP provides you with the prefix. Your devices complete the rest of the address automatically. You can also use a DHCPv6 server, but I don’t and some devices don’t support it anyway. Yes, all those adresses are globally routable, they are “Internet” adresses. You can still use locally routable adresses too if you want, called Unique local address (look it up on Wikipedia), but that requires manual configuration.
I don’t think unique local addresses require manual configuration. On linux at least, I get an
fe80::address derived from the interface’s MAC address even if there it can’t find any router. If the host receives a router advertisement, it will add a local address (the same suffix as the fe80 but with afd8b:something::/64)and the “internet”2003::.I’m not an expert and this may be just the configuration of my router, but all my linux installs automatically got these three addresses without manual configuration or issues.
That’s SLAAC not a ULA
You are confusing unique local adresses and link local addresses. Unique local adresses can only be configured manually or, in theory, with DHCPv6. On Debian, I edit the file “/etc/network/interfaces.d/<interface name>”:
- auto <interface name>
- iface <interface name> inet dhcp
- iface <interface name> inet6 static
-
address <unique local adress of your choice within the official range> -
autoconf 1 -
accept_ra 2 -
privext
This gives you: autoconfigured IPv4 address, autoconfigured (slaac) IPv6 address, an IPv6 unique local address, temporary IPv6 adresses (privacy extensions) and your IPv6 link local address.
fe80::That’s a link local address [0].
fd8b::That’s a ULA [1]
2003::This one is a globally routable address (Global Unicast Address, or GUA) [2].
As you observed, link-local addresses are generated completely independently. ULAs and GUAs are self-assigned using SLAAC or assigned by a server using DHCPv6 after your host has seen a router.
For a GUA or ULA to be assigned, the router or DHCP server has to have a prefix delegated to it. A GUA prefix would come from your ISP. A ULA prefix would be configured on the router itself. If yours has one without you setting it up, maybe it does that by default?
[0] https://en.wikipedia.org/wiki/Link-local_address [1] https://en.wikipedia.org/wiki/Unique_local_address [2] https://www.iana.org/assignments/ipv6-unicast-address-assignments/ipv6-unicast-address-assignments.xhtml
Yeah, I guess my router just decided on an ULA prefix on its own. Thank you for providing the right terminology and explaining how a host gets these addresses.
Correct, the ISP would assign you a /56 of public IPs that all share a prefix which you can slice and dice into however you see fit. All devices receive a publicly routable IP which your router/firewall would limit access to. So no running out of IPs ever, no network/IP collisions if you have to connect to another private network, etc.
Why can’t you just use it on your local network? Don’t need ISP for that.
“IPv4 is running out of IP addresses so therefore every local network needs to move to IPv6” is a full clown move.
IPv6 is literally designed to solve the scalability issues with IPv6
First of all, enterprises usually have at least one public IP (the one I work at right now has more public IPs than they have server VMs)
Secondly enterprises have big enough and complex enough networks to see other benefits of IPv6. For example IPv4 has some problems when broadcast domains are too large, so your internal network sizes are artificially limited when following best practices. Without private networks you don’t have to worry about IP collisions between different private networks that you have to route between (comes up more than you’d think!) etc etc.
I will happily enable and use it once doing so doesn’t break any of my connectivity.
I’m not managing an enterprise network, it’s just my home, but my ISP doesn’t support IPv6 so that’s one extra layer of complexity right off the hop. On top of that internal services switch which previously required no manual configuration just seem to randomly not work.
IPv6 is not going to see widespread adoption unless it can be implemented completely transparently for the end user, full stop.
IPv6 is widely adopted
In a local network there is no point in using ipv6.
It is interesting when you run out of ip addresses for the amount of devices you have.
So in the open Internet.
Unless I am missing something.
Why not? It’s as easy as ipv4
A couple of things that IPv6 does better for local networks is link local addressing (fe80::). and multicasting.
In IPv4, they kind of hacked something out of 169.254, but if you have more than one NIC, it pretty much becomes useless.
If you have a service designed explicitly never to be accessed over a router, then you can live in fe80:: a lot more easily than trying to do the same thing with 169.254.
Is this real? I thought that Linus has improved his demeanor a bit in the recent years?
You can easily tell it’s fake. Linus would never censor “fuck”.
and it would be Linus telling þe committer þeir patch sucks ass.
Why do your th’s look like that?
Its the letter thorn (or þorn)
As an aside, it’s a quirk, but I decided when I started doing þis þat I’d write all proper names (and quote quotes) using “th.” So alþough I use “þ”, I write it “thorn”. And “Matthew”, and so on.
It’s an arbitrary decision. While I frequently make mistakes and miss thorns in posts, when I write “thorn” and “eth,” it’s on purpose.
In Middle English, it would have been “þorn” and “eþ”, þough. Maybe I should make an exception for þose?
