Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Wednesday November 07 2018, @03:12AM   Printer-friendly
from the BAD-design-(Broken-As-Designed) dept.

In June 2012, an owner of a Samsung Galaxy Nexus running Android 4.0.2 opened a case in Google's Issue Tracker requesting support for Dynamic Host Configuration Protocol for IPv6, otherwise known as DHCPv6, or RFC 3315, which allows for stateful address and connection configuration on devices joining an IPv6 network. DHCPv6, like DHCPv4, is commonly used in enterprise networks for connecting devices.

For the last six years — including through five new major versions of Android — that request has languished, when this week it was marked as "Won't Fix (Intended Behavior)" by Google engineer Lorenzo Colitti. Android is effectively the only platform which lacks support for DHCPv6, making the IPv6 implementation on Android incomplete.

https://www.techrepublic.com/article/androids-lack-of-dhcpv6-support-frustrates-enterprise-network-admins/


Original Submission

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 2) by MichaelDavidCrawford on Wednesday November 07 2018, @03:50AM (8 children)

    by MichaelDavidCrawford (2339) Subscriber Badge <mdcrawford@gmail.com> on Wednesday November 07 2018, @03:50AM (#758820) Homepage Journal

    That's two major releases ago. Desktop Macs have supported IPv6 for eons. I recall quite a while ago the Airport Wireless Access Point disabled IPv6 by default but it was still available as a configuration option.

    Comcast as far as I can tell supports it.

    Dan Bernstein is _never_ going to get his magic moment [cr.yp.to]. I expect the Internet of Things will make the IPv4 address crunch far worse.

    --
    Yes I Have No Bananas. [gofundme.com]
    • (Score: 0) by Anonymous Coward on Wednesday November 07 2018, @03:55AM (1 child)

      by Anonymous Coward on Wednesday November 07 2018, @03:55AM (#758821)

      The only reason "macOS" has had it so long is that it used to be developed by ideologues, you know, before Apple started making serious money in the modern era. They've all left or been pushed out.

      Android, however, was always total trash. Just look through AOSP for yourself.

      • (Score: 2, Interesting) by realDonaldTrump on Wednesday November 07 2018, @07:03AM

        by realDonaldTrump (6614) on Wednesday November 07 2018, @07:03AM (#758859) Homepage Journal

        Apple was losing MAJOR, MAJOR market share. I said, you have to make the IPhone screen bigger. Because it was too small for a Grown Man's hands. They listened. Now it's MAGNIFICENT. And very successful. Trillion Dollar Company -- because of me!!!!

    • (Score: 2, Interesting) by Anonymous Coward on Wednesday November 07 2018, @03:56AM

      by Anonymous Coward on Wednesday November 07 2018, @03:56AM (#758822)

      Psst. Hint... Read the article CAREFULLY and understand what it is trying to say.

    • (Score: 2) by driverless on Wednesday November 07 2018, @06:15AM

      by driverless (4770) on Wednesday November 07 2018, @06:15AM (#758853)

      Bah, they should just all use IPv4. If it was good enough for Dennis, Brian, Ken, and Rich (W.Richard), it should be good enough for let-me-take-a-selfie phone user.

    • (Score: 2) by insanumingenium on Wednesday November 07 2018, @05:03PM (2 children)

      by insanumingenium (4824) on Wednesday November 07 2018, @05:03PM (#759048) Journal

      The Internet of Things is a buzzword, but I have never seen an IoT device that needed a public IPv4 IP. And I have never seen a user who signed up for an internet connection just to get their IoT device online. The crunch is not effected in the slightest by IoT marketing wank.

      There isn't a single magic moment for everyone, we are in the middle of a decades long journey. Mobile IPv6 support is ubiquitous, desktop support is ubiquitous, server (application) support is still a work in progress on the whole. Residential deployment of IPv6 is widespread in my area, almost no-one who is using it even realizes it. The desktops and phones are connected to home routers which (when less than ~5 years old) on the whole tend to support IPv6 and everything "just works". Every year more and more IPv6 is getting lit, and for the most part no-one notices, and that is exactly how it should be. The big resistance I see is from the business side of the house, but they will catch up in time, first we need to train a generation of engineers on IPv6.

      What is broken on the whole is IPv6 DNS. Not the servers, the client config. So the end result is that all the real world IPv6 usage I have ever seen resolved DNS over IPv4. DHCPv6 is one solution for this problem, perhaps not my first pick, but a solution none the less.

      • (Score: 1) by Mike on Thursday November 08 2018, @01:48AM (1 child)

        by Mike (823) on Thursday November 08 2018, @01:48AM (#759258)

        I'm curious what is missing on the client config? My understanding is that the ipv6 autoconfig can supply a DNS server (i.e. at a ipv6 address).

        • (Score: 2) by insanumingenium on Thursday November 08 2018, @05:26PM

          by insanumingenium (4824) on Thursday November 08 2018, @05:26PM (#759433) Journal

          The IETF went through the trouble of demanding strict separation along layers and then decided to adopt an RFC for RA-DNS after the fact. I'd rather adopt SSDP or equivalent to advertise arbitrary config options instead, no need to add a RFC for every new config parameter, while still getting away from our layer bluring DHCP all together in stateless deployments. Using a generalized service discovery protocol to discover DNS/NTP/PXE/SIP/Whatever either all in one place, or directly from the relevant servers.

          ND-RDNSS or RFC 6106/8106 support is growing, admittedly faster that I realized according to the usual unreliable sources [wikipedia.org]. Big win that Win10 finally got it in April. But I still haven't seen it used in the wild, likely my experience is stale. IPv6 prefix delegation got pushed out to the hoi polloi while I had my back turned, mayhaps NS-RDNSS did as well.

          For better or worse DHCPv4 became a vehicle for arbitrary configuration strings. Which makes it likely that if only via inertia DHCPv6 is still going to make more sense for the distribution of higher level config parameters if you have any DHCPv4 options other than DNS/Gateway.

    • (Score: 2) by hendrikboom on Wednesday November 07 2018, @10:06PM

      by hendrikboom (1125) Subscriber Badge on Wednesday November 07 2018, @10:06PM (#759172) Homepage Journal

      Mass adoption of IPv6 will happen only when the 128-bit address space is far too small to be usable.

  • (Score: 2) by Rosco P. Coltrane on Wednesday November 07 2018, @05:58AM (2 children)

    by Rosco P. Coltrane (4757) on Wednesday November 07 2018, @05:58AM (#758850)

    It makes them much easier to track.

    • (Score: 5, Insightful) by coolgopher on Wednesday November 07 2018, @07:18AM

      by coolgopher (1157) on Wednesday November 07 2018, @07:18AM (#758865)

      Then they'd be better off on IPv6, where you don't end up with infernal aliasing of internal addresses all the !&#^ time due to NAT.

    • (Score: 5, Interesting) by Anonymous Coward on Wednesday November 07 2018, @07:28AM

      by Anonymous Coward on Wednesday November 07 2018, @07:28AM (#758868)

      My understanding of it is that Google knew how ephemeral IP numbers would be as Android mobile devices transit across multiple networks, during the course of a day/week/month they knew they'll change them multiple times and that they'd probably be NAT'ed (and my mobiles always seem to get IPv4 addresses in the 10.x.x.x range from the phone company, so we're looking at CGNAT there).

      So, to track you, they use the play store and associated software garbage to track your (ab)usage by your google account, no google account? no problem...it's amazing how 'chatty' the average Android phone is out of the box (especially some of the Chinese ones)....why, hello there Mr IMSI, what are *you* doing lurking there in that datastream? (fuck knows what else is lurking in the encrypted ones...)

      Put a firewall on your phone, set it to block all traffic by default, watch what happens in the logs....(though to be more thorough here, you also need to monitor the phones connection via an external device to catch 'OOB' stuff, rule 1: never trust *any* software running on *any* phone)

      Fixed IP?, they don't need *no* steenkin' IP numbers (v4 or v6, fixed or otherwise) to track you, they're well beyond that.

      I'll just rearrange my tinfoil hat...

  • (Score: 3, Insightful) by aim on Wednesday November 07 2018, @09:59AM (4 children)

    by aim (6322) on Wednesday November 07 2018, @09:59AM (#758892)

    So, why use DHCPv6 rather than IPv6 autoconfig and anycast/multicast, other than admin's familiarity with DHCP in IPv4 networks?

    Way back when I learned about IPv6, DHCPv6 wasn't done yet, and as I haven't been deeply involved with IPv6 networking since, I honestly want to know.

    • (Score: 5, Informative) by ledow on Wednesday November 07 2018, @10:53AM

      by ledow (5567) on Wednesday November 07 2018, @10:53AM (#758902) Homepage

      Leases and reserved addresses.

      It's not a question of how a machine gets an address, that's easy. It's a question of knowing what machine got what address and how you talk to that machine and what if you want that machine to have a very particular address without having to make it use a (manually entered) static IP.

      For instance, if you want a machine to have a certain IP because it is the end-point of a port-forward, but you also know that that machine might in its life change IP randomly, then you have a problem. If you have a dozen such machines managing static IPs for them is a chore even on a small network of IPv4. With a large network of IPv6, it's a nightmare. So you want to say "this MAC address only ever gets this IP address", which stops the fluxing, doesn't need you to enter ANYTHING into the machine itself (which might be on the other side of the room or the other side of the planet), and you know that nobody will be able to stamp over it with an autoconfigured address, or another DHCP allocation.

    • (Score: 2) by coolgopher on Wednesday November 07 2018, @10:56AM

      by coolgopher (1157) on Wednesday November 07 2018, @10:56AM (#758903)

      Typically to push out DNS and/or NTP settings, and updating internal resolvers (if mDNS is not in use).

      These days there's also RFC8106, where DNS settings can be carried in the Route Advertisements, further reducing the need for DHCPv6.

    • (Score: 2, Interesting) by Anonymous Coward on Wednesday November 07 2018, @11:16AM (1 child)

      by Anonymous Coward on Wednesday November 07 2018, @11:16AM (#758906)

      To allow fixed addresses according to a network's acceptable range. Better auditing and configuration of systems to use the network. And better anonymity of incoming/outgoing packets since they aren't based on a real or partially obfuscated hardware address.

      If we wanted hardware addresses in our network ids we would still be using IPX after all.

(1)