• Re: RPi associating two IPs with its one and only wifi interface

    From Richard Kettlewell@3:633/10 to All on Sat Dec 27 20:46:18 2025
    Jim Diamond <zsd@jdvb.ca> writes:
    I was looking at my network and discovered an IP which I didn't know
    about; after a few seconds of investigation I discovered that one of
    my Pis (which is on wifi only, and only has one wifi card) has two
    IPs.

    Where are you finding the two addresses?

    What does
    ip addr show
    display?

    Two of my other Pis are running the same version of Raspberry Pi OS
    (i.e., "Debian GNU/Linux 12 (bookworm)"). They don't do this.

    Looking around the net, there are claims that this is because Pis
    might try to netboot, and that later on in the boot process they also
    get their usual IP the "usual" way. (In my case I am using
    networkmanager.)

    What does
    rpi-eeprom-config
    display?

    https://www.raspberrypi.com/documentation/computers/raspberry-pi.html#BOOT_ORDER
    describes how the boot order is configured.

    I can't imagine what I did to make one of my Pis want to (try to)
    netboot.

    The web suggests some have been shipped with netbook enabled.
    --
    https://www.greenend.org.uk/rjk/

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From John R Walliker@3:633/10 to All on Sat Dec 27 19:03:28 2025
    On 27/12/2025 18:17, The Natural Philosopher wrote:
    On 27/12/2025 17:34, Jim Diamond wrote:
    Has anyone here seen this, and, if so, know what grievous sins I have
    committed to make this happen?ÿ And how to make it stop?

    For years my main *86 server would register itself with DHCP first, then manually set its IP address to something else.

    In the end I plugged in a screen and used the GUI and net manager to
    sort it all out, and it stayed sorted out.

    The problem is that there have been so many ways to set up IP over the
    last few years that any online guidance has at best a 25% chance of
    doing it the way the distro thinks it ought to be done,

    Did the Pi itself think that it had two addresses on the WiFi interface
    or was it your router / dhcp server that had two records?


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Jim Diamond@3:633/10 to All on Sun Dec 28 12:00:00 2025
    I was looking at my network and discovered an IP which I didn't know about; after a few seconds of investigation I discovered that one of my Pis (which
    is on wifi only, and only has one wifi card) has two IPs.

    Two of my other Pis are running the same version of Raspberry Pi OS (i.e., "Debian GNU/Linux 12 (bookworm)"). They don't do this.

    Looking around the net, there are claims that this is because Pis might try
    to netboot, and that later on in the boot process they also get their usual
    IP the "usual" way. (In my case I am using networkmanager.)

    I can't imagine what I did to make one of my Pis want to (try to) netboot.

    Has anyone here seen this, and, if so, know what grievous sins I have
    committed to make this happen? And how to make it stop?

    Thanks.
    Jim

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Sat Dec 27 19:28:30 2025
    On 27/12/2025 19:03, John R Walliker wrote:
    On 27/12/2025 18:17, The Natural Philosopher wrote:
    On 27/12/2025 17:34, Jim Diamond wrote:
    Has anyone here seen this, and, if so, know what grievous sins I have
    committed to make this happen?ÿ And how to make it stop?

    For years my main *86 server would register itself with DHCP first,
    then manually set its IP address to something else.

    In the end I plugged in a screen and used the GUI and net manager to
    sort it all out, and it stayed sorted out.

    The problem is that there have been so many ways to set up IP over the
    last few years that any online guidance has at best a 25% chance of
    doing it the way the distro thinks it ought to be done,

    Did the Pi itself think that it had two addresses on the WiFi interface
    or was it your router / dhcp server that had two records?

    [It isnt a pi: it is an *x86 mobo running linux mint, with ethernet connectivity]

    I am nor sure. I think I could at least ping the DHCP assigned interface.

    The router only has DHCP records - statics are not recorded

    But its all gone these days...

    My point being this is not Pi or wifi specific, it is an issue with
    network configuration



    --
    ?There are two ways to be fooled. One is to believe what isn?t true; the
    other is to refuse to believe what is true.?

    ?Soren Kierkegaard


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Sat Dec 27 18:17:02 2025
    On 27/12/2025 17:34, Jim Diamond wrote:
    Has anyone here seen this, and, if so, know what grievous sins I have committed to make this happen? And how to make it stop?

    For years my main *86 server would register itself with DHCP first, then manually set its IP address to something else.

    In the end I plugged in a screen and used the GUI and net manager to
    sort it all out, and it stayed sorted out.

    The problem is that there have been so many ways to set up IP over the
    last few years that any online guidance has at best a 25% chance of
    doing it the way the distro thinks it ought to be done,



    --
    "When one man dies it's a tragedy. When thousands die it's statistics."

    Josef Stalin



    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lawrence D?Oliveiro@3:633/10 to All on Mon Dec 29 15:30:00 2025
    On Sun, 28 Dec 2025 23:14:09 -0400, Jim Diamond wrote:

    No, the Pi thought two IPs were associated with the one and only
    wifi card.

    Just a note that this itself is not an error; the Pi doesn?t *think*
    the interface has two IPs, it *knows*, because it was configured that
    way.

    There are legitimate uses for this sort of thing, even though it
    was not a deliberate setup in your case.

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Jim Diamond@3:633/10 to All on Mon Dec 29 15:30:00 2025
    On 2025-12-27 at 15:03 AST, John R Walliker <jrwalliker@gmail.com> wrote:
    On 27/12/2025 18:17, The Natural Philosopher wrote:
    On 27/12/2025 17:34, Jim Diamond wrote:
    Has anyone here seen this, and, if so, know what grievous sins I have
    committed to make this happen?ÿ And how to make it stop?

    For years my main *86 server would register itself with DHCP first, then
    manually set its IP address to something else.

    In the end I plugged in a screen and used the GUI and net manager to
    sort it all out, and it stayed sorted out.

    The problem is that there have been so many ways to set up IP over the
    last few years that any online guidance has at best a 25% chance of
    doing it the way the distro thinks it ought to be done,

    Did the Pi itself think that it had two addresses on the WiFi interface
    or was it your router / dhcp server that had two records?

    Just in case that question was directed at me...

    No, the Pi thought two IPs were associated with the one and only wifi card.
    (In fact, I discovered which device it was by ssh-ing in after nmap told me that it was listening on the sshd port.)

    Jim

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Jim Diamond@3:633/10 to All on Tue Dec 30 15:30:02 2025
    On 2025-12-29 at 19:48 AST, Pancho <Pancho.Jones@protonmail.com> wrote:
    On 12/27/25 23:25, Theo wrote:
    Jim Diamond <zsd@jdvb.ca> wrote:
    I was looking at my network and discovered an IP which I didn't know about; >>> after a few seconds of investigation I discovered that one of my Pis (which >>> is on wifi only, and only has one wifi card) has two IPs.

    Two of my other Pis are running the same version of Raspberry Pi OS (i.e., >>> "Debian GNU/Linux 12 (bookworm)"). They don't do this.

    Looking around the net, there are claims that this is because Pis might try >>> to netboot, and that later on in the boot process they also get their usual >>> IP the "usual" way. (In my case I am using networkmanager.)

    I can't imagine what I did to make one of my Pis want to (try to) netboot. >>>
    Has anyone here seen this, and, if so, know what grievous sins I have
    committed to make this happen? And how to make it stop?

    Is this two IPv4 addresses? Having multiple IPv6 addresses is completely
    routine. As is having one IPv4 and one or more IPv6s.


    Yes, I have recently been experimenting with IPv6.

    IPv6 uses: Static Addresses, Link-Local, DHCPv6, Router Advertisements,
    and SLAAC. Proving me with multiple, often random, IPv6 addresses. This totally broke my IP rule based routing.

    Unlike my Linux machines, My Android devices take a whole IPv6 block
    (prefix delegation). I think I'll change all my Linux hosts to this too.
    I quite like the protection of random addresses, but want them
    constrained to a recognisable range for each host.

    Some websites fail, some apt repos fail under IPv6.

    IPv6 seems like a world of pain.

    Did someone say "new and improved" ? ;-)

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Richard Kettlewell@3:633/10 to All on Mon Dec 29 10:59:28 2025
    Jim Diamond <zsd@jdvb.ca> writes:
    Richard Kettlewell <invalid@invalid.invalid> wrote:
    Jim Diamond <zsd@jdvb.ca> writes:
    I was looking at my network and discovered an IP which I didn't know
    about; after a few seconds of investigation I discovered that one of
    my Pis (which is on wifi only, and only has one wifi card) has two
    IPs.

    Where are you finding the two addresses?

    What does
    ip addr show
    display?

    Two of my other Pis are running the same version of Raspberry Pi OS
    (i.e., "Debian GNU/Linux 12 (bookworm)"). They don't do this.

    Looking around the net, there are claims that this is because Pis
    might try to netboot, and that later on in the boot process they also
    get their usual IP the "usual" way. (In my case I am using
    networkmanager.)

    What does
    rpi-eeprom-config
    display?

    BOOT_UART=0
    WAKE_ON_GPIO=1
    POWER_OFF_ON_HALT=0

    What model of Pi is it?

    (Nothing there about BOOT_ORDER, altho my Pi5 (not the computer that showed up with two IPs) does have BOOT_ORDER=0xf461. But it would appear I'd need
    a 0x2 in there fairly early for it to try a netboot.)

    Elsewhere:

    No, the Pi thought two IPs were associated with the one and only wifi
    card. (In fact, I discovered which device it was by ssh-ing in after
    nmap told me that it was listening on the sshd port.)

    i.e. the interface has two addresses after booting has completed. I
    would put network boot at the bottom of the list of possible causes.
    AIUI the kernel starts from a blank slate, it doesn?t inherit interface configuration from the boot loader.

    --
    https://www.greenend.org.uk/rjk/

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Mon Dec 29 17:31:04 2025
    On 29/12/2025 15:08, Jim Diamond wrote:

    It was in fact two IPv4 addresses. One which the Network Manager config
    file specifies, and the other was some other address which I speculate
    came from the router's DHCP server.
    You can check the routers DHCP table and the MAC addresses associated
    with it.

    As I did.


    Jim

    --
    ?People believe certain stories because everyone important tells them,
    and people tell those stories because everyone important believes them. Indeed, when a conventional wisdom is at its fullest strength, one?s
    agreement with that conventional wisdom becomes almost a litmus test of
    one?s suitability to be taken seriously.?

    Paul Krugman


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lars Poulsen@3:633/10 to All on Tue Dec 30 15:30:02 2025
    On 2025-12-30, Jim Diamond <zsd@jdvb.ca> wrote:
    My router's DHCP function has a serious case of brain damage. (It is supplied by my ISP and is not easy to avoid using, because of the way they control their fiber optic cable based system.) Right now it thinks the Pi
    in question is using the "mystery" IPv4 address, and it resolutely declares that the "proper" address is not connected. The Pi itself disagrees.

    My current ISP is Frontier Fiber, and I have a business account with
    a static IP address. While they supply a fiber termination unit in
    my house wiring closet, they also supply a Sagemcom router/WiFi
    device, which I *must* use because my static IP is the creation of a VPN feature inside the router, for which they do not provide any
    usable documentation. So I had them bridge that IP to one of the
    ethernet ports on the Sagemcom device, and then have my own edge
    router attached to that. I can't turn off the WiFi on the Sagemcom, but
    I have my own WiFi access point behind my router.

    I am not very happy with the arrangement, but my bandwidth is 500 Mbps symmetric, although my (old) edge router is only 10/100 ethernet.
    But after spending years on a 15Mbps symmetric network, I am happy to
    get 100/100 that WORKS.

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Theo@3:633/10 to All on Sat Dec 27 23:25:44 2025
    Jim Diamond <zsd@jdvb.ca> wrote:
    I was looking at my network and discovered an IP which I didn't know about; after a few seconds of investigation I discovered that one of my Pis (which is on wifi only, and only has one wifi card) has two IPs.

    Two of my other Pis are running the same version of Raspberry Pi OS (i.e., "Debian GNU/Linux 12 (bookworm)"). They don't do this.

    Looking around the net, there are claims that this is because Pis might try to netboot, and that later on in the boot process they also get their usual IP the "usual" way. (In my case I am using networkmanager.)

    I can't imagine what I did to make one of my Pis want to (try to) netboot.

    Has anyone here seen this, and, if so, know what grievous sins I have committed to make this happen? And how to make it stop?

    Is this two IPv4 addresses? Having multiple IPv6 addresses is completely routine. As is having one IPv4 and one or more IPv6s.

    Theo

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Pancho@3:633/10 to All on Mon Dec 29 23:48:28 2025
    On 12/27/25 23:25, Theo wrote:
    Jim Diamond <zsd@jdvb.ca> wrote:
    I was looking at my network and discovered an IP which I didn't know about; >> after a few seconds of investigation I discovered that one of my Pis (which >> is on wifi only, and only has one wifi card) has two IPs.

    Two of my other Pis are running the same version of Raspberry Pi OS (i.e., >> "Debian GNU/Linux 12 (bookworm)"). They don't do this.

    Looking around the net, there are claims that this is because Pis might try >> to netboot, and that later on in the boot process they also get their usual >> IP the "usual" way. (In my case I am using networkmanager.)

    I can't imagine what I did to make one of my Pis want to (try to) netboot. >>
    Has anyone here seen this, and, if so, know what grievous sins I have
    committed to make this happen? And how to make it stop?

    Is this two IPv4 addresses? Having multiple IPv6 addresses is completely routine. As is having one IPv4 and one or more IPv6s.


    Yes, I have recently been experimenting with IPv6.

    IPv6 uses: Static Addresses, Link-Local, DHCPv6, Router Advertisements,
    and SLAAC. Proving me with multiple, often random, IPv6 addresses. This totally broke my IP rule based routing.

    Unlike my Linux machines, My Android devices take a whole IPv6 block
    (prefix delegation). I think I'll change all my Linux hosts to this too.
    I quite like the protection of random addresses, but want them
    constrained to a recognisable range for each host.

    Some websites fail, some apt repos fail under IPv6.

    IPv6 seems like a world of pain.


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Mon Dec 29 13:08:24 2025
    On 29/12/2025 10:59, Richard Kettlewell wrote:
    Jim Diamond <zsd@jdvb.ca> writes:
    Richard Kettlewell <invalid@invalid.invalid> wrote:
    Jim Diamond <zsd@jdvb.ca> writes:
    I was looking at my network and discovered an IP which I didn't know
    about; after a few seconds of investigation I discovered that one of
    my Pis (which is on wifi only, and only has one wifi card) has two
    IPs.

    Where are you finding the two addresses?

    What does
    ip addr show
    display?

    Two of my other Pis are running the same version of Raspberry Pi OS
    (i.e., "Debian GNU/Linux 12 (bookworm)"). They don't do this.

    Looking around the net, there are claims that this is because Pis
    might try to netboot, and that later on in the boot process they also
    get their usual IP the "usual" way. (In my case I am using
    networkmanager.)

    What does
    rpi-eeprom-config
    display?

    BOOT_UART=0
    WAKE_ON_GPIO=1
    POWER_OFF_ON_HALT=0

    What model of Pi is it?

    (Nothing there about BOOT_ORDER, altho my Pi5 (not the computer that showed >> up with two IPs) does have BOOT_ORDER=0xf461. But it would appear I'd need >> a 0x2 in there fairly early for it to try a netboot.)

    Elsewhere:

    No, the Pi thought two IPs were associated with the one and only wifi
    card. (In fact, I discovered which device it was by ssh-ing in after
    nmap told me that it was listening on the sshd port.)

    i.e. the interface has two addresses after booting has completed. I
    would put network boot at the bottom of the list of possible causes.
    AIUI the kernel starts from a blank slate, it doesn?t inherit interface configuration from the boot loader.

    IIRC in my setup I had static setup via a hand edited file somewhere,
    and it was booting into DHCP before it read that file

    --
    For in reason, all government without the consent of the governed is the
    very definition of slavery.

    Jonathan Swift



    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lawrence D?Oliveiro@3:633/10 to All on Tue Dec 30 17:30:02 2025
    On Mon, 29 Dec 2025 10:59:29 +0000, Richard Kettlewell wrote:

    i.e. the interface has two addresses after booting has completed. I
    would put network boot at the bottom of the list of possible causes.
    AIUI the kernel starts from a blank slate, it doesn?t inherit
    interface configuration from the boot loader.

    If my reading of this <https://en.wikipedia.org/wiki/Preboot_Execution_Environment> article
    on PXE is correct, the kernel can indeed inherit network settings from
    a PXE boot. But if you?re not using PXE, then that?s not relevant, of
    course. I don?t know of any other booting setup that could pass across
    network parameters from the bootloader to the kernel.

    Places I know of where the kernel can get network interface setups:
    * DHCP client
    * /etc/network/interfaces file (or distro-specific equivalent)
    * systemd.network files
    * NetworkManager

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Tue Dec 30 11:03:28 2025
    On 30/12/2025 01:00, Jim Diamond wrote:
    On 2025-12-29 at 13:31 AST, The Natural Philosopher <tnp@invalid.invalid> wrote:
    On 29/12/2025 15:08, Jim Diamond wrote:

    It was in fact two IPv4 addresses. One which the Network Manager config >>> file specifies, and the other was some other address which I speculate
    came from the router's DHCP server.

    You can check the routers DHCP table and the MAC addresses associated
    with it.

    My router's DHCP function has a serious case of brain damage. (It is supplied by my ISP and is not easy to avoid using, because of the way they control their fiber optic cable based system.) Right now it thinks the Pi
    in question is using the "mystery" IPv4 address, and it resolutely declares that the "proper" address is not connected. The Pi itself disagrees.

    Routers do not retain memory of statically assigned addresses. You might
    find it in the arp table


    However, it was worth a look. Maybe. According to the router, the
    "mystery" address is paired with the wifi card's actual ethernet (MAC) address, whereas the "proper" address is (currently) paired with the same ethernet address, except the last octet is 8C instead of 8D. This makes me think that it is showing "Connected" or "Disconnected" according to the ethernet address which is working, and it is not careful about pairing that with the correct IPv4 address.

    I thought this was a wifi issue. There is no 'Ethernet' address.

    Jim

    --
    Socialism is the philosophy of failure, the creed of ignorance and the
    gospel of envy.

    Its inherent virtue is the equal sharing of misery.

    Winston Churchill



    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Tue Dec 30 11:01:08 2025
    On 30/12/2025 00:40, Jim Diamond wrote:
    On 2025-12-28 at 23:51 AST, Lawrence D?Oliveiro <ldo@nz.invalid> wrote:
    On Sun, 28 Dec 2025 23:14:09 -0400, Jim Diamond wrote:

    No, the Pi thought two IPs were associated with the one and only
    wifi card.

    Just a note that this itself is not an error; the Pi doesn?t *think*
    the interface has two IPs, it *knows*, because it was configured that
    way.

    OK... If we want to be really careful with our English, I suppose we should say "The Pi associated two IPv4 addresses with the one and only WiFi card."

    Well that depends on what constitutes the 'Pi'' Its software did the association, not the Pi...

    Not running any AI on my Pi, I don't think it knows or thinks anything. ;-)

    You never know, with Heffalumps

    There are legitimate uses for this sort of thing, even though it
    was not a deliberate setup in your case.

    Quite true, I have used it once or twice in the past myself. (Why I did it is now lost in the depths of time.)

    Handy way to talk to an unconfigured router on some random default address.

    Jim

    --
    There is nothing a fleet of dispatchable nuclear power plants cannot do
    that cannot be done worse and more expensively and with higher carbon emissions and more adverse environmental impact by adding intermittent renewable energy.


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Jim Diamond@3:633/10 to All on Wed Dec 31 10:00:00 2025
    On 2025-12-30 at 01:42 AST, Lawrence D?Oliveiro <ldo@nz.invalid> wrote:
    On Mon, 29 Dec 2025 21:00:19 -0400, Jim Diamond wrote:

    My router's DHCP function has a serious case of brain damage. (It is
    supplied by my ISP and is not easy to avoid using, because of the
    way they control their fiber optic cable based system.)

    I wonder why you would use an ISP-supplied router -- don?t you get to
    connect your own?

    No. The fiber comes into a box and the box also has the wifi router.
    (I attached a second wifi router to give me better coverage, but that one
    just operates in some "pass through" mode (if that is the correct term).)

    According to things I've read on the internet, some people have bought
    their own fiber hardware and then managed to do some end-run around the security (and so on) that the phone company puts in place. But I decided
    that I wasn't going down that road.

    That?s what I have done, here in ??. I have turned off its DHCP
    function, and run my own DHCP server on a separate Linux box, so that I
    can more easily keep track of lists of statically-assigned addresses for
    my trusted machines.

    I could try turning off the DHCP server in the ISP's box. However, they
    have a habit of resetting and updating the software in their box, and I'm
    not sure how long turning off that DHCP server would last. I am not sure
    what would happen when two DHCP servers are on the same LAN, but I imagine
    Bad Things would happen.

    Jim

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From David Higton@3:633/10 to All on Tue Dec 30 20:27:40 2025
    In message <10j1bob$223dq$1@dont-email.me>
    Lawrence D?Oliveiro <ldo@nz.invalid> wrote:

    On Tue, 30 Dec 2025 20:00:52 GMT, David Higton wrote:

    What I particularly like about IPv6 is that NAT/NAPT are simply not necessary, and it's possible to have multiple servers on the same port (e.g. multiple web servers on port 80/443) on one site, because you have effectively unlimited internet-accessible addresses.

    What happens if/when you switch provider? Are you allowed to take your IPv6 address block with you?

    No.

    Because if not, you end up having to assign new addresses to everything on your LAN.

    That's what DHCPv6 is for.

    In terms of external DNS, most routers support dynamic DNS with several protocols; but it's normally a very simple operation to update your
    AAAA records manually. Or even by a shell script, like I run on my
    Ubuntu and RISC OS boxes, to automate it. Most domestic users have
    very few internet-facing IP addresses and therefore very few DNS entries
    to update, so the task doesn't comsume much time.

    David

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From David Higton@3:633/10 to All on Tue Dec 30 20:00:52 2025
    In message <10iv40e$1e1ba$1@dont-email.me>
    Pancho <Pancho.Jones@protonmail.com> wrote:

    IPv6 seems like a world of pain.

    In my experience it just works.

    There was one (very recent) exception, where a website had been moved to
    a new provider and IPv6 was now available. They copied across their
    config files from the old site. What they forgot was that the old
    config files didn't enable IPv6. A simple fault that was quickly fixed,
    and was not a fault of IPv6 in itself. Since then the site has just
    worked over IPv6 and IPv4.

    I moved my own website to a new provider a few weeks ago, specifically
    because the old provider didn't do IPv6 on their low cost shared servers (despite saying on their own website that IPv6 was supported everywhere!).
    The new website just worked from the get-go over IPv6 and IPv4.

    The analysis I've seen of the world's internet traffic indicates that
    over 50% is now IPv6, and rising.

    What I particularly like about IPv6 is that NAT/NAPT are simply not
    necessary, and it's possible to have multiple servers on the same port
    (e.g. multiple web servers on port 80/443) on one site, because you
    have effectively unlimited internet-accessible addresses.

    David

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lawrence D?Oliveiro@3:633/10 to All on Wed Dec 31 14:00:02 2025
    On Tue, 30 Dec 2025 17:50:15 -0400, Jim Diamond wrote:

    On 2025-12-30 at 01:42 AST, Lawrence D?Oliveiro <ldo@nz.invalid> wrote:

    I wonder why you would use an ISP-supplied router -- don?t you get
    to connect your own?

    No. The fiber comes into a box and the box also has the wifi router.

    Here in NZ, the fibre terminates in a separate box called the ?ONT?.
    Everything up to and including that box is the responsibility of a
    company called ?Tuatahi Fibre?, which is *not* an ISP (and is not
    allowed to become one). Basically they provide a layer-2 connection
    between my house and the routers/switches/whatever for the service
    providers, and their job is to treat all service providers equally.

    Also, that box is to be considered part of the fittings of the house,
    so it stays with the house, and doesn?t move if/when the residents
    move house. The installation of the box (and the physical fibre
    connection to it) was done free of charge, under a Government-funded
    plan.

    The particular ONT box in my house has 4 Ethernet ports and 2
    telephone landline ports. These can be independently assigned to
    different services, coming from different providers. E.g. the one
    that?s live for my ISP connection has an Ethernet cable running
    between it and my actual router, which I bought from a local store.

    I could try turning off the DHCP server in the ISP's box. However, they
    have a habit of resetting and updating the software in their box, and I'm
    not sure how long turning off that DHCP server would last. I am not sure what would happen when two DHCP servers are on the same LAN, but I imagine Bad Things would happen.

    Hmmm ... I?m thinking it might be possible to isolate that box on a
    dedicated Ethernet port on a Linux box, so you could use a packet
    filter to block anything DHCP-related, and only let through the stuff
    you want ...

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Richard Kettlewell@3:633/10 to All on Wed Dec 31 20:18:52 2025
    Pancho <Pancho.Jones@protonmail.com> writes:
    The Natural Philosopher wrote:
    David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not
    necessary
    So making the implementation of a firewall absolutely mandatory


    Linux IPv6 does appear to use random IPv6 address for outbound
    connections, which have a limited lifespan. This appears to be
    something like 1-7 days, but if very short lifespans were used it
    could offer a protection similar to NAT. I need to investigate a bit
    further, but I don't think IPv6 needs to be inherently less safe.

    NAT does not offer any protection. The reason that a typical domestic NAT-equipped router protects you from inbound connections is that it has
    a firewall as well.

    (Getting a packet addressed to your internal addresses to your external interface is inconvenient for many attackers, for sure, but
    striaghtforward for your ISP or anyone who can hack or coerce them.)

    --
    https://www.greenend.org.uk/rjk/

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Jim Diamond@3:633/10 to All on Thu Jan 1 22:00:00 2026
    On 2025-12-31 at 07:58 AST, The Natural Philosopher <tnp@invalid.invalid> wrote:
    On 30/12/2025 21:07, Jim Diamond wrote:
    All good thoughts, thanks. But...
    - There is no dhcp client running (at least by the time I am able to ssh in >> to the machine after it boots).
    - All /etc/network/interfaces does is source any files in interfaces.d,
    and there are no files there.
    - "sudo locate systemd.network" only shows the man page.
    - the "mystery" IPv4 address doesn't show up anywhere in the
    /etc/NetworkManager directory or sub-directories.

    I guess the mystery continues.

    That was where I got to with my one.

    At some stage that mobo died and I took the opportunity to switch mobos
    and install an updated linux version, using a GUI and network manager to
    set up the fixed IP, and the problem vanished.

    If you can do a fresh install its probably the shortest route.

    That will be a consideration, should push come to shove. So far, this
    mystery IP hasn't caused any problems, but it is anomalous, which is bothersome.

    Now even if its headless there is a CLI to network manager and you might investigate that.

    It's called in a fit of stunning originality, 'nmcli'

    Try
    #nmcli device show

    Yes, I know about nmcli and even use it occasionally. Thanks.

    Also ifconfig -a should show up any active interfaces on odd addresses.

    BUT IIRC I never could identify that interface that way - it seemed to
    be some sort of low level zombie.

    It existed in the router DHCP table, showing it had been issues by the router in response to a request from the machine, but it only ever
    responded to pings, IIRC.

    No listening process beyond that was ever bound to it.

    I assumed it was some bug either induced by me hand editing files that network manager was supposed to edit, or as a changeover from earlier methods of setting up IP, not fully ignored by the new NM control system

    All I know is that rigorous adherence to the GUI CLI on a fresh install eliminated it. Whether it was one or the other factor that was crucial,
    I cannot say.
    As with most transient bugs, life is to frikkin short...

    I agree 100% with that.

    I am sorry I cannot help beyond noting that yes, I have seen it happen,
    and no, I cant reproduce it any more, and at a given point it vanished, never to reappear...

    Well, I haven't seen my mystery addr recently. Maybe a neutrino hit
    the wrong spot during boot.

    Jim

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Wed Dec 31 11:36:32 2025
    On 30/12/2025 20:00, David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not
    necessary

    So making the implementation of a firewall absolutely mandatory

    --
    Ideas are more powerful than guns. We would not let our enemies have
    guns, why should we let them have ideas?

    Josef Stalin


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Wed Dec 31 21:54:20 2025
    On 31/12/2025 20:18, Richard Kettlewell wrote:
    Pancho <Pancho.Jones@protonmail.com> writes:
    The Natural Philosopher wrote:
    David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not
    necessary
    So making the implementation of a firewall absolutely mandatory


    Linux IPv6 does appear to use random IPv6 address for outbound
    connections, which have a limited lifespan. This appears to be
    something like 1-7 days, but if very short lifespans were used it
    could offer a protection similar to NAT. I need to investigate a bit
    further, but I don't think IPv6 needs to be inherently less safe.

    NAT does not offer any protection. The reason that a typical domestic NAT-equipped router protects you from inbound connections is that it has
    a firewall as well.

    (Getting a packet addressed to your internal addresses to your external interface is inconvenient for many attackers, for sure, but
    striaghtforward for your ISP or anyone who can hack or coerce them.)

    How?
    Genuine question.



    --
    A lie can travel halfway around the world while the truth is putting on
    its shoes.


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Pancho@3:633/10 to All on Wed Dec 31 19:26:48 2025
    On 12/30/25 20:00, David Higton wrote:
    In message <10iv40e$1e1ba$1@dont-email.me>
    Pancho <Pancho.Jones@protonmail.com> wrote:

    IPv6 seems like a world of pain.

    In my experience it just works.


    I'm surprised. Accepting that you do not do some of the things I do,
    like policy routing rules based upon a host computer IP, I'm still
    seeing servers on the internet that advertise they should work with IPv6
    but don't. This means they don't fall back to IPv4.

    I'm not far enough along in my understanding to be entirely confident,
    but I would be surprised if I were wrong.


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Richard Kettlewell@3:633/10 to All on Wed Dec 31 13:15:26 2025
    druck <news@druck.org.uk> writes:
    On 30/12/2025 01:00, Jim Diamond wrote:
    However, it was worth a look. Maybe. According to the router, the
    "mystery" address is paired with the wifi card's actual ethernet (MAC)

    MAC's aren't just Ethernet, WiFi and Bluetooth interfaces also have them.

    address, whereas the "proper" address is (currently) paired with the
    same ethernet address, except the last octet is 8C instead of 8D.
    This makes me think that it is showing "Connected" or "Disconnected"
    according to the ethernet address which is working, and it is not
    careful about pairing that with the correct IPv4 address.

    You often see a difference of 1 when something creates a virtual
    network interface for use by a virtual machine or container. The
    virtual network card is assigned the second IP address and can operate independently from anything using the hosts primary interface and IP
    address.

    At least on my 3B and 4B, the wired and wireless interfaces have
    adjacent MACs.

    PS C:\Users\rjk> ssh shairo ip link show
    1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    2: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
    link/ether dc:a6:32:cb:73:6b brd ff:ff:ff:ff:ff:ff
    3: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP mode DORMANT group default qlen 1000
    link/ether dc:a6:32:cb:73:6c brd ff:ff:ff:ff:ff:ff

    If both interfaces were connected to the same network then I might see something similar to Jim?s situation.

    I did ask Jim for ?ip addr show? output but it has not appeared.

    --
    https://www.greenend.org.uk/rjk/

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Wed Dec 31 11:58:52 2025
    On 30/12/2025 21:07, Jim Diamond wrote:
    All good thoughts, thanks. But...
    - There is no dhcp client running (at least by the time I am able to ssh in
    to the machine after it boots).
    - All /etc/network/interfaces does is source any files in interfaces.d,
    and there are no files there.
    - "sudo locate systemd.network" only shows the man page.
    - the "mystery" IPv4 address doesn't show up anywhere in the
    /etc/NetworkManager directory or sub-directories.

    I guess the mystery continues.

    That was where I got to with my one.

    At some stage that mobo died and I took the opportunity to switch mobos
    and install an updated linux version, using a GUI and network manager to
    set up the fixed IP, and the problem vanished.

    If you can do a fresh install its probably the shortest route.

    Now even if its headless there is a CLI to network manager and you might investigate that.

    It's called in a fit of stunning originality, 'nmcli'

    Try
    #nmcli device show

    Also ifconfig -a should show up any active interfaces on odd addresses.

    BUT IIRC I never could identify that interface that way - it seemed to
    be some sort of low level zombie.

    It existed in the router DHCP table, showing it had been issues by the
    router in response to a request from the machine, but it only ever
    responded to pings, IIRC.

    No listening process beyond that was ever bound to it.

    I assumed it was some bug either induced by me hand editing files that
    network manager was supposed to edit, or as a changeover from earlier
    methods of setting up IP, not fully ignored by the new NM control system

    All I know is that rigorous adherence to the GUI CLI on a fresh install eliminated it. Whether it was one or the other factor that was crucial,
    I cannot say.
    As with most transient bugs, life is to frikkin short...

    I am sorry I cannot help beyond noting that yes, I have seen it happen,
    and no, I cant reproduce it any more, and at a given point it vanished,
    never to reappear...


    --
    Microsoft : the best reason to go to Linux that ever existed.


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From David Higton@3:633/10 to All on Wed Dec 31 20:23:18 2025
    In message <10j3tmk$29ec2$1@dont-email.me>
    Pancho <Pancho.Jones@protonmail.com> wrote:

    On 12/31/25 11:36, The Natural Philosopher wrote:
    On 30/12/2025 20:00, David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not necessary

    So making the implementation of a firewall absolutely mandatory


    Linux IPv6 does appear to use random IPv6 address for outbound
    connections, which have a limited lifespan. This appears to be something like 1-7 days, but if very short lifespans were used it could offer a protection similar to NAT. I need to investigate a bit further, but I
    don't think IPv6 needs to be inherently less safe.

    I use Ubuntu, which gives itself two routable IPv6 addresses. One is
    always the same (good if you have an internet-accessible server) and
    the other is new each reboot (good for privacy). Maybe it's the same
    for you?

    David

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Wed Dec 31 21:48:58 2025
    On 31/12/2025 19:31, Pancho wrote:
    On 12/31/25 11:36, The Natural Philosopher wrote:
    On 30/12/2025 20:00, David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not
    necessary

    So making the implementation of a firewall absolutely mandatory


    Linux IPv6 does appear to use random IPv6 address for outbound
    connections, which have a limited lifespan. This appears to be something like 1-7 days, but if very short lifespans were used it could offer a protection similar to NAT. I need to investigate a bit further, but I
    don't think IPv6 needs to be inherently less safe.

    No, its more that it imposes a different discipline, And that has to be learnt, taught and implemented.

    And that takes time, costs money and always has gotchas.

    Which is why most people (consumers and even businesses) are still
    running V4 + NAT....


    --
    ?A leader is best When people barely know he exists. Of a good leader,
    who talks little,When his work is done, his aim fulfilled,They will say,
    ?We did this ourselves.?

    ? Lao Tzu, Tao Te Ching


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Pancho@3:633/10 to All on Wed Dec 31 19:31:30 2025
    On 12/31/25 11:36, The Natural Philosopher wrote:
    On 30/12/2025 20:00, David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not
    necessary

    So making the implementation of a firewall absolutely mandatory


    Linux IPv6 does appear to use random IPv6 address for outbound
    connections, which have a limited lifespan. This appears to be something
    like 1-7 days, but if very short lifespans were used it could offer a protection similar to NAT. I need to investigate a bit further, but I
    don't think IPv6 needs to be inherently less safe.

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lawrence D?Oliveiro@3:633/10 to All on Thu Jan 1 22:00:00 2026
    On Wed, 31 Dec 2025 16:00:45 -0400, Jim Diamond wrote:

    rpi4-4b ls -l /etc/systemd/network
    total 0
    lrwxrwxrwx 1 root root 9 Dec 5 2023 73-usb-net-by-mac.link -> /dev/null lrwxrwxrwx 1 root root 9 Dec 5 2023 99-default.link -> /dev/null

    Interesting that you have anything in that directory at all; the one
    on my Debian Unstable system is empty.

    This symlinking to /dev/null is called ?masking?. It explicitly
    disables those service entries (originals present in
    /lib/systemd/network in this case) so they cannot be
    (easily/accidentally) enabled.

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From David Higton@3:633/10 to All on Wed Dec 31 11:44:38 2025
    In message <10j31s0$2gptg$9@dont-email.me>
    The Natural Philosopher <tnp@invalid.invalid> wrote:

    On 30/12/2025 20:00, David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not necessary

    So making the implementation of a firewall absolutely mandatory

    Of course, as possessed by all routers. The default state is to reject
    all incoming connections. If you want to expose a device to the
    internet, you punch a pinhole in the firewall, normally for one address
    and one port.

    David

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Wed Dec 31 21:52:10 2025
    On 31/12/2025 20:04, Jim Diamond wrote:
    Maybe a neutrino hit
    the wrong spot during boot.

    Not sure a neutrino would do it. A muon however...

    --
    A lie can travel halfway around the world while the truth is putting on
    its shoes.


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From David Higton@3:633/10 to All on Wed Dec 31 20:19:14 2025
    In message <10j3tdq$29dt0$1@dont-email.me>
    Pancho <Pancho.Jones@protonmail.com> wrote:

    On 12/30/25 20:00, David Higton wrote:
    In message <10iv40e$1e1ba$1@dont-email.me>
    Pancho <Pancho.Jones@protonmail.com> wrote:

    IPv6 seems like a world of pain.

    In my experience it just works.


    I'm surprised. Accepting that you do not do some of the things I do,
    like policy routing rules based upon a host computer IP, I'm still
    seeing servers on the internet that advertise they should work with IPv6
    but don't. This means they don't fall back to IPv4.

    I'm not far enough along in my understanding to be entirely confident,
    but I would be surprised if I were wrong.

    I've not encountered anything that's more difficult in IPv6 than in IPv4.
    I'm certain that IPv6 works just as well and as reliably as IPv4; after
    all, it's been in global-scale use for many years now, so all the issues
    have been solved. But there's always scope for something to go wrong,
    like in the example I quoted earlier where there was an IPv6 interface
    that didn't previously exist, and configuring it (which was no more
    difficult than the original IPv4 config, which was done so long ago that everyone had forgotten it) simply hadn't been done. Since everything mainstream now defaults to IPv6, there were two fault symptoms, depending
    which browser and OS were in use:

    1) The site appeared unavailable;

    2) The site was reached, but only after a delay.

    Nothing about it was a problem of IPv6 per se.

    I'd be interested to know what "policy routing rules based upon a host
    computer IP" you're using. My router runs OpenWRT. Everything gets
    its IPv6 address via DHCPv6. The traffic rules pick up the device by
    name, so, if the IPv6 address changes, the rules change automatically
    to match.

    David

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Wed Dec 31 21:50:50 2025
    On 31/12/2025 20:00, Jim Diamond wrote:
    Given that this mystery IP didn't show up the next time I rebooted (or
    again just now (to test)), I consider that unlikely. And grepping through /etc I don't see any unexpected instances of either (a) the mystery IP, or (b) the string "wlan". Not that this is an exhaustive hunt.

    Linux has had a fit about the naming of interfaces. wlan seems deprecated.

    --
    ?A leader is best When people barely know he exists. Of a good leader,
    who talks little,When his work is done, his aim fulfilled,They will say,
    ?We did this ourselves.?

    ? Lao Tzu, Tao Te Ching


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Pancho@3:633/10 to All on Thu Jan 1 11:02:44 2026
    On 12/31/25 20:23, David Higton wrote:
    In message <10j3tmk$29ec2$1@dont-email.me>
    Pancho <Pancho.Jones@protonmail.com> wrote:

    On 12/31/25 11:36, The Natural Philosopher wrote:
    On 30/12/2025 20:00, David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not
    necessary

    So making the implementation of a firewall absolutely mandatory


    Linux IPv6 does appear to use random IPv6 address for outbound
    connections, which have a limited lifespan. This appears to be something
    like 1-7 days, but if very short lifespans were used it could offer a
    protection similar to NAT. I need to investigate a bit further, but I
    don't think IPv6 needs to be inherently less safe.

    I use Ubuntu, which gives itself two routable IPv6 addresses. One is
    always the same (good if you have an internet-accessible server) and
    the other is new each reboot (good for privacy). Maybe it's the same
    for you?


    Yes, I'm using an Ubuntu derivative. That is what I had. I had the IPv6 address DHCPv6/RA assigned, plus there was a temporary one. Most
    external connections to the WAN went over the temporary one. I guess
    this is to not expose my permanent routable IPv6 address, the one I'm
    likely to have open ports on.

    AIUI, the temporary IPv6 address is preferred for one day, but hangs
    around for 7 days.


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Pancho@3:633/10 to All on Thu Jan 1 10:50:08 2026
    On 12/31/25 20:19, David Higton wrote:
    In message <10j3tdq$29dt0$1@dont-email.me>
    Pancho <Pancho.Jones@protonmail.com> wrote:

    On 12/30/25 20:00, David Higton wrote:
    In message <10iv40e$1e1ba$1@dont-email.me>
    Pancho <Pancho.Jones@protonmail.com> wrote:

    IPv6 seems like a world of pain.

    In my experience it just works.


    I'm surprised. Accepting that you do not do some of the things I do,
    like policy routing rules based upon a host computer IP, I'm still
    seeing servers on the internet that advertise they should work with IPv6
    but don't. This means they don't fall back to IPv4.

    I'm not far enough along in my understanding to be entirely confident,
    but I would be surprised if I were wrong.

    I've not encountered anything that's more difficult in IPv6 than in IPv4.

    In general, I'm not claiming IPv6 is more difficult. However, I have 30+
    years of dealing with and solving the problems of IPv4. Switching to
    dual stack IPv6 is introducing new problems, which I need to understand
    and solve.

    The reason I was looking at IPv6 was to solve some poorly understood
    problems I appear to have with a new VoIP SIP provider.

    I'm certain that IPv6 works just as well and as reliably as IPv4; after
    all, it's been in global-scale use for many years now, so all the issues
    have been solved. But there's always scope for something to go wrong,
    like in the example I quoted earlier where there was an IPv6 interface
    that didn't previously exist, and configuring it (which was no more
    difficult than the original IPv4 config, which was done so long ago that everyone had forgotten it) simply hadn't been done. Since everything mainstream now defaults to IPv6, there were two fault symptoms, depending which browser and OS were in use:

    1) The site appeared unavailable;

    2) The site was reached, but only after a delay.

    Nothing about it was a problem of IPv6 per se.


    But in practice, I am having IPv6 problems. Not caused by IPv6 itself,
    but apparently by third party misconfiguration. It is no good IPv6 being
    good, or my routing implementation being good, if in practice I'm
    banging my head against third party problems.

    At this stage it is quite possible I have misconfigured something, but
    it takes a lot of effort for me to understand each problem, so I discuss
    my experience in this public forum to gauge other people's practical experience.

    I'd be interested to know what "policy routing rules based upon a host computer IP" you're using. My router runs OpenWRT. Everything gets
    its IPv6 address via DHCPv6. The traffic rules pick up the device by
    name, so, if the IPv6 address changes, the rules change automatically
    to match.


    I use permanent VPN tunnels to access the WAN, something like NordVPN.
    Some LAN services I specifically want to route through the VPN, some I specifically want to route over the standard WAN. One way I have
    traditionally done this is to using a pfSense router to use a service
    internal LAN IP address to decide which external gateway to route over.
    Other people might do something similar using VLANs, but my switch
    hardware strips VLAN tags.

    For the avoidance of doubt, I'm relatively naive w.r.t. networking. I
    just knock up the first thing that works, rather than do something
    technically orthodox.

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Richard Kettlewell@3:633/10 to All on Thu Jan 1 11:34:04 2026
    The Natural Philosopher <tnp@invalid.invalid> writes:
    On 31/12/2025 20:18, Richard Kettlewell wrote:
    Pancho <Pancho.Jones@protonmail.com> writes:
    The Natural Philosopher wrote:
    David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not
    necessary
    So making the implementation of a firewall absolutely mandatory


    Linux IPv6 does appear to use random IPv6 address for outbound
    connections, which have a limited lifespan. This appears to be
    something like 1-7 days, but if very short lifespans were used it
    could offer a protection similar to NAT. I need to investigate a bit
    further, but I don't think IPv6 needs to be inherently less safe.

    NAT does not offer any protection. The reason that a typical domestic
    NAT-equipped router protects you from inbound connections is that it
    has a firewall as well. (Getting a packet addressed to your internal
    addresses to your external interface is inconvenient for many
    attackers, for sure, but straightforward for your ISP or anyone who
    can hack or coerce them.)

    How?
    Genuine question.

    Same as routing any other packet. Make sure there?s an appropriate
    routing table entry for the customer addresses on the ISP?s
    customer-facing router (and whatever intermediate routers there are
    between that and the attack source), then call socket/connect/write.

    The question is then what the customer router does with it.

    * If it follows the strong end system then the packet is discarded
    before NAT even comes into the question.
    Linux follows the weak end system model by default, so this
    possibility doesn?t apply to Linux-based router unless someone has
    taken the trouble to change its behavior somehow.

    * If there?s a basically competent firewall on the customer router then
    the packet is discard by that.

    * If there?s a NAT then it gets to look at the packet, but it won?t
    match any of the rules that enable translation, so it will not be
    modified at this stage.

    * All that?s now left is normal routing, so the packet passes on to its
    destination on the customer network.

    https://www.greenend.org.uk/rjk/tech/nat.html has a worked example.

    --
    https://www.greenend.org.uk/rjk/

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Thu Jan 1 13:21:10 2026
    On 01/01/2026 12:26, Richard Kettlewell wrote:
    The internal network destination address_is_ in its routing table.

    But not its port. And not in its NAT table,

    The essence of NAT is that the originating port cannot be accessed directly.

    And that *everything* gets translated, There is no concept of 'allowing
    some shit through even though it's not in my NAT tables' beyond
    theoretical.

    Whether that is a feature of NAT or a firewall, is semantics, but *in practice* it is not a security risk.

    If the police want to examine my server, they simply knock on the door
    with a warrant. No matter what protocol is on the interface



    --
    All political activity makes complete sense once the proposition that
    all government is basically a self-legalising protection racket, is
    fully understood.



    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Jim Diamond@3:633/10 to All on Fri Jan 2 10:00:00 2026
    On 2025-12-31 at 17:50 AST, The Natural Philosopher <tnp@invalid.invalid> wrote:
    On 31/12/2025 20:00, Jim Diamond wrote:
    Given that this mystery IP didn't show up the next time I rebooted (or
    again just now (to test)), I consider that unlikely. And grepping through >> /etc I don't see any unexpected instances of either (a) the mystery IP, or >> (b) the string "wlan". Not that this is an exhaustive hunt.

    Linux has had a fit about the naming of interfaces. wlan seems deprecated.

    I don't think "Linux" has had a fit about naming of interfaces. All of my machines (RPi and otherwise) have "wlan0". But I have seen more exotic
    names on some distros.

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Richard Kettlewell@3:633/10 to All on Thu Jan 1 13:48:36 2026
    The Natural Philosopher <tnp@invalid.invalid> writes:
    On 01/01/2026 12:26, Richard Kettlewell wrote:
    The internal network destination address_is_ in its routing table.

    But not its port. And not in its NAT table,

    Irrelevant.

    I literally did the experiment, you can see the results in the link
    posted earlier.

    --
    https://www.greenend.org.uk/rjk/

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Richard Kettlewell@3:633/10 to All on Thu Jan 1 12:26:52 2026
    The Natural Philosopher <tnp@invalid.invalid> writes:
    On 01/01/2026 11:34, Richard Kettlewell wrote:
    The Natural Philosopher <tnp@invalid.invalid> writes:
    On 31/12/2025 20:18, Richard Kettlewell wrote:
    NAT does not offer any protection. The reason that a typical domestic
    NAT-equipped router protects you from inbound connections is that it
    has a firewall as well. (Getting a packet addressed to your internal
    addresses to your external interface is inconvenient for many
    attackers, for sure, but straightforward for your ISP or anyone who
    can hack or coerce them.)

    How?
    Genuine question.
    Same as routing any other packet. Make sure there?s an appropriate
    routing table entry for the customer addresses on the ISP?s
    customer-facing router (and whatever intermediate routers there are
    between that and the attack source), then call socket/connect/write.
    The question is then what the customer router does with it.
    * If it follows the strong end system then the packet is discarded
    before NAT even comes into the question.
    Linux follows the weak end system model by default, so this
    possibility doesn?t apply to Linux-based router unless someone has
    taken the trouble to change its behavior somehow.
    * If there?s a basically competent firewall on the customer router
    then
    the packet is discard by that.
    * If there?s a NAT then it gets to look at the packet, but it won?t
    match any of the rules that enable translation, so it will not be
    modified at this stage.

    Ah. I misunderstood your original post. Sure the ISP can send an
    internally addressed packet to your router. If it wants to lose its
    customer base.

    * The police turn up with a warrant and inform you that if you don?t
    help them break into a certain customer?s network, you will go to
    prison.

    * The mafia turn up with a gun, and inform you that if you don?t help
    them break into a certain customer?s network, you will be shot.

    * A teenager who follows full-disclosure exploits the latest zero day to
    break into the ISP?s network, and from there to as many customers as
    they can reach.

    * An ISP staff member suspects that a friend who happens to be a
    customer is having an affair with their wife. Overwhelmed by jealousy
    they decide to attempt to hack the customer?s computers.

    You can probably imagine more scenarios.

    But will that get forwarded along?

    * All that?s now left is normal routing, so the packet passes on to its
    destination on the customer network.
    https://www.greenend.org.uk/rjk/tech/nat.html has a worked example.

    I can't believe that my router would accept arbitrary packets with
    an internal destination address on its external facing port...

    It probbaly doesn?t: it might use the strong end system model, and if
    not it might have a firewall preventing it. But whatever its
    configuration, it?s not NAT that stops it.

    if the destination is not in its tables, it will be automatically discarded...

    The internal network destination address _is_ in its routing table.
    That?s how it sends legitimate packets back to the internal network
    (e.g. when an internal host pings the router).

    On my router:

    $ ip route show|grep 172.17.207.0
    172.17.207.0/24 dev br0 proto kernel scope link src 172.17.207.1

    (172.17.207.0 is my internal network.)

    --
    https://www.greenend.org.uk/rjk/

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Thu Jan 1 20:56:46 2026
    On 01/01/2026 17:54, Richard Kettlewell wrote:
    Jim Diamond <zsd@jdvb.ca> writes:
    The Natural Philosopher <tnp@invalid.invalid> wrote:
    On 31/12/2025 20:00, Jim Diamond wrote:
    Given that this mystery IP didn't show up the next time I rebooted (or >>>> again just now (to test)), I consider that unlikely. And grepping through >>>> /etc I don't see any unexpected instances of either (a) the mystery IP, or >>>> (b) the string "wlan". Not that this is an exhaustive hunt.

    Linux has had a fit about the naming of interfaces. wlan seems
    deprecated.

    I don't think "Linux" has had a fit about naming of interfaces. All
    of my machines (RPi and otherwise) have "wlan0". But I have seen more
    exotic names on some distros.

    TNP is thinking of the ?predictable interface names? scheme, which is a misnomer since predicting the name depends on information you don?t necessarily have readily to hand. Really it is about keeping the mapping between names and specific hardware interfaces stable, which is valuable
    in some situations and inconvenient in others.

    Er no. On my HP laptop the interface is something like (runs to laptop
    and returns) "wlo1"...


    IMHO they should have taken a lesson from filesystem naming, and allowed interfaces to be identified by their attributes, rather than depending
    on a unique name.


    --
    "Fanaticism consists in redoubling your effort when you have
    forgotten your aim."

    George Santayana


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Richard Kettlewell@3:633/10 to All on Thu Jan 1 17:54:34 2026
    Jim Diamond <zsd@jdvb.ca> writes:
    The Natural Philosopher <tnp@invalid.invalid> wrote:
    On 31/12/2025 20:00, Jim Diamond wrote:
    Given that this mystery IP didn't show up the next time I rebooted (or
    again just now (to test)), I consider that unlikely. And grepping through >>> /etc I don't see any unexpected instances of either (a) the mystery IP, or >>> (b) the string "wlan". Not that this is an exhaustive hunt.

    Linux has had a fit about the naming of interfaces. wlan seems
    deprecated.

    I don't think "Linux" has had a fit about naming of interfaces. All
    of my machines (RPi and otherwise) have "wlan0". But I have seen more
    exotic names on some distros.

    TNP is thinking of the ?predictable interface names? scheme, which is a misnomer since predicting the name depends on information you don?t
    necessarily have readily to hand. Really it is about keeping the mapping between names and specific hardware interfaces stable, which is valuable
    in some situations and inconvenient in others.

    IMHO they should have taken a lesson from filesystem naming, and allowed interfaces to be identified by their attributes, rather than depending
    on a unique name.

    --
    https://www.greenend.org.uk/rjk/

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Tauno Voipio@3:633/10 to All on Thu Jan 1 22:37:44 2026
    On 31.12.2025 22.09, Jim Diamond wrote:
    On 2025-12-31 at 09:15 AST, Richard Kettlewell <invalid@invalid.invalid> wrote:
    druck <news@druck.org.uk> writes:
    On 30/12/2025 01:00, Jim Diamond wrote:
    However, it was worth a look. Maybe. According to the router, the
    "mystery" address is paired with the wifi card's actual ethernet (MAC)

    MAC's aren't just Ethernet, WiFi and Bluetooth interfaces also have them. >>>
    address, whereas the "proper" address is (currently) paired with the
    same ethernet address, except the last octet is 8C instead of 8D.
    This makes me think that it is showing "Connected" or "Disconnected"
    according to the ethernet address which is working, and it is not
    careful about pairing that with the correct IPv4 address.

    You often see a difference of 1 when something creates a virtual
    network interface for use by a virtual machine or container. The
    virtual network card is assigned the second IP address and can operate
    independently from anything using the hosts primary interface and IP
    address.

    At least on my 3B and 4B, the wired and wireless interfaces have
    adjacent MACs.

    PS C:\Users\rjk> ssh shairo ip link show
    1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN mode DEFAULT group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    2: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode DEFAULT group default qlen 1000
    link/ether dc:a6:32:cb:73:6b brd ff:ff:ff:ff:ff:ff
    3: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc fq_codel state UP mode DORMANT group default qlen 1000
    link/ether dc:a6:32:cb:73:6c brd ff:ff:ff:ff:ff:ff

    If both interfaces were connected to the same network then I might see
    something similar to Jim?s situation.

    I did ask Jim for ?ip addr show? output but it has not appeared.

    Mea culpa, I thought I did.

    Here is today's output... but I have long since gotten rid of that extra
    IP, so I'm not sure if this is at all interesting:

    1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host noprefixroute
    valid_lft forever preferred_lft forever
    2: eth0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc mq state DOWN group default qlen 1000
    link/ether dc:a6:32:37:93:8c brd ff:ff:ff:ff:ff:ff
    3: wlan0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
    link/ether dc:a6:32:37:93:8d brd ff:ff:ff:ff:ff:ff
    inet 192.168.2.74/24 brd 192.168.2.255 scope global noprefixroute wlan0
    valid_lft forever preferred_lft forever
    inet6 fe80::d10a:4386:b7f7:43f9/64 scope link noprefixroute
    valid_lft forever preferred_lft forever

    Should it happen again I'll capture this output in case it helps find the source.

    Jim

    If your system is running NetworkManager, it is the culprit.

    In my RasPi3B+ router, I disabled and stopped NetworkManager.
    systemd-networkd is perfectly capable to handle the DHCP
    client duties.

    After this, you have to create the needed interface and
    network descriptions into /etc/systemd/network, and that's it.

    The two links in the directory are fine, do not mess with them.

    --

    Tauno Voipio


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From The Natural Philosopher@3:633/10 to All on Thu Jan 1 12:09:50 2026
    On 01/01/2026 11:34, Richard Kettlewell wrote:
    The Natural Philosopher <tnp@invalid.invalid> writes:
    On 31/12/2025 20:18, Richard Kettlewell wrote:
    Pancho <Pancho.Jones@protonmail.com> writes:
    The Natural Philosopher wrote:
    David Higton wrote:
    What I particularly like about IPv6 is that NAT/NAPT are simply not >>>>>> necessary
    So making the implementation of a firewall absolutely mandatory


    Linux IPv6 does appear to use random IPv6 address for outbound
    connections, which have a limited lifespan. This appears to be
    something like 1-7 days, but if very short lifespans were used it
    could offer a protection similar to NAT. I need to investigate a bit
    further, but I don't think IPv6 needs to be inherently less safe.

    NAT does not offer any protection. The reason that a typical domestic
    NAT-equipped router protects you from inbound connections is that it
    has a firewall as well. (Getting a packet addressed to your internal
    addresses to your external interface is inconvenient for many
    attackers, for sure, but straightforward for your ISP or anyone who
    can hack or coerce them.)

    How?
    Genuine question.

    Same as routing any other packet. Make sure there?s an appropriate
    routing table entry for the customer addresses on the ISP?s
    customer-facing router (and whatever intermediate routers there are
    between that and the attack source), then call socket/connect/write.

    The question is then what the customer router does with it.

    * If it follows the strong end system then the packet is discarded
    before NAT even comes into the question.
    Linux follows the weak end system model by default, so this
    possibility doesn?t apply to Linux-based router unless someone has
    taken the trouble to change its behavior somehow.

    * If there?s a basically competent firewall on the customer router then
    the packet is discard by that.

    * If there?s a NAT then it gets to look at the packet, but it won?t
    match any of the rules that enable translation, so it will not be
    modified at this stage.

    Ah. I misunderstood your original post. Sure the ISP can send an
    internally addressed packet to your router. If it wants to lose its
    customer base.

    But will that get forwarded along?

    * All that?s now left is normal routing, so the packet passes on to its
    destination on the customer network.

    https://www.greenend.org.uk/rjk/tech/nat.html has a worked example.

    I can't believe that my router would accept arbitrary packets with an internal destination address on its external facing port...if the
    destination is not in its tables, it will be automatically discarded...



    --
    "In our post-modern world, climate science is not powerful because it is
    true: it is true because it is powerful."

    Lucas Bergkamp


    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)
  • From Lawrence D?Oliveiro@3:633/10 to All on Fri Jan 2 10:00:00 2026
    On Thu, 01 Jan 2026 17:54:34 +0000, Richard Kettlewell wrote:

    IMHO they should have taken a lesson from filesystem naming, and
    allowed interfaces to be identified by their attributes, rather than depending on a unique name.

    What attributes would you use?

    --- PyGate Linux v1.5.2
    * Origin: Dragon's Lair, PyGate NNTP<>Fido Gate (3:633/10)