I'm assuming that no output is bad here. Still, I don't see why a device that works shouldn't be able to participate in a bond. As a network interface, the wifi device produces and responds to network traffic. Are you saying the bonding takes place below the driver level?
On 2023-10-16 18:52, Igor Cicimov wrote:
Hi,
On Tue, Oct 17, 2023 at 12:12 PM Gary Dale <ga...@extremeground.com> wrote:
On 2023-10-16 18:52, Igor Cicimov wrote:
Hi,
Apparently neither interface supports it. According to what I
have read, calling mii-tool with no parameters should return a
terse list of all interfaces that support it. However, when I try
that, I get "No interface specified". Moreover,
# mii-tool enp10s0
SIOCGMIIPHY on 'enp10s0' failed: Operation not supported
# mii-tool wlxc4411e319ad5
SIOCGMIIPHY on 'wlxc4411e319ad5' failed: Operation not supported
which seems weird given that I have a recent, mainstream ASUS mainboard with a generic realtek onboard NIC that seems to be participating in the bonding.
I've also not seen any warnings that bonding requires a specific (and apparently rare) type of NIC. Indeed, my laptop seems to fail over nicely between ethernet and wifi.
Perhaps mii-tool is broken on Trixie?
I use systemd-networkd to configure bonding in the same way. I use the
"active-backup" mode and one parameter that I don't *think* you've set
is the "primary". According to
https://www.kernel.org/doc/Documentation/networking/bonding.txt, you'd
set "primary" to the interface which is always active if it's available.
So you probably want to set "bond-primary enp10s0" so that the system
will switch to the cable when it's connected; when the cable disconnects
it should switch over to the wifi. Without "primary" being set, I
suspect the system doesn't have any motivation to prefer the cable when
both are connected.