ARP requests never succeed (timeout exceeded)

[dvdsk]

Stack: Embassy commit a4f8fdd, using the embassy-net-wiznet crate (0.1).

Context:
this happens after about 15 hours of working perfectly though that period is not fixed.

What I need:
Some help debugging this. Is there a place in smolltcp I should add extra tracing? I tried some already but I did not get wiser from it. I might also need some reading up on ARP etc, forgotten most of that stuff now. Is there a resource you can recommend?

source snippet:

pub async fn handle(
    stack: &Stack<'_>,
    publish: &Queues,
    led: LedHandle<'_>,
    driver_orderers: &slow::DriverOrderers,
) {
    let mut rx_buffer = [0; 1024];
    let mut tx_buffer =
        [0; max(SensMsg::ENCODED_SIZE, ErrorReport::ENCODED_SIZE) * 2];

    let mut socket = TcpSocket::new(*stack, &mut rx_buffer, &mut tx_buffer);
    socket.set_timeout(Some(Duration::from_secs(5)));
    socket.set_keep_alive(Some(Duration::from_secs(1)));
    let host_addr = Ipv4Address::new(192, 168, 1, 43);
    let host_port = 1234;

    debug!("Configured socket and connecting");
    loop {
        debug!("socket state: {:?}", socket.state());
        if let Err(e) = socket.connect((host_addr, host_port)).await {
            warn!("connect error: {}", e);
            Timer::after_secs(5).await;
            continue;
        }

        info!("(re-)connected");
        // Prevent out-dated data from being send
        publish.clear().await;

        let (reader, writer) = socket.split();
        match select(
            send_messages(writer, publish),
            receive_orders(reader, &led, &driver_orderers),
        )
        .await
        {
            select::Either::First(e) => {
                warn!("Error while sending messages: {}", e)
            }
            select::Either::Second(e) => warn!("Error receiving orders: {}", e),
        };
        // Or the socket will hang for a while waiting to close this makes sure
        // we can reconnect instantly
        socket.abort();
        Timer::after_secs(60).await; // Experiment: does this help?
    }
}

send_messages and receive_orders are rather simple. The former using socket.write_all() then socket.flush() the latter only does socket.read().

trace:
(note I edited the paths /home/david/.cargo/registry/src/index.crates.io-6f17d22bba15001f/smoltcp-0.12.0/src/macros.rs:18 becomes smoltcp/src/macros.rs:18

WARN  connect error: ConnectionReset
└─ large_bedroom_sensor_node::network::handle::{async_fn#0} @ src/network.rs:103
DEBUG timeout exceeded
└─ smoltcp::socket::tcp::{impl#9}::dispatch @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG timeout exceeded
└─ smoltcp::socket::tcp::{impl#9}::dispatch @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG timeout exceeded
└─ smoltcp::socket::tcp::{impl#9}::dispatch @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG timeout exceeded
└─ smoltcp::socket::tcp::{impl#9}::dispatch @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG timeout exceeded
└─ smoltcp::socket::tcp::{impl#9}::dispatch @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG socket state: Closed
└─ large_bedroom_sensor_node::network::handle::{async_fn#0} @ src/network.rs:95
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
DEBUG timeout exceeded
└─ smoltcp::socket::tcp::{impl#9}::dispatch @ smolltcp/src/macros.rs:18
DEBUG address 192.168.1.43 not in neighbor cache, sending ARP request
└─ smoltcp::iface::interface::{impl#2}::lookup_hardware_addr @ smolltcp/src/macros.rs:18
WARN  connect error: ConnectionReset

[Dirbaio]

this is more likely a bug in embassy-net-wiznet. Do you have a packet capture from the other host that would show whether the ARP requests are coming through or not?