Re: Progress on tailscale on NetBSD (plus $ available)

[Kevin Bloom <ktnb%netbsd.org@localhost>]

> Been hammering the new tailscale package "in production" (which is to
> say using it as a preferred transport for a selection of remote access
> and deploy actions), and noticed three cases similar to the below when
> running a moderate sized rsyncs across tailscale (~50M changed out
> of~180M)
>
> client_loop: ssh_packet_write_poll: Connection to 100.82.213.3 port
> 22: Address family not supported by protocol family
> rsync: [sender] write error: Broken pipe (32)
> rsync error: unexplained error (code 255) at io.c(849) [sender=3.4.1]
>
> This is between a NetBSD-amd64 and some Ubuntu 24.04 hosts running
> 1.96.4. Restarting the process completed the rsync in each case.
>
> Previous transport for this was wireguard-go with a Ubuntu 22.04 host
> acting as a wireguard relay. I have had that case rarely go into a
> state where all traffic would stop and even restarting wireguard-go
> would not recover, so there *could* be a common cause and tailscale is
> just better at handling some "bad state"
>
> David

I'm not sure why that is happning. Is there anything in the
tailscaled log? At least I don't see anything obvious causing it just
snooping around the code.