On Wed, Apr 15, 2020 at 10:53:49PM +0200, Ben wrote: <snip> > I have exactly one device - an Apple smartphone - within one of the > subnets, that Unbound is not able to send "some" data. The log tells us > "sendto failed: No buffer space available". Beside the error message, > the device seems to work without any issues. It gets its DNS queries > resolved. And it is the only device triggering this unbound behaviour.
AFAIU, ENOBUFS happens when the NIC transmit queue is full. Have you looked at the interface statistics to see if there are many dropped packets? Try, e.g., $ netstat -ni > > WHAT HAS BEEN DONE SO FAR: <snip> > (4) pf.conf > bnd_flows = "1024" > bnd_qlimit = "1024" > # respective queue where the traffic from router to device should pass > queue int_guests parent int_ingress bandwidth $int_guests max > $int_guests_max flows $bnd_flows qlimit $bnd_qlimit I don't have any experience with queueing, but you might try disabling it to rule it out. Without knowing anything about the implementation, I wouldn't be surprised if it dropped UDP packets in a way that triggered ENOBUFS, resembling (or emulating) the behavior of a full TX queue.
