I deployed two changes in my PF config.
(1) Bigger Queue
I rearranged some queues and gave the queue holding the DNS traffic more
bandwidth and a higher qlimit on the affected interface.
bnd_flows = "1024"
bnd_qlimit = "1024"
guest_local = "850M"
queue guest_local parent guest_root bandwidth $guest_local max
$guest_local flows $bnd_flows qlimit $bnd_qlimit default
(2) Packet Prioritization
I added the following rule.
match inet proto { tcp, udp } user _unbound set prio 6
# Summary
After these changes I currently do not see "sendto failed" notices. My
guess is that change (2) for itself might fix the issue, as some people
reported that they experience the notice, but do not use queuing.
Thoughts?
