[nsd-users] Enabling recvmmsg logs errors from sendmmsg call

[Anand Buddhdev]

I've built NSD 4.1.14 on CentOS 7 with the --enable-recvmmsg option to
configure. Running with this NSD on a server produces such log entries:

error: sendmmsg [0]=::A.B.C.D count=1 failed: No route to host
error: sendmmsg [0]=::a count=1 failed: No route to host

In the above log lines, A.B.C.D is a real IP address that I have obscured.

I wasn't seeing any such log entries before. They've only appeared after
building NSD with --enable-recvmmsg.

The server has a static default route, so it should always be able to
send DNS responses. If I query the kernel, it says it has a route:

# ip route get A.B.C.D from 193.0.14.129
A.B.C.D from 193.0.14.129 via E.F.G.H dev em2
    cache

E.F.G.H is the static default gateway.

Has anyone else seen such errors? Any idea why NSD is emitting them?

Regards,
Anand