I'm not so sure it's bound to that issue. But nevertheless I now opened an issue myself. Let's see.
That was quick. Analzed by Ondřej:
No, BIND 9 is not broken. The upstream servers are:
$ dig +norec A _.drive.pro.io @151.80.197.88
=> Timeout
then it retries the query over TCP:
$ dig +tcp +norec A _.drive.pro.io @151.80.197.88
=> End of File (that's what I found in the bind logs)
That is because the upstream server closes the TCP connection after accepting.
"As a temporary workaround, you can disable QNAME minimization on the server.
And kudos to the pro.io for breaking the DNS in a new innovative way..."
Mit freundlichen Grüssen
-Benoît Panizzon-