Hi Aki, On Fri, 11 Oct 2013 21:01:02 +0300 Aki Tuomi <[email protected]> wrote:
> Are you actually using AXFR to something on pipebackend? If not, why > not just send FAIL when you get one? I can't imagine this being > performance issue as it's clearly identifiable. It's hardly worth the > trouble you are going thru now. No - I'm trying to actually disable AXFR on PipeBackend as it's not needed/used. The reason why initially I did not go with FAIL was because I thought backend got cycled after failed zone transfer when FAIL was used: pdns[6965]: TCP Connection Thread unable to answer a question because of a backend error, cycling (In fact all backend processes are intact after such failed transfer, so I'm a bit confused here) This seems to be only happening when something was already sent back to the client (SOA which precedes AXFR). When END is used instead I did not get that message, so that why I went with END instead. Just noticed when I filter out SOA when remote_ip == '0.0.0.0' both END and FAIL give identical results: transfer failure (nothing gets send back to the client) and no cycle message, so most likely I with FAIL instead. Thanx, Marcin _______________________________________________ Pdns-users mailing list [email protected] http://mailman.powerdns.com/mailman/listinfo/pdns-users
