[Pdns-users] Problem with pipe backend scripts...

[Laimonas Selenis]

Hello,

I am using PowerDNS 2.9.17 with PipeBackend on a Linux machine. Backend
is a perl script, very similar to this one, which is in documentation
(plus several small modifications). Now when I am running AXFR query -
server is responding each second time only. Once - I get normal
response, and then I get an error:

dig @ns.domain.com domain.com axfr
;; communications error to xxx.xxx.xxx.xxx#53: end of file

PowerDNS log then looks like this:

Jun 16 12:43:34 xx pdns[21447]: AXFR of domain 'domain.com' initiated
by yyy.yyy.yyy.yyy
Jun 16 12:43:34 xx pdns[21447]: [PIPEBackend] Error from coprocess:
Unable to ascertain status of coprocess 21436 from 21447: No child
processes
Jun 16 12:43:34 s15147160 pdns[21447]: TCP nameserver: Unable to
ascertain status of coprocess 21436 from 21447: No child processes

Next query runs normally and after - error again, and so on...

After each such error, server is opening several new processes, running
my perl backend script and old processes are just hanging... So after
some time I have a lot of hanging processes in my system. Usually, after
each error, number of processes in the system increases by 2.

May this be a bug in PDNS server?

Thanks!
Laimonas