We moved our complete office last week and changed a lot of the infrastructure. Since then we have regular timeslots with slow responding web pages.
By analysing we found out that our nameserver (pdns_server 3.3) floods the syslog in these times with
Aug 13 12:41:00 dns1 pdns[6938]: Recursive query for remote 172.16.2.10:2055 with internal id 269 was not answered by backend within timeout, reusing id
Setup is a master/slave setup with amongst others params:
allow-recursion= ...# here are all of our different networks
recursor= ...# here are two nameservers from our ISP datacentre
We don't use a separate recursor daemon, it's the pdns_server itself.
Name resolving is working, both with internal/external addresses. Zone replication, everything - but just sometimes it runs in these issues and is really spamming the syslog - like flushing a queue with these recursive queries.
If I restart the pdns_server during this flushing everything is fine again. (But no, it's not a valid solution to check the syslog for pdns spam and restart the daemon ;-) )
Do you have any hints how I can solve this?
