PowerDNS replication failing (TCP nameserver had error, cycling backend: GSQLBackend unable to lookup)

Written by Claudio Kuenzler - 0 comments

Published on September 17th 2024 - Listed in Linux MySQL Databases DNS PowerDNS

---

As I'm currently in the midst of upgrading a PowerDNS cluster, I came across a replication error between the Primary (formerly known as Master) and the Secondary (previously Slave) DNS servers.

After creating a new DNS record on the Primary (10.10.10.53), the change was not replicated to the Secondary (10.20.20.53).

Logs on the Primary showed:

Sep 17 15:12:39 ip-10-10-10-53 pdns_server[3104867]: Received NOTIFY for example.com from 10.10.10.53 but slave support is disabled in the configuration
Sep 17 15:12:40 ip-10-10-10-53 pdns_server[3104867]: AXFR-out zone 'example.com', client '10.20.20.53', transfer initiated
Sep 17 15:12:40 ip-10-10-10-53 pdns_server[3104867]: TCP nameserver had error, cycling backend: GSQLBackend unable to lookup 'example.com|SOA':Could not prepare statement: SELECT content,ttl,prio,type,domain_id,disabled,name,auth FROM records WHERE disabled=0 and type=? and name=?: The client was disconnected by the server because of inactivity. See wait_timeout and interactive_timeout for configuring this behavior.
Sep 17 15:12:40 ip-10-10-10-53 pdns_server[3104867]: Received unsuccessful notification report for 'example.com' from 10.10.10.53:53, error: Query Refused

On the Secondary, the logs showed:

Sep 17 15:12:40 ip-10-20-20-53 pdns_server[48671]: XFR-in zone: 'example.com', primary: '10.10.10.53', unable to xfr zone (ResolverException): Remote nameserver closed TCP connection (This was attempt number 1. Excluding zone from slave-checks until 1726578820)

The relevant information is found on the Primary. The GSQLBackend (in this case using MySQL) was unable to run the mentioned SELECT query.

It turned out that on the Primary, MySQL was upgraded from 5.7 to 8.0 in order to support the database schema change in PowerDNS 4.7 - but after the MySQL restart, PowerDNS itself required a restart, too.

After a restart of the pdns service, the DNS replication worked correctly again:

root@ip-10-10-10-53:~# systemctl restart pdns
root@ip-10-10-10-53:~# tail -f /var/log/syslog
Sep 17 15:17:00 ip-10-10-10-53 pdns_server[3244998]: Received NOTIFY for example.com from 10.10.10.53 but slave support is disabled in the configuration
Sep 17 15:17:00 ip-10-10-10-53 pdns_server[3244998]: AXFR-out zone 'example.com', client '10.20.20.53', transfer initiated

Similar success log entry on the Secondary:

root@ip-10-20-20-53:~# tail -f /var/log/syslog
Sep 17 15:17:00 ip-10-20-20-53 pdns_server[48671]: AXFR-in zone: 'example.com', primary: '10.10.10.53', zone committed with serial 2024091702

Add a comment

Show form to leave a comment

Comments (newest first)

No comments yet.