[Pdns-users] Powerdns stops slaving
bert.hubert at netherlabs.nl
Thu Jun 28 12:27:38 UTC 2012
In the 3.1 release notes you can find a MySQL related bug fix which might be related to your issue.
• It turns out we were using libmysqlclient in a thread-unsafe manner. This issue was reported and painstakingly debugged by Marc Haber. Presumably fixed incommit 2591
Since these release notes have been written, we can confirm that this issue is fixed by that commit.
Could you check if 3.1, which contains other worthwhile fixes, makes your problem go away?
Netherlabs Computer Consulting BV - http://www.netherlabs.nl/
On Jun 28, 2012, at 10:59 AM, Nick Douma wrote:
> We have a production setup using PowerDNS 3.0-1 static. We're using a
> hidden master setup, with three public facing auth servers which
> actually serve the data. The hidden master is responsible for slaving
> zones from remote masters, and contains the master MySQL db which is
> synced to the three public servers using native replication mode.
> It occurs that, after an interval I have not yet determined, the pdns
> auth server stops slaving. All the usual slaving output (and normal
> output) to /var/log/syslog stops, and pdns stops responding to manual
> slave update requests using `pdns_control retrieve`. The only fix for
> this is restarting powerdns.
> For now I have added a monitoring check which generates a monitoring
> failure if no powerdns output is detected in syslog for two hours.
> However, I would like to understand why this is happening.
> As said, the server which is experiencing the problem is running
> pdns-static 3.0-1, downloaded from the website some time ago. We're
> running in non-dnssec mode, with the appropriate older db schema. The
> host system is a physical server running Ubuntu 10.04 LTS with
> 2.6.32-34-server x64 kernel. MySQL version is 5.1.63-0ubuntu0.10.04.1,
> all tables use InnoDB.
> Kind regards,
> Nick Douma
> Pdns-users mailing list
> Pdns-users at mailman.powerdns.com
More information about the Pdns-users