Discussion:
Handling expired domains
rohan.henry cwjamaica.com
2018-06-28 22:48:03 UTC
Permalink
Hello All,

If all zones on a slave server expire because the slave could not reach the master shouldn't the slave start working again once the master becomes reachable without having to tweak anything like the serial?

Thanks.

Regards,
Rohan
Browne, Stuart via bind-users
2018-06-29 00:18:16 UTC
Permalink
Assuming the slave can retrieve the SOA and zone, yup. It should just come right back online.

Stuart

From: bind-users [mailto:bind-users-***@lists.isc.org] On Behalf Of rohan.henry cwjamaica.com
Sent: Friday, 29 June 2018 8:48 AM
To: bind-***@lists.isc.org
Subject: Handling expired domains

Hello All,

If all zones on a slave server expire because the slave could not reach the master shouldn't the slave start working again once the master becomes reachable without having to tweak anything like the serial?

Thanks.

Regards,
Rohan
Niall O'Reilly
2018-06-29 19:03:30 UTC
Permalink
Post by rohan.henry cwjamaica.com
If all zones on a slave server expire because the slave could not reach the master shouldn't the slave start working again once the master becomes reachable without having to tweak anything like the serial?
The slave should start working again once it discovers that the master has become reachable. According to the circumstances, this moment may differ, either grossly or subtly, from the moment when the master actually becomes reachable.

For example, if the master itself has failed, been recovered, and been restarted, it will likely send NOTIFY messages to the slaves, which will then be aware of restored reachability, and will be able to resume service directly.

On the other hand, if the reachability failure is due to a network fault, the master will have continued running, and will have no reason to send NOTIFY on restoration of reachability. In this case, resumption of normal service will depend on how the slave server software implements recovery from an expiry event. I expect, but have never had occasion to confirm, that this would depend on the REFRESH and RETRY timers. this might involve a delay of some, or even many, hours.

In any recovery situation, I would be minded to check slave status within a few minutes of restoration of reachability, and to force the master to send NOTIFY messages in case any slaves had not yet resumed service.


Niall O'Reilly

Loading...