Package: bind9 Severity: normal Dear Maintainer,
As suggested in a downstream bug [1], I am forwarding this here. Consider the following use case: - A user installs bind9 on his system; - bind is up and running normally; - The user makes changes to configuration files, bringing them to a bogus state, without restarting the service. - The bind9 package is upgraded, and therefore, restarted. Since the configuration file was in a bogus state before the upgrade, the package will be upgraded and the service will fail to start. Would it be reasonable to run a configuration file check in the postinst step before proceeding to restarting the service? Then, in case a config check fails, the service would not be restarted, which could prevent service disruption. [1] https://bugs.launchpad.net/ubuntu/+source/bind9/+bug/1492212 -- Athos Ribeiro
