Bind cache delays and disaster recovery


During the disaster recovery exercise last month, @gm ran into a problem that needs fixing.

  • @gm was working from a laptop connected to an Enough instance setup with a VPN
  • the laptop was configured to use the DNS provided by the VPN, because he is connected via Free who suppresses private IP resolution
  • during the disaster recovery exercise (and in a real situation too), a new sub domain of is created and delegated to a new DNS server created for the occasion
  • services created for the recovery obtain their LE certificates, meaning propagation happened and the configuration of the DNS is valid
  • @gm was however unable to resolve the new name because the DNS he was using did not know about it. It still had the older view of the zones and failed to resolve

The obvious solution is to just wait for the refresh :slight_smile: But it would be much better if Enough DNS was able to refresh as quickly as LE. How can they be quicker ?