Andrew Gallagher wrote:
> Dear Maintainer,
>
> Unbound does not recover gracefully from network-level errors. For example, if
> it is running as a local resolver on a laptop it can easily be thrown into a
> broken state when connecting to a new network, or when trying to browse the
> web
> while accidentally disconnected from a network.
>
> 1. Disconnect laptop from a network
> 2. Restart unbound (to flush cache, simulates leaving it disconnected
> overnight)
> 3. Connect laptop to a network
> 4. Simulate a transient network failure by deleting the default route
> 5. `host www.google.com` - it does not work (obviously)
> 6. Fix the network failure by reinstating the default route or running "sudo
> dhclient eth0"
> 7. `host www.google.com` again. It is still broken.
>
> The problem can only be fixed by either manually flushing all the offending
> records or restarting unbound. This is not user friendly.
>
> Andrew.
Hi, Andrew:
Can you try your sequence again, and then try the following:
8. `unbound-control flush_infra all`
9. `host www.google.com`
If the "flush_infra" unbound-control command fixes the problem, we might
be able to insert this as a hook to be run after an interface is
reconfigured.
You might also try lowering the "infra-host-ttl" value in the Unbound
configuration.
--
Robert Edmonds
edmo...@debian.org
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
