how keep unbound up if internet down ?

Unbound
1

Fedora 29, unbound 1.8.3

A machine rebooted while we had a brief (20 second ) glitch in our router.

systemd[1]: Starting Unbound recursive Domain Name Server...
Apr 23 18:49:54 asterisk unbound-checkconf[957]: unbound-checkconf: no errors in /etc/unbound/unbound.conf
Apr 23 18:50:10 asterisk systemd[1]: Started Unbound recursive Domain Name Server.
Apr 23 18:50:10 asterisk unbound[1282]: [1282:0] notice: init module 0: ipsecmod
Apr 23 18:50:10 asterisk unbound[1282]: [1282:0] notice: init module 1: validator
Apr 23 18:50:10 asterisk unbound[1282]: [1282:0] notice: init module 2: iterator
Apr 23 18:50:10 asterisk unbound[1282]: [1282:0] info: start of service (unbound 1.8.3).
Apr 23 18:50:10 asterisk unbound[1282]: [1282:0] info: generate keytag query _ta-4f66. NULL IN
Apr 23 18:50:33 asterisk unbound[1282]: [1282:0] error: .: failed lookup, cannot probe to master c.root-servers.net
Apr 23 18:50:33 asterisk unbound[1282]: [1282:0] error: .: failed lookup, cannot probe to master b.root-servers.net
Apr 23 18:50:36 asterisk unbound[1282]: [1282:0] notice: sendto failed: Address family not supported by protocol
Apr 23 18:50:36 asterisk unbound[1282]: [1282:0] notice: remote address is (inet_ntop error) port 48988
Apr 23 18:50:36 asterisk systemd[1]: unbound.service: Main process exited, code=killed, status=11/SEGV
Apr 23 18:50:36 asterisk systemd[1]: unbound.service: Failed with result 'signal'.

The router came up about 5 seconds later.

How do I /can I configure unbound so that it keeps trying for at least some period of time ? This is actually a remote machine, but by sheer luck one of our guys was there. The only thing the users knew was that nothing worked, even though the net was up.

Any help appreciated.

2

This might more of topic for systemd rather than Unbound. You can configure restart counts, exit values, and time intervals to retry crashes. Begin by reviewing options including keyword "restart"
https://www.freedesktop.org/software/systemd/man/systemd.service.html
-Eric

3

Well, thanks. Learned about systemd restart codes. "Restart=on-failure" should work. And it's better than the cron job I hacked together.

But it still seems a hack. I'd like to set unbound to be more robust in case the net went down, at least for some period of time.

sean

4

Just added Restart=on-abort to unbound.service.

But that made me realize another issue if you're depending on systemd and using a distribution, like I am : Fedora.

Any time unbound is upgraded ( And why is Fedora 29 still at 1.8.3 ? ), the unbound.service file can get replaced. So unless unbound.service is changed upstream to include Restart, it may happen at an upgrade.

sean

5

Hey Sean,

Just added Restart=on-abort to unbound.service.

But that made me realize another issue if you're depending on systemd
and using a distribution, like I am : Fedora.

Any time unbound is upgraded ( And why is Fedora 29 still at 1.8.3 ? ),
the unbound.service file can get replaced. So unless unbound.service is
changed upstream to include Restart, it may happen at an upgrade.

You need to learn more about systemd. With systemd, you can have a local
override.

If you run "systemctl edit unbound.service" then it will open an editor
where you can add your "Restart=on-abort". This then gets merged with
the system-supplied unit. Your overlay is recorded in:

/etc/systemd/system/unbound.service.d/override.conf

Even if unbound is later upgraded, your local change persists.

Regards,
Anand

6

Additionally, you could file a bug with fedora, as this seems like a useful addition to the unbound rpm in general.

Kind regards,
  Tom