You can disable it by specifying an empty control {} section.

This prevents inconsistency when using DDNS and you update your zone
as well before data in journal is flushed.
To choice is to invalidate data in the journal. If one choses to
revert to previous zone, he has to revert changes in the zone file
and overwrite current journal with the backed up one.
Note that the journal has to have newer mtime.

Submitted by Ondrej Caletka

This caused a bug when a SOA query answer was received before
reloading was complete, so the zone wasn't visible in the zone
database yet.

This prevents mixing up the current IXFR and started AXFR.

This was used before for knotc and zcompile, so the only problem is
when sources are configured and not installed, restart would expect
binary to be at $(sbindir)/knotd.

Change-Id: Ia44220989918f9eb53b3f2a2b1d74b092aef4246

Change-Id: I760f21c2dc4823c6b1c014c629f30c55827ee094

reffs #36

refs #36

refs #36

refs #36

AF_UNIX is now preferred over AF_INET sockets, although it is still
off by default.
When TCP/IP is used, default port changed from 5553->5533 (free).
AF_UNIX socket based control doesn't use (naturally) 'allow' section.

refs #36

refs #36

It is still possible to control knotd using signals though,
but the 'knotc' will use just the one configured channel not a mix
of signals and channel.

refs #36

Removed 'pidfile' config option in favour of 'rundir' for later
storage of UNIX sockets and other stuff.
Default of 'rundir' is $(localstatedir)/run/knot
Default of 'storage' is $(sharedstatedir)/knot

refs #36