o Respect the 'dnslocalhost' option, which wasn't previously possible
because dhcpleases clobbered /etc/hosts directly anyway.
o Align the code with how unbound does it: use a private file for
the hosts generation.
o HUP dnsmasq at the end of the hosts file regeneration.
o `--dhcp-hostsfile' was really really wrong as /etc/hosts is read
automatically, and the file specified here should have been adhering
to the dhcp leases format of on `--dhcp-host' argument per line.
From the docs:
Starting with version 4.2.9, strongSwan provides a much more flexible configuration of the loggers. Logger configurations in strongswan.conf have a higher priority than charondebug in ipsec.conf: If you define any loggers in strongswan.conf, charondebug does not have any effect at all.
Turns out unbound reload flushes cache as well. Also, unbound-control
may hang when it runs into an "undefined state" (stoping a stopped
service, right), and lastly it wants to chroot after being chrooted
on reload, something that cannot be fixed with using unbound-control.
Keep unbound-control-setup in case somebody finds the facility useful.
IPFW and PF are used for more than just firewall and shaper, e.g.
NAT, policy routing or shaping, and the plugin framework supports
registering multiple user-facing services per file.
pkg can display to-be-deleted packages, but won't resolve these
issues when triggered using "upgrade -n". Thus, it's pointless
to try to keep this code until a newer pkg release can actually
cope with resolving the full upgrade cycle before upgrading.
This is the glue needed to be entirely plugin-agnostic, it only
needs new hooks for spots we haven't pluginified yet. The dns
subsystem is another likely candidate in the long run.
Allows to move these plugins away from core: bsnmpd, igmpproxy,
relayd, miniupnpd. ACL and menu entries pending...
