On a system having both ntp-client and rp-pppoe started in the same runlevel, it is probable that the administrator wants NTP to reach a host over the PPPoE connection. This fails with the current stable ebuilds because init wants to start ntp before rp-pppoe. A simple change in the dependencies in /etc/init.d/ntp does the trick. Reproducible: Always Steps to Reproduce: 1. add ntp-client and rp-pppoe to myrunlevel 2. configure rp-pppoe ; leave ntp-client's config as-is 3. rc myrunlevel Actual Results: NTP complains it can't reach pool.ntp.org, because the PPPoE connection has not yet been established Expected Results: Establish PPPoE connection and then sync clocks Using up-to date portage tree, stable ebuilds, x86, gentoo-sources kernel. A quick fix is to change the dependencies in /etc/init.d/ntp-client : depend() { before cron need net use dns logger >> after rp-pppoe } A "before" dependency in /etc/init.d/rp-pppoe should do the trick just as well.
rp-pppoe is not a supported net script
