On 06/12/16 09:13, jose wrote: > I experienced similar behavior when I first set up a Linux server for the > pool. > > > > It turns out that I was dropping NTP packets due to the nf_conntrack_max > kernel parameter value being too low for the amount of traffic I was getting. > > > > I set it at 256000 and the server has been stable since then. I've configured > the server for 1 Gig network speed and am getting about 3 Mbps sustained NTP > traffic.
If you're running a dedicated server for NTP, there's really no need to have connection tracking enabled for NTP. The local NTP server will already be tracking client connections, and all conntrack will do is cause problems like the one you found. If you need conntrack for other purposes, you can exclude NTP from it with something like: iptables -t raw -I PREROUTING -j NOTRACK -p udp --dport 123 Paul _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
