Ntpdate is "deprecated", it was working but you had to invoke it as ntpdate-debian; In case your Pi has been sitting offline for a long time, the time jump it would cause could disrupt the machine to some degree.
I'm not sure it still works so well on jessie, I think it's better to use ntpd only, and let it adjust the time progressively. In my experience ntpd works fine on raspbian (no worse than standard debian/x86: from time to time it quits on me...)
Great news: systemd is futzing with NTP too. If you have installed the ntp package and wonder wtf are those unknown servers among your peers, just look at /etc/systemd/timesyncd.conf.
I have disabled the systemd-timesyncd service and now my list of peers is the one I've set in ntp.conf. (really this is appalling.)
What's missing from your post is your ntp.conf file... This is what I use for servers:
Code: Select all
server refclock.lan1 iburst
server ntp.lan2 iburst
server 127.127.1.1 # Local LCL clock
fudge 127.127.1.1 stratum 6 # Local clock policy:
# 1. Do not void time from our selected servers (stratum 1 or 2), but
# 2. do not trust broadcasted time or random stratum 15 hosts
(I'm not sure about the use of iburst. I wonder if it is not the cause of ntp quitting from time to time, but it's in this file and this ntpd doesn't behave worse than others on my network.)
and this is my peers list:
Code: Select all
# ntpq --peers
remote refid st t when poll reach delay offset jitter
*refclock.lan1 .DCFa. 1 b 40 64 77 1.121 -3.262 1.407
+ntp.lan2 192.168.0.1 2 b 29 64 177 1.055 -3.266 1.405
+some.reputa xxx.xxx.xxx.xxx 2 u 37 64 377 6.518 2.631 2.256
LOCAL(1) .LOCL. 6 l 503 64 200 0.000 0.000 0.000
172.16.255.255 .BCST. 16 b - 64 0 0.000 0.000 0.000
NTP uses port 123/udp. Make sure your firewall lets it through.