NTP users are strongly urged to take immediate action to ensure that their NTP daemons are not susceptible to being used in distributed denial-of-service (DDoS) attacks. Please also take this opportunity to defeat denial-of-service attacks by implementing Ingress and Egress filtering through BCP38.
ntp-4.2.8p15
was released on 23 June 2020. It addresses 1 medium-severity security issue in ntpd, and provides 13 non-security bugfixes over 4.2.8p13.
Are you using Autokey in production? If so, please contact Harlan - he's got some questions for you.
See ConfiguringEndrunPraecisCTRefclocksDev for discussion of this topic.
6.1.3. Installing an Endrun Pręcis CT CDMA based refclock.
I usually prefer Mototola Oncore reference clocks, but this one needed to be installed in the 32nd floor of a 40+ storied building, so it would be very expensive to get the GPS antenna installed.
So, instead I had ordered the little CDMA cell phone unit made by Endrun Technologies, and hoped that it would install easily.
Well, it sure did!
Here's the entire setup script:
1) Install the server from scratch with a standard FreeBSD 5.3 ISO CD, enable sshd and define a userid for myself.
2) Left the noisy server room and did the rest of the work over the SSH connection.
3) Compiled a new kernel (NTP), made by copying the
GENERIC
setup file to
NTP
, and then add
options PPS_SYNC
, before
config; make depend; make; make install
and reboot.
4) Downloaded a recent ntp-dev tarball and used scp to put on the new server, then did the usual
./configure; make; make install
5) Got rid of the default ntp* programs in
/usr/sbin
, so that the newly installed program in
/usr/local/bin
would be picked up. I made a tiny ntp.conf file with just
driftfile=/etc/ntp.drift
server=ntp9.xxxxxxx # One of my current GPS clocks in Norway
I also added a few lines to /etc/rc.conf to enable ntpd on reboot:
ntpd_enable="YES"
ntpd_flags="-g -p /var/run/ntpd.pid"
ntpd_program="/usr/local/bin/ntpd"
This worked at once, with delay times of about 150 ms.
6) At this point I attached the Pręcis CT (PCT) unit to my (Windows) laptop using the supplied serial cable, and started up a terminal (HyperTerminal) window. After a short while the CDMA clock started to blink green regularly every second, and the terminal window showed a date&time string at the same time.
7)
PROBLEM AREA! The docs for the PCT states that entering the command
ctime=off
will stop the regular date/time string output, but this did not work with Hyperterminal. Instead I started up a Dos-based terminal emulator I wrote more than 20 years ago, and that was the key. I believe
CKERMIT
would almost certainly work the same way.
8) I followed the PCT docs and enabled TrueTime emulation, followed by
ctime=on
, and the unit started to print out TrueTime format date/time strings.
9) I then moved the PCT to the new FreeBSD server, and added one more line to the
ntp.conf
file:
server 127.127.5.0 prefer
fudge 127.127.5.0 refid CDMA
I also created a soft link:
ln -s /dev/cuaa0 /dev/true0
After
/etc/rc.d/ntpd restart
the PCT serial signal was picked up, but the offset and jitter was pretty bad, in the sub-10 ms range.
10) The documentation for the PCT suggest using the Palisade driver, but that would have required me to move the unit back to my laptop and reconfigure it, and I knew that using a true PPS driver should be even better, so I tried to follow the documentation which tells you to add a pps line to
ntp.conf
.
I found it much easier to use the regular ATOM driver instead, this worked at once:
server 127.127.22.0
fudge 127.127.22.0 refid CDMA
and
ln -s /dev/cuaa0 /dev/pps0
This also worked at once, and the PPS interface got down well below the ms level almost at once.
11) The only remaining item was to copy the two
ln -s
commands into
/etc/rc.local
, since this is needed to make sure that the links are recreated after every reboot. (FreeBSD 5+ use a
devfs
virtual filesystem instead of a regular
/dev
directory, so all symbolic devices must be recreated each time on startup!)
usntp1# cat rc.local
ln -s /dev/cuaa0 /dev/pps0
ln -s /dev/cuaa0 /dev/true0
Today is friday, four days later, and it seems that the (supposedly OCXO/TCXO?) crystal within the PCT has stabilized nicely: I get consistent
ntpq -p
displays like this:
C:\>ntpq -p usntp1
remote refid st t when poll reach delay offset jitter
======================================================================
+ntp9.hda.hydro. .GPS. 1 u 19 64 377 150.080 -0.736 0.145
oPPS(0) .CDMA. 0 l 39 64 377 0.000 -0.004 0.004
+TRUETIME(0) .CDMA. 0 l 16 16 377 0.000 0.043 6.078
--
TerjeMathisen - 01 May 2005
Related Topics: EndrunPraecisRefclockUsers