Home / ubuntu / Questions / 254826
Accepted
ysap
Asked: 2013-02-14 13:58:12 +0800 CST

How to force a clock update using ntp?

  • 772

I am running Ubuntu on an ARM based embedded system that lacks a battery backed RTC. The wake-up time is somewhere during 1970. Thus, I use the NTP service to update the time to the current time.

I added the following line to /etc/rc.local file:

sudo ntpdate -s time.nist.gov

However, after startup, it still takes a couple of minutes until the time is updated, during which period I cannot work effectively with tar and make.

How can I force a clock update at any given time?

UPDATE 1: The following (thanks to Eric and Stephan) works fine from command line, but fails to update the clock when put in /etc/rc.local:

$ date ; sudo service ntp stop ; sudo ntpdate -s time.nist.gov ; sudo service ntp start ; date
Thu Jan  1 00:00:58 UTC 1970
 * Stopping NTP server ntpd     [ OK ] 
 * Starting NTP server          [ OK ] 
Thu Feb 14 18:52:21 UTC 2013

What am I doing wrong?

UPDATE 2: I tried following the few suggestions that came in response to the 1st update, but nothing seems to actually do the job as required. Here's what I tried:

  1. Replace the server to us.pool.ntp.org
  2. Use explicit paths to the programs
  3. Remove the ntp service altogether and leave just sudo ntpdate ... in rc.local
  4. Remove the sudo from the above command in rc.local

Using the above, the machine still starts at 1970. However, when doing this from command line once logged in (via ssh), the clock gets updated as soon as I invoke ntpdate.

Last thing I did was to remove that from rc.local and place a call to ntpdate in my .bashrc file. This does update the clock as expected, and I get the true current time once the command prompt is available.

However, this means that if the machine is turned on and no user is logged in, then the time never gets updates. I can, of course, reinstall the ntp service so at least the clock is updated within a few minutes from startup, but then we're back at square 1.

So, is there a reason why placing the ntpdate command in rc.local does not perform the required task, while doing so in .bashrc works fine?

ntp

18 Answers

  1. Instead of ntpdate (which is deprecated), use ntpd:

    sudo service ntp stop
sudo ntpd -gq
sudo service ntp start

    The -gq tells the ntp daemon to correct the time regardless of the offset (g) and exit immediately (q) after setting the time.

    • 572
  2. Best Answer

    Probably the ntp service is running, that's why ntpdate can't open the socket (port 123 UDP) and connect to ntp server.

    Try from command line:

    sudo service ntp stop
sudo ntpdate -s time.nist.gov
sudo service ntp start

    If you want to put this in /etc/rc.local use the following:

    ( /etc/init.d/ntp stop
until ping -nq -c3 8.8.8.8; do
   echo "Waiting for network..."
done
ntpdate -s time.nist.gov
/etc/init.d/ntp start )&
    • 364

  3. Use sntp to set the time immediately. Some examples from its man page:

    USAGE
     sntp ntpserver.somewhere
             is the simplest use of this program and can be run as an unprivileged command to check the current time and error in the local clock.

     sntp -Ss -M 128 ntpserver.somewhere
             With suitable privilege, run as a command or from a cron(8) job, sntp -Ss -M 128 ntpserver.somewhere will request the time from the server, and if that server reports that it is
             synchronized then if the offset adjustment is less than 128 milliseconds the correction will be slewed, and if the correction is more than 128 milliseconds the correction  will
             be stepped.

     sntp -S ntpserver.somewhere
             With suitable privilege, run as a command or from a cron(8) job, sntp -S ntpserver.somewhere will set (step) the local clock from a synchronized specified server, like the (dep‐
             recated) ntpdate(8), or rdate(8) commands.

    It does work with any ntp time server. A convenient list of servers can be found on ntppool.org.

    You need sudo privileges, for example:

    sudo sntp -Ss -M 128 0.de.pool.ntp.org
    • 70

  4. Use timedatectl (systemd service unit) to set the time. ntp is deprecated.

    sudo systemctl restart systemd-timesyncd.service

    You can check the time was updated reading the logs with journalctl -xe | tail.

    Status with timedatectl status and config in /etc/systemd/timesyncd.conf.

    Reference

    • 63

  5. As others have pointed out the best solution is to instruct ntpd to ignore the panic threshold, which is 1000 seconds by default. You can configure the panic threshold in one of two ways:

    • edit /etc/default/ntp and ensure that the -g option is present.
    • edit /etc/ntp.conf and place tinker panic 0 at the top

    So far this is essentially what others have recommended however there is one more step I think you should take. Install the fake-hwclock program:

    # apt-get install fake-hwclock


fake-hwclock: Save/restore system clock on machines without working RTC hardware

 Some machines don't have a working realtime clock (RTC) unit, or no
 driver for the hardware that does exist. fake-hwclock is a simple set
 of scripts to save the kernel's current clock periodically (including
 at shutdown) and restore it at boot so that the system clock keeps at
 least close to realtime. This will stop some of the problems that may
 be caused by a system believing it has travelled in time back to
 1970, such as needing to perform filesystem checks at every boot.

 On top of this, use of NTP is still recommended to deal with the fake
 clock "drifting" while the hardware is halted or rebooting.

    With fake-hwclock installed your machine will not start up thinking it is 1970 all over again. When your machine boots up it will set its clock to the timestamp fake-hwclock wrote during the last reboot/shutdown. This means you can have a somewhat correct clock in case there are network issues when you boot up.

    • 42

  6. ntpdate is a program different from the net dameon. NTPDate is probably erroring out on boot because ntpd is running on that socket.

    From the command line, run

    # sudo service ntp stop ; sudo ntpdate -s time.nist.gov ; sudo service ntp start

    You could also uninstall ntpd all together (apt-get remove ntp) and add a cron script to use ntpdate every hour or so.

    UPDATE

    ntp service probably won't have meaningful value for you on this system, so remove that first.

    # sudo apt-get remove ntp

    Now add the command:

    ntpdate -sb time.nist.gov

    to /etc/rclocal

    Reboot. Should be good at that point.

    • 13 
  7. rdate -s tick.greyware.com

    if all you want to do is set the clock once, simple

    • 10

  8. Note that some current Ubuntu based systems don't even use the NTP service by default now. On my Linux Mint 19 (Ubuntu 18.04) machine, time is kept by systemd-timesyncd.

    So to get an up to date time after it has lost sync, I just run 

    sudo systemctl restart systemd-timesyncd

    Since 15.04 Ubuntu uses systemd by default. Therefore critical systems like time are managed through systemd. To find what service your system is using run something like 

    systemctl list-unit-files | grep -E 'ntp|timesyncd'

    For TechJS on 16.04 the service was ntp. For myself on Ubuntu 18.04 (Mint 19) the service is systemd-timesyncd. Interestingly, I logged into a 16.04 server I have and it uses systemd-timesyncd as well.

    • 9

  9. Check out this one (no apt, no service):

    sudo date -s "$(wget -qSO- --max-redirect=0 google.com 2>&1 | grep Date: | cut -d' ' -f5-8)Z"

    Cheers to Prashant, who solved that issue

    • 7

  10. The correct way to do this on a Debian / Mint / Ubuntu (or other Debian derivative) system is to have the line

    NTPD_OPTS="-g"

    in the file

    /etc/default/ntp

    This ensures that when ntpd is started from the /etc/init.d/ntp script, it runs with the "-g" option, viz

     start-stop-daemon --start --quiet --oknodo --pidfile /var/run/ntpd.pid --startas /usr/sbin/ntpd -- -p /var/run/ntpd.pid -g -u 124:128

    to allow ntpd to correct the system time when it is more than 1000 s out, eg when the system time is January 1st 1970 on startup because there is no hardware RTC.

    • 7