Selecting 'Daemon + Drift' as ntp method does not set date

[borpin]

Selecting 'Daemon + Drift' as ntp method does not set date

Required Information:

• DietPi Version | 157
• SBC Device? OrangePi Zero
• Power supply used (EG: 5v 1A RAVpower)? Good!
• SD card used (EG: Sandisk ultra)? Kingston
• Distro (EG: Jessie) | uname -a? Linux DietPi 3.4.113-sun8i v64 #18 SMP PREEMPT Thu Jun 15 02:16:06 CEST 2017 armv7l GNU/Linux

Additional Information (if applicable):

• Can this issue be replicated on a fresh installation of DietPi? Yes

Expected behavior:

Selecting 'daemon + drift' as time syncronization method sets the correct date time

Actual behavior:

Date / Time is not being set correctly. Daemon does not seem to be running.

Steps to reproduce:

When creating a new image, in the dietpi setup configuration file, select 'Daemon + Drift' as ntp method.

Did you submit a dietpi-bugreport?

31ce7a2b-3cd9-4754-a9b4-24ac5eba4b88-0

Extra details:

A message relating to ntp is seen during setup indicating it has been successful.

On a fresh install as tzdata is installed the following message is seen (indicating date/time has not been set)

Current default time zone: 'Europe/London'
Local time is now:      Sat Mar 11 12:16:44 GMT 2017.
Universal Time is now:  Sat Mar 11 12:16:44 UTC 2017.
Run 'dpkg-reconfigure tzdata' if you wish to change it.

Not sure what should be running as the daemon but there is no sign of ntpd or systemd-timesyncd

Package ntp is installed.

It seems reasonable to use systemd-timesyncd

root@DietPi:/lib/systemd/system# systemctl status systemd-timesyncd.service
● systemd-timesyncd.service - Network Time Synchronization
   Loaded: loaded (/lib/systemd/system/systemd-timesyncd.service; disabled)
   Active: inactive (dead)
     Docs: man:systemd-timesyncd.service(8)

My Solution locally

• Edit /etc/systemd/timesyncd.conf and enabled the time servers line.
• Enable systemd-timesyncd systemctl enable systemd-timesyncd.service
• Start systemd-timesyncd systemctl start systemd-timesyncd.service
• Check timedatactl

root@DietPi:/usr/sbin# timedatectl status
      Local time: Sun 2017-10-22 09:33:52 BST
  Universal time: Sun 2017-10-22 08:33:52 UTC
        RTC time: Thu 1970-01-01 00:34:09
       Time zone: Europe/London (BST, +0100)
     NTP enabled: yes
NTP synchronized: no
 RTC in local TZ: no
      DST active: yes
 Last DST change: DST began at
                  Sun 2017-03-26 00:59:59 GMT
                  Sun 2017-03-26 02:00:00 BST
 Next DST change: DST ends (the clock jumps one hour backwards) at
                  Sun 2017-10-29 01:59:59 BST
                  Sun 2017-10-29 01:00:00 GMT

• Note NTP synchronized: no
• Enable syncronization timedatectl set-ntp true

Reference used [(https://feeding.cloud.geek.nz/posts/time-synchronization-with-ntp-and-systemd/)] although he intimates that ntp cannot be installed alongside systemd-timesyncd.

[MichaIng]

Jep, choosing 'Daemon + Drift' doesn't seem to have any effect. No service/process of ntp can be found active.

[Fourdee]

Quick test on Odroid C2, unable to replicate:

root@DietPi:~# timedatectl status
      Local time: Thu 2017-11-09 20:02:34 GMT
  Universal time: Thu 2017-11-09 20:02:34 UTC
        RTC time: n/a
       Time zone: Europe/London (GMT, +0000)
     NTP enabled: yes
NTP synchronized: yes
 RTC in local TZ: no
      DST active: no
 Last DST change: DST ended at
                  Sun 2017-10-29 01:59:59 BST
                  Sun 2017-10-29 01:00:00 GMT
 Next DST change: DST begins (the clock jumps one hour forward) at
                  Sun 2018-03-25 00:59:59 GMT
                  Sun 2018-03-25 02:00:00 BST

Could be specific to our OPi Zero image.

[MichaIng]

• Okay, for the first, dbus need to be installed for systemd to initiate systemd-timesyncd
• On Jessie then the service is starting as expected.
• But on Stretch, it fails if ntp is still installed and active: https://superuser.com/questions/938926/systemd-timesyncd-service-start-fails
  This error really just shows up on Stretch. On Jessie everything works, even if the mentioned /usr/sbin/ntpd is present.

https://feeding.cloud.geek.nz/posts/time-synchronization-with-ntp-and-systemd/

In any case the ntp package is not necessary anymore, so should be purged if timesyncd is used, and pool server can be changed via /etc/systemd/timesyncd.

[Fourdee]

@MichaIng

Great find 👍

But on Stretch, it fails if ntp is still installed and active:
In any case the ntp package is not necessary anymore, so should be purged if timesyncd is used

So in theory, we could simply apt-get purge ntp when user selects daemon + drift, and visa versa, reinstall as needed?

[MichaIng]

Jep, this also works on Jessie and we can keep installation slim this way. ntp pools need to be set in /etc/systemd/timesyncd.conf then.

But in terms of memory consumption ntp called by cron is still the preferred way, as systemd-timesyncd consumes up to 5M on my VMs + additional dbus memory.

[Fourdee]

@MichaIng

👍 I'll add this to v159 milestone

[borpin]

It has been suggested to me that timesyncd may not work on a Read Only filesystem https://community.openenergymonitor.org/t/ensuring-emonpi-emonbase-time-is-synchronised-with-an-ntp-server/5487/11

Not sure if this is useful information or not.

[Fourdee]

Tested 2 > 4 > 0 > 2 > after all commits on:

• 🈯️ Jessie
• 🈯️ Stretch

[MichaIng]

Works great, also tested with Jessie and Stretch VMs back and forth 🙂.

[Fourdee]

@MichaIng

Excellent 👍

I'll mark this as completed.