timedatectl reports wrong status for timesyncd in core18

[Oliver Grawert]

when calling timedatectl in Ubuntu Core 18 the tool reports the status of the
systemd-timesyncd.service ...

sadly it does report it wrongly (the tool reports it inactive while the service is actually running):

ogra@pi4:~$ timedatectl
                      Local time: Wed 2020-07-01 10:52:48 UTC
                  Universal time: Wed 2020-07-01 10:52:48 UTC
                        RTC time: n/a
                       Time zone: n/a (UTC, +0000)
       System clock synchronized: yes
systemd-timesyncd.service active: no
                 RTC in local TZ: no
ogra@pi4:~$ systemctl status systemd-timesyncd.service
● systemd-timesyncd.service - Network Time Synchronization
   Loaded: loaded (/lib/systemd/system/systemd-timesyncd.service; disabled; vendor preset: enabled)
   Active: active (running) since Tue 2020-06-30 18:06:18 UTC; 16h ago
     Docs: man:systemd-timesyncd.service(8)
 Main PID: 779 (systemd-timesyn)
   Status: "Synchronized to time server 91.189.94.4:123 (ntp.ubuntu.com)."
    Tasks: 2 (limit: 4915)
   CGroup: /system.slice/systemd-timesyncd.service
           └─779 /lib/systemd/systemd-timesyncd

[Oliver Grawert (ogra)]

any fix should be SRUed into the respective LTS releases for Ubuntu Core, added tasks for them

[Balint Reczey (rbalint)]

Works in Ubuntu Classic it seems.
Where are Ubuntu Core bugs tracked? Please reassign this but there.

[Balint Reczey (rbalint)]

Sorry, it may be a valid bug, but may need more triaging.

[Balint Reczey (rbalint)]

So far this was unreproducible in an armhf lxc container running on arm64 host.
Do you have a reproducer for Ubuntu Classic?

[Balint Reczey (rbalint)]

BTW does it work in Core 20?

[Oliver Grawert (ogra)]

i dont know where you guys track core bugs either since foundations took over the responsibility Ubuntu Core with core 18...

i can not reproduce it on classic anywhere but on each core18 image i have running at my house ...

the customer that initially reported it uses core18, so it is important to get it fixed there, regardless if it works in core 20 (i'll do some testing though, but i currently do not have any core20 installs around).

it is likely just core specific (confinement, the different way of assembling the rootfs or the fact that timedatectl has a wrapper on core)

[Dimitri John Ledkov (xnox)]

bugs specific to core are at https://github.com/snapcore/core18/issues/new or https://github.com/snapcore/core20/issues/new as appropriate

[Oliver Grawert (ogra)]

theer we go ... thanks a lot dimitri, i opened

https://github.com/snapcore/core18/issues/157

[Dimitri John Ledkov (xnox)]

Also in core, timedatectl is a wrapper shell script, instead of the real one. So, some things that work regular timedatectl may not work with the one in the code. For example timezone name mangling / changes.

Also units are moved around a lot, such that things are not "enabled" via symlinks in /etc, but in /lib.

So yeah, ideally bugs about core should go first into the core* projects, as they are very different from classic file layout.

[Dan Streetman (ddstreet)]

please reopen if this is still an issue