cloudron 6.0.1 fail to install on a new fresh ubuntu 20.04 server

[TheNils]

hi there,
i'm trying to install cloudron on a new fresh ubunntu 20.04 server
the installation fail with the following error :

echo "==> Configuring host"
==> Configuring host
sed -e 's/^#NTP=/NTP=0.ubuntu.pool.ntp.org 1.ubuntu.pool.ntp.org 2.ubuntu.pool.ntp.org 3.ubuntu.pool.ntp.org/' -i /etc/systemd/timesyncd.conf
timedatectl set-ntp 1
Failed to set ntp: NTP not supported

i have tried to manually setup timedatectl with a ntp service but the same error append.
keep up the great work !

[nebulon]

on which VPS provider is this?

[TheNils]

@nebulon scaleway dedicated server (see here the store-1-s)

[nebulon]

Hm maybe the specific ubuntu image from scaleway does not pre-install ntp. Can you run apt-get install ntp and then try to install Cloudron again?

[TheNils]

@nebulon said in cloudron 6.0.1 fail to install on a new fresh ubuntu 20.04 server:

apt-get install ntp

i already done that but no change.
after installing $: timedatectl status rerpot :

               Local time: Tue 2020-12-08 13:50:22 CET
           Universal time: Tue 2020-12-08 12:50:22 UTC
                 RTC time: Tue 2020-12-08 12:50:22    
                Time zone: Europe/Paris (CET, +0100)  
System clock synchronized: no                         
              NTP service: n/a                        
          RTC in local TZ: no 

relating to this answer on serverfault :

timedatectl command reports that NTP is in use when either chronyd or systemd-timesyncd

apt-get install systemd-timesyncd seems to solve my problem ( chronyd does not work for me)...

[girish]

I did test installation on DEV1-M yesterday and it went fine. maybe the bare metal ones have a different ubuntu.

[girish]

On DEV1-M, the package is already there and timedatectl set-ntp 1 works on a vanilla ubuntu 20

root@scw-tender-euclid:~# systemctl status systemd-timesyncd
● systemd-timesyncd.service - Network Time Synchronization
     Loaded: loaded (/lib/systemd/system/systemd-timesyncd.service; enabled; vendor preset: enabled)
     Active: active (running) since Tue 2020-12-08 18:25:01 UTC; 1min 22s ago
       Docs: man:systemd-timesyncd.service(8)
   Main PID: 270 (systemd-timesyn)
     Status: "Initial synchronization to time server 10.194.3.3:123 (10.194.3.3)."
      Tasks: 2 (limit: 614)
     Memory: 1.7M
     CGroup: /system.slice/systemd-timesyncd.service
             └─270 /lib/systemd/systemd-timesyncd

Dec 08 18:25:00 ubuntu systemd[1]: Starting Network Time Synchronization...
Dec 08 18:25:01 ubuntu systemd[1]: Started Network Time Synchronization.
Dec 08 18:25:04 scw-tender-euclid systemd-timesyncd[270]: Network configuration changed, trying to establish connection.
Dec 08 18:25:04 scw-tender-euclid systemd-timesyncd[270]: Network configuration changed, trying to establish connection.
Dec 08 18:25:06 scw-tender-euclid systemd-timesyncd[270]: Network configuration changed, trying to establish connection.
Dec 08 18:25:36 scw-tender-euclid systemd-timesyncd[270]: Initial synchronization to time server 10.194.3.3:123 (10.194.3.3).

[girish]

It worked on a STARDUST1-S instance as well. Mmm, let's see if we have others reporting similar issue before making a fix.

[robi]

@girish may be easier to make a check for it or just add to list of apt-get install prerequisites

[TheNils]

@girish it may only be my server... the scalway support is working on my server since my last message because the server is randomly going offline for no apparent reason after(?) installing cloudron. they even tryed to change the server because the ubuntu reinstalaltion process got broken at some point.

on their ubuntu image openntpd is installed in place of systemd-timesyncd.

have you ever encountered this kind of probleme (server randomly going offline ) ?

[girish]

@thenils haven't encountered such problems, no. Looks like some scaleway issue. Maybe just create a new server install?

[TheNils]

i have watched journalctl until it crash and ...
Apparently it was systemd-sleep kicking in and suspending the server ...
i previsously missed it from logs because i was searching for errors.
( also i was assuming that it came from scalway because the first server was completly broken and they had to replace it)

i have disabled power management with :
$: systemctl mask sleep.target suspend.target hibernate.target hybrid-sleep.target

it should go fine now.

Thanks for your time

[robi]

@thenils wow that's quite the oversight from Scaleway.

A server OS doesn't normally come with power management by default.

There may be other things hidden there, so check if you can install your own clean OS with less surprises.

[TheNils]

@robi yes i'll might do it soon. i've seen in the logs xorg complaining about not finding any cursor. there's definitelly some funny stuff going on with this ubuntu image...

[TX-Hermit]

The exact same ntp issue also happens with the Ubuntu 20.04 image being used by Atlantic.net.

After initial VM spin-up, the output of timedatectl was showing as:

Local time: Tue 2021-01-05 00:04:37 UTC
Universal time: Tue 2021-01-05 00:04:37 UTC
RTC time: Tue 2021-01-05 00:04:37    
Time zone: Etc/UTC (UTC, +0000)       
System clock synchronized: yes                        
NTP service: n/a                        
RTC in local TZ: no

Due to the NTP service showing as n/a, even though it is verified as running and enabled, your setup script passing the timedatectl set-ntp 1 would fail with:

Failed to set ntp: NTP not supported

After wasting too much time on NTP issues by verifying configs and the working of the NTP daemon, I pulled the default NTP package and installed Chrony. Problems vanished.

For whatever reason, the default ntp package used in the image:

Package: ntp
Version: 1:4.2.8p12+dfsg-3ubuntu4

is not being recognized by timedatectl. I'm sure it is a simple fix, but it is just so much faster to replace ntp.

Hope this helps someone.

[girish]

@tx-hermit thanks for reporting, let me test out atlantic.net to see if we can fix this to make it work out of the box.

[girish]

I could reproduce the issue on atlantic.

root@cloudron:~# systemctl is-active ntp
active
root@cloudron:~# timedatectl set-ntp 1
Failed to set ntp: NTP not supported

Then, I saw this error in ntp service logs:

● ntp.service - Network Time Service
     Loaded: loaded (/lib/systemd/system/ntp.service; enabled; vendor preset: enabled)
     Active: active (running) since Tue 2021-01-05 01:28:44 UTC; 9min ago
       Docs: man:ntpd(8)
   Main PID: 23144 (ntpd)
      Tasks: 2 (limit: 2353)
     Memory: 1.5M
     CGroup: /system.slice/ntp.service
             └─23144 /usr/sbin/ntpd -p /var/run/ntpd.pid -g -u 107:115

Jan 05 01:28:49 cloudron ntpd[23144]: Soliciting pool server 69.164.213.136
Jan 05 01:28:49 cloudron ntpd[23144]: Soliciting pool server 91.189.89.199
Jan 05 01:28:49 cloudron ntpd[23144]: Soliciting pool server 104.194.8.227
Jan 05 01:28:49 cloudron ntpd[23144]: Soliciting pool server 50.205.244.39
Jan 05 01:28:50 cloudron ntpd[23144]: Soliciting pool server 91.189.89.198
Jan 05 01:28:50 cloudron ntpd[23144]: Soliciting pool server 173.255.215.209
Jan 05 01:28:50 cloudron ntpd[23144]: Soliciting pool server 68.54.100.49
Jan 05 01:28:50 cloudron ntpd[23144]: Soliciting pool server 66.79.136.235
Jan 05 01:28:50 cloudron ntpd[23144]: Soliciting pool server 4.53.160.75
Jan 05 01:34:27 cloudron ntpd[23144]: kernel reports TIME_ERROR: 0x41: Clock Unsynchronized

ntptime also returns some strange errors:

root@cloudron:~# ntptime
ntp_gettime() returns code 5 (ERROR)
  time e39e4293.b8e9e5e4  Tue, Jan  5 2021  1:42:43.722, (.722319846),
  maximum error 16000000 us, estimated error 16000000 us, TAI offset 37
ntp_adjtime() returns code 5 (ERROR)
  modes 0x0 (),
  offset 0.000 us, frequency 0.000 ppm, interval 1 s,
  maximum error 16000000 us, estimated error 16000000 us,
  status 0x2041 (PLL,UNSYNC,NANO),
  time constant 3, precision 0.001 us, tolerance 500 ppm,

[girish]

@tx-hermit Ah, here's the explanation - https://serverfault.com/questions/1024770/ubuntu-20-04-time-sync-problems-and-possibly-incorrect-status-information . Uninstalling/disabling ntp service does the trick! I will make a fix for next release.