Concurrent execution of DietPi-Run_NTPD detected

[Sopor]

I'm running nginx on DietPi but all the log files suddenly change to be 0 bytes. I also noticed that all the log files for the system also are 0 bytes.

I then run dietpi-software (to change the log issue), but it did have some problems with the NTP (NTPD: Waiting for completion of systemd-timesyncd (1/60) and it will try 60 times) so i follow the instructions, but now i get this after selecting my gateway as an NTP server:

Could you tell me how i make the log files to be persistent? This seems to not be an easy task to do :(
I was trying to follow your instructions here

[Sopor]

I manage to change it on another DietPi and when it work as it should it wasn't so hard :)

[MichaIng]

@Sopor
Many thanks for your report.

Jep logfiles are cleared every every hour due to RAMlog, otherwise RAM would fill up unlimited.
So you need to disable/uninstall DietPi-RAMlog via dietpi-software. Choose "Custom", as rsyslog is actually a very obsolete thing on modern Debian. systemd-journald is active anyway which stores all system logs well sortable and readable via journalctl in RAM. rsyslog just doubles this, making all syslogs a more-than-double effort.
I will change this to remove the rsyslog option completely and instead allow to make journald boot persistent, which is quite easy as well.
EDIT: Ah yeah RAMlog option 2 of course is an alternative as well, /var/log/ files will still be cleared hourly but content stored to disk for later review if required.

About the NTPD error:
Jep I think I know what causes this loop:

• When you run into a time sync error, you can open dietpi-config to change settings.
• But when you change time sync settings, the same script is called again.
• What we need to do is from within dietpi-config, check if the script already runs and in case skip the additional call.

But in your particular case you managed to solve everything now?

[Sopor]

I still not manage to get it to work

[ INFO ] DietPi-Run_NTPD | NTPD: Waiting for completion of systemd-timesyncd (10/10)

I run pfsense as a router and i have an NTP server running... I'm not sure what it want?

It doesn't matter what i try to add to mirror it won't work. I even tried ntp.stupi.se but no success.

[MichaIng]

@Sopor
Please paste: journalctl -u systemd-timesyncd

[Sopor]

$ sudo journalctl -u systemd-timesyncd
-- Logs begin at Thu 2016-11-03 18:16:42 CET, end at Sat 2019-07-27 14:23:13 CEST. --
Jun 08 00:17:14 DietPi systemd[1]: Starting Network Time Synchronization...
Jun 08 00:17:14 DietPi systemd[1]: Started Network Time Synchronization.
Jun 08 00:23:23 DietPi systemd-timesyncd[662]: Synchronized to time server 192.36.143.130:123 (0.debian.pool.ntp.org).
Jun 08 00:23:24 DietPi systemd[1]: Stopping Network Time Synchronization...
Jun 08 00:23:24 DietPi systemd[1]: Stopped Network Time Synchronization.

[MichaIng]

@Sopor

Jun 08 00:17:14 DietPi systemd[1]: Started Network Time Synchronization.
Jun 08 00:23:23 DietPi systemd-timesyncd[662]: Synchronized to time server 192.36.143.130:123 (0.debian.pool.ntp.org).

It looks like it took 6 minutes to sync time...
It should be ~1 second.
It syncs with 0.debian.pool.ntp.org currently, which is default on Debian but not optional. "Gateway" (local router) should be best in very most cases.

All instances of run_ntpd are killed, right? killall -w run_ntpd

and i have an NTP server running

What you mean by this? An NTP server in the router of on the DietPi machine? Or the "ntp" package installed?

[Sopor]

Instead of syncing to time.windows.com i sync my computers to my pfsense.

[MichaIng]

@Sopor
Yeah this makes sense for the DietPi system as well. When you choose "Gateway" in dietpi-config > Network Options: Misc > NTP mirror it should do this, if the router is the default internet gateway for the network.
Switching this and then running journalctl -u systemd-timesyncd should show:

systemd-timesyncd[662]: Synchronized to time server <ip_of_your_router>

as one of the last lines then.

[MichaIng]

PR up: #3095
Some other enhancements are and need to be done with this, however concurrent execution from both sides are resolved.

[codebeat-nl]

@MichaIng Thanks! journalctl -u systemd-timesyncd did the trick.

[MichaIng]

journalctl -u systemd-timesyncd only prints the services logs but does not change something 😉. So I guess the time it took to search and run the command was the solution, as the concurrent script finished in the meantime 🙂.