You need to provide the -h parameter to chronyc.
E.g. chronyc -h 1.2.3.4 sources
Or: chronyc -h 1.2.3.4 tracking
(when the chronyd you want to watch is on a system with IP address
1.2.3.4)
Again, you DO NOT need to run the daemon on the system (pod) where
you run chronyc.
You only need to provide IP connectivity between the two, and to use
"cmdallow" in the
chronyd configuration file(s) to allow the IP address where chronyc
is used.
Example: chronyc -h 44.137.72.10 tracking
Reference ID : 50505300 (PPS)
Stratum : 1
Ref time (UTC) : Tue Aug 05 09:27:29 2025
System time : 0.000000054 seconds slow of NTP time
Last offset : +0.000000108 seconds
RMS offset : 0.000001205 seconds
Frequency : 10.403 ppm fast
Residual freq : -0.000 ppm
Skew : 0.025 ppm
Root delay : 0.000000001 seconds
Root dispersion : 0.000025486 seconds
Update interval : 16.0 seconds
Leap status : Normal
Rob
On 2025-08-05 10:22, Remush wrote:
I'll try to elaborate more about what I'm trying
to achieve.
First of all, I'm very grateful for the detailed
and well-explained comments you made.
My objective is to monitor a drift between 3 NTP
servers.
`chronyc sources` provides a wonderful and simple
output just for that purpose.
The issue I'm having is the following:
I'm using Openshift as my base, and my solution
was to simply set a pod with a chrony image and run the
command,
sadly to my understanding, openshift pods doesn't
have access to systemd, meaning I can't set a chronyd daemon
on the Pod it self.
"chrony tracking" also returns the error message
"506 can't talk with daemon".
Honestly I really think that because of that
Openshift limitation I won't find any way to actually run
"chronyc sources".
I'm not completely understanding what it means to
run `chronyd -U -x`.
I actually run it successfully, and it created a
chronyd.pid, but "chronyc sources" still return "506 Cannot
talk to daemon"
Am I missing something?
My guess was that he was attempting monitoring of an
existing chronyd outside
his container.
I have done that (way) in the past to monitor time service
using nagios.
I just installed (copied) the chronyc binary to the
monitoring system, which
itself was running ntpd, and made a check_chrony script that
did a
chronyc call (from a perl skeleton available for nagios).
Probably not the most efficient way, but it works. I used
"chronyc tracking",
of course with a -h parameter.
But maybe I am completely wrong guessing his objectives...
it is not very
clear from the explanation.
Rob
On 2025-08-05 09:20, Mingye Wang wrote:
So uh Rob gave a good explanation of how chrony works, but honestly:
this smells like an "XY problem" to me. What are you attempting to do,
actually, by getting `chronyc sources`? In other words, what
information do you *really* need?