RE: [chrony-users] Chronyc Call to Add Refclock

["Chang, Benjamin" <Benjamin.Chang@xxxxxxxxxxxxxxx>]

I am just calling makestep once after so it's essentially:
Systemctl restart chronyd
Chronyc delete <ip address>
Chronyc add server <ip address> minpoll -6 maxpoll -6 offset -54321
Chronyc makestep

And it then becomes 54321 seconds ago

>Is it synchronizing to NTP when it ignores PPS?
Yes.

>I'd expect the sources to be marked both as falsetickers first. Then PPS will stop accepting new samples, after some time it will select NTP, which will allow PPS to collect new measurements again and then it can be selected..
That's what I figured at some point, but it never really fully ever syncs to the PPS after. The LastRX goes up to the 100s and then may reset once and then go back up, or it'll be correct but never sync to it (i.e. go from 4-12s before LastRX reset which is right since im at poll 3).

>How large is the "+/-" value in the sources output? Maybe there are old samples mixed with new samples, causing a very large dispersion.
Here is an example one of my chronyc sources calls:

MS Name/IP address         Stratum Poll Reach LastRx Last sample               
===============================================================================
#x NTPS                          0   3   377     6    +60us[  +60us] +/- 1920ns
^* 192.168.1.200                 1  -6   377     1  +2660ns[+2710ns] +/- 5398ns

I will investigate/delve deeper with the debug option.

-----Original Message-----
From: Miroslav Lichvar <mlichvar@xxxxxxxxxx> 
Sent: Monday, December 7, 2020 9:37 AM
To: chrony-users@xxxxxxxxxxxxxxxxxxxx
Subject: Re: [chrony-users] Chronyc Call to Add Refclock

On Mon, Dec 07, 2020 at 02:18:45PM +0000, Chang, Benjamin wrote:
> I am essentially using the offset to be like the settime call (so by hours, days, etc). Its easier/more consistent to use the offset to change the date/time across multiple systems than using settime due to one being a constant while the other potentially having a delay depending on when the other systems receive the call. 

That's an interesting use case. After readding the NTP source do you issue chronyc makestep or do you have an unlimited makestep allowed?

Is it synchronizing to NTP when it ignores PPS?

I'd expect the sources to be marked both as falsetickers first. Then PPS will stop accepting new samples, after some time it will select NTP, which will allow PPS to collect new measurements again and then it can be selected.

> > That indicates the offset between the NTP source and PPS source is too large.
> 
> Interesting because if I have the offset established in the chrony file, it will sync to the PPS after with the time/day offset. For examples, I am in EST, I currently have an offset of -54321 in the .conf file and upon a systemctl restart chronyd, it is now that far behind (Sunday), synced to my PPS with around 0-10us measured offset. 
> So if I somehow change the offset prior to the first PPS after I reset chronyd, then that should work maybe?
> 
> Even when it syncs up occasionally, the measured chronyc sources offset is only about 50 us which should be synchronize-able no?

How large is the "+/-" value in the sources output? Maybe there are old samples mixed with new samples, causing a very large dispersion.

The best way to see what's going on would be to get the debug output from chronyd (when compiled with the --enable-debug option).

--
Miroslav Lichvar


-- 
To unsubscribe email chrony-users-request@xxxxxxxxxxxxxxxxxxxx 
with "unsubscribe" in the subject.
For help email chrony-users-request@xxxxxxxxxxxxxxxxxxxx 
with "help" in the subject.
Trouble?  Email listmaster@xxxxxxxxxxxxxxxxxxxx.


--
To unsubscribe email chrony-users-request@xxxxxxxxxxxxxxxxxxxx
with "unsubscribe" in the subject.
For help email chrony-users-request@xxxxxxxxxxxxxxxxxxxx
with "help" in the subject.
Trouble?  Email listmaster@xxxxxxxxxxxxxxxxxxxx.