RE: [chrony-users] Chrony offset and stability adjustments?

[ Thread Index | Date Index | More chrony.tuxfamily.org/chrony-users Archives ]


Unfortunately root dispersion is a lousy way of determining the accuracy. It
is sort a maximum uncertainty but it could be much less than that, or it could
be greater even.
To figure out what the accuracy is, you need a better clock. Buy a GPS-- it
need not be anything expensive but must hav a PPS line. Hook that up to your
system as a source and compare your sources to that. Note that a VM is liable
to have horrible time excursions. You should be disciplining the underlying
hardware (the OS running the VM) and read that sytem clock and not the VM
clock. The computer can swap out your VM and it stays out for an indeterminate
amount of time, and when it is swapped back in, it has not idea how much time
has gone by. It is like using a pendulum clock on a ship.


On Mon, 7 Jan 2019, LeBlanc, Daniel James wrote:

Hi Miroslav.

The requirement that we are trying to meet is " Time accuracy must be within 1 ms of true time.", as described in Section 5.18 of https://www.nena.org/resource/resmgr/standards/NENA-STA-010.2_i3_Architectu.pdf.

I am currently connected to the NRC Internet-accessible time servers from our lab environment.  I am so close to meeting the spec (~1.3ms reported root dispersion) that I thought I would reach out to the experts to see if there was some adjustment that I might try.  :-)

I am running chrony within a virtual environment (VM) so have disabled rtcsync and enabled rtcfile within the .conf file.  I will look at reducing the maxclockerror and increasing the minsamples setting.  Any additional items to consider would be appreciated!

Thanks.

Daniel J. LeBlanc, P.Eng., MBA, DTME | Senior Network Architect | Bell Canada


-----Original Message-----
From: Miroslav Lichvar [mailto:mlichvar@xxxxxxxxxx]
Sent: January-07-19 10:47 AM
To: chrony-users@xxxxxxxxxxxxxxxxxxxx
Subject: Re: [chrony-users] Chrony offset and stability adjustments?

On Mon, Jan 07, 2019 at 02:32:00PM +0000, LeBlanc, Daniel James wrote:
Thanks for your quick reply.  I have done a bit of searching and have not been able to find anything regarding how we can improve the root dispersion (as you have described).  Based on the description of how root dispersion is calculated (https://serverfault.com/questions/768280/what-is-ntp-dispersion-and-how-do-i-control-it - see ladder diagrams and associated description half way down the page), even reducing the polling interval on the local server will not improve the calculated root dispersion amount.  Is there anything that can be done short of using a closer and lower stratum NTP server as a source to further reduce the root dispersion?

If the computer had a very stable clock, you could set the
maxclockerror option to a smaller value and possibly also increase the
minsamples option to force chronyd to keep more samples and reduce the
uncertainty in the estimated frequency.

But that is unlikely to be the case in a VM.

What problem are you trying to solve? A root dispersion of a few
milliseconds is usually fine.

--
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.


--
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.


Mail converted by MHonArc 2.6.19+ http://listengine.tuxfamily.org/