From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <cfec672c163a4af9d0e91c7f002bbc8e@quanstro.net>
From: erik quanstrom <quanstro@quanstro.net>
Date: Fri, 26 Jun 2009 09:41:58 -0400
To: 9fans@9fans.net
In-Reply-To: <a560a5d00906260453m5c198ee8ofa0b157a1339185@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 8bit
Subject: Re: [9fans] synchronization of time, setting time
Topicbox-Message-UUID: 0f8fb3c0-ead5-11e9-9d60-3106f5b1d025

> 1) If I want to set the hardware (on-board) clock by hand, how can I?

date -n>/dev/rtc

> 2) If I want to synchronize the hardware time with a ntp server (once
> / periodically), how can I?

modify timesync.

> 3) If I run the 'timesync -n [ntp server]' command, how is the
> frequency of synchronization determined?

by the error in the measurement.  here's the output from my
log.  looks like its running about once a minute.

chula Jun 26 09:34:30 δ -1049724 avgδ 652049 hz 2660094075
chula Jun 26 09:35:30 δ -1233447 avgδ 942747 hz 2660093874

i can think of two reasons for bad measurements — specifying
an accuracy that is not attainable or running on a machine
with an unsyncronized tsc.  for example, certain pentium iv.

- erik