Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 02 Mar 2005 19:17:37 -0600
From:      Kevin Kinsey <kdk@daleco.biz>
To:        LukeD@pobox.com
Cc:        Tom Trelvik <ttt@cwru.edu>
Subject:   Re: Received mail timestamp is off by 7 hours
Message-ID:  <422665B1.7040807@daleco.biz>
In-Reply-To: <Pine.NEB.4.61.0503021420230.15178@norge.freeshell.org>
References:  <20050302102908.GF30896@alzatex.com> <LOBBIFDAGNMAMLGJJCKNCEKCFAAA.tedm@toybox.placo.com> <Pine.NEB.4.61.0503021253040.11146@norge.freeshell.org> <42262D74.3050907@cwru.edu> <Pine.NEB.4.61.0503021420230.15178@norge.freeshell.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Luke wrote:

>>     That suggests larger problems on your system, to me, but I dunno.
>
> 3) If your clock tends to run noticably fast or slow, constant NTP 
> corrections
> tend to do more harm than good, at least in my experience.  It got to 
> where
> I couldn't even run a buildworld because NTP kept tinkering with the 
> clock
> in the middle of the process.
>
> You're right.  This machine did have serious problems.  The clock was 
> wild. 
> Using the NTP daemon to try to correct it just aggravated the situation
> because calibration was just about impossible.


This occurs on some machines (older ones in my experience) because FreeBSD
(I guess) selects the wrong type of hardware timecounter for the CPU 
it's on. 

Now, it's likely that it's really the hardware's fault, but changing the 
sysctl
"kern.timecounter.hardware" to another value (`sysctl 
kern.timecounter.choice`
will give you options) has worked for me on at least two different 
occasions
when we faced this problem.

IIRC, one was an old AMD K6-2/475 on an Asus P5-A mobo; the other was an
original Pentium, but I don't know the mobo model off the top of my head.

HTH,

Kevin Kinsey



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?422665B1.7040807>