Olympus-OM
[Top] [All Lists]

Re: [OM] Somewhat OT: Computer Clock Management

Subject: Re: [OM] Somewhat OT: Computer Clock Management
From: Jez Cunningham <jez@xxxxxxxxxxxxxx>
Date: Sun, 17 Aug 2014 11:24:55 +0100
Good info Scott!   I've bookmarked you as a trusted source ;-)
ex-IT Jez (pursued by a bear?)


On 17 August 2014 03:25, Scott Gomez <sgomez.baja@xxxxxxxxx> wrote:

> Best practice is to get time sync from the nearest time server to you, and
> not query the "source" directly for it. Time servers have a hierarchy, like
> DNS servers, because of loading issues, and as turns out, most of the major
> ISPs seem to provide this service from their DNS servers. Then you set one
> machine in your network up to query the time servers directly, on fairly
> long interval, and set all other machines on your network to query your
> configured master machine. That's probably the best method for the average
> person wanting accurate time service. Another way to do it would be to set
> a network switch, or your router or firewall to be the master time server
> for your home network, and point all other devices to it for time service,
> where possible.
>
> Note that upper level time servers can and will block connection from
> devices they determine to a) poll too often, b) not be sufficiently high in
> the hierarchy to poll high-level servers directly, or c) be sending
> malformed queries. Whether you know if you've been blocked is entirely up
> to you to determine, which, on Windows machines, often meant not knowing at
> all until you started getting bizarre behavior on downstream machines due
> to being too far out of sync.
>
> Microsoft run a time server for the purpose. There are also ntp pool
> servers that can be queried. See the (many!) notes and instructions
> available via searching on "network time protocol" or "network time
> service", but keep in mind that when it comes to use of critical internet
> services like ntp, adherence to the RFCs are usually the "law" as far as
> responsible system administrators are concerned. Relevant RFCs are numbers
> 958, 1305 (version 3) and 5905 (version 4). Unlike application programmers
> (who have quite accurately been referred to by Chris Trask as "cyber
> toddlers") responsible for the user-facing software, Syadmins (like myself,
> for instance) are a famously cranky lot, and tenacious in their pursuit of
> up-time. See XKCD, here:
>
> http://www.xkcd.com/705/
>
> or the BOFH series:
>
> http://bofh.ntk.net/BOFH/
>
> In our defence, I will say that we come by our reputation honestly.
>
> Microsoft's note on use of Windows Time Service (which has been provided as
> part of Windows since the XP days):
>
>  http://support.microsoft.com/kb/307897
>
> Properly set up, it's accurate enough for most uses (including sniping bids
> on Ebay, which I often used to do). It has been a long time since I delved
> further into it, but I believe some versions (probably newer than XP?) even
> have a means of compensating for drift rate, which is usually the largest
> error factor in home PCs, which aren't notoriously stable when it comes to
> time. NOTE: there are significant differences in setup between "flavors"
> such as XP Home and XP Professional. I would assume that that's true as
> well for WinVista, Win7, and Win8.
>
> Linux machines (and by extension, Macs, which use a very similar method)
> are quite easy to configure to get time sync from a network source,
> unless/until one needs extremely high accuracy. As far as I know, both have
> a check box where it can be selected, and from then on it's usually
> "automagic".
>
> ---
> Scott
>
>
> On Fri, Aug 15, 2014 at 11:02 AM, Chris Barker <ftog@xxxxxxxxxxxxxx>
> wrote:
>
> > Thanks for more useful information, Ken.
> >
> > But you needn't worry about us in the UK: we get all our time from Big
> > Ben, which is always on time.
> >
> > :-)
> >
> > Chris
> >
> > On 15 Aug 14, at 16:37, Ken Norton <ken@xxxxxxxxxxx> wrote:
> >
> > > Like seriously!  Take a frequency band meant for Space-to-ground
> > > transmission, immediately next to the GPS band (where the signal
> > > strength is somewhere just above the universe's noise floor) and
> > > instead repurpose it to ground-to-ground transmittions with high power
> > > transmitters. It wasn't just cheap, inexpensive unfiltered GPS units
> > > that suffered. During the tests, it was wiping out everything and the
> > > GPS BITS clock receivers were getting completely toasted. But that
> > > didn't influence the regulators. It took the threat of airplanes
> > > falling out of the sky before anybody would listen. We faced
> > > opposition that was mind-boggling.
> > >
> > > In almost every case, the NTP sites you are getting your clock updates
> > > from are primary timed from a GPS timed BITS Clock system.
> >
> > --
> > _________________________________________________________________
> > Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
> > Archives: http://lists.thomasclausen.net/mailman/private/olympus/
> > Themed Olympus Photo Exhibition: http://www.tope.nl/
> >
> >
> --
> _________________________________________________________________
> Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
> Archives: http://lists.thomasclausen.net/mailman/private/olympus/
> Themed Olympus Photo Exhibition: http://www.tope.nl/
>
>
-- 
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/

<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz