CMOS (non-volatile harware) clock sycronisation

I have a number of Windows member servers in an AD forrest and in steady
state w32time keeps there clocks approximately right (+-200ms). However when
they are rebooted their clocks are (for a short time) all over the place (for
10-30 minutes) because, it would appear, that nothng is keeping the CMOS
(non-volatile harware) clock in sycronisation with the Windows clock.

On Linux I'd just create a cron job to do "clock -uw" periodically but I
can't find any equivalent on Windows.

Please note: I'm asking how to get Windows to keep the *CMOS* clock synced
so that when servers are rebooted their clocks start out *approximately*
right *before* w32time manages to syncronise to a time server.

Pegasus
Tue Apr 29 08:09:47 PDT 2008


"Brian McCauley" <my-name-with-a-dot-in-it@uhb.nhs.uk.invalid> wrote in
message news:E7523F1F-9CBF-44A0-9DFE-71C85C8E5F49@microsoft.com...
>I have a number of Windows member servers in an AD forrest and in steady
> state w32time keeps there clocks approximately right (+-200ms). However
> when
> they are rebooted their clocks are (for a short time) all over the place
> (for
> 10-30 minutes) because, it would appear, that nothng is keeping the CMOS
> (non-volatile harware) clock in sycronisation with the Windows clock.
>
> On Linux I'd just create a cron job to do "clock -uw" periodically but I
> can't find any equivalent on Windows.
>
> Please note: I'm asking how to get Windows to keep the *CMOS* clock synced
> so that when servers are rebooted their clocks start out *approximately*
> right *before* w32time manages to syncronise to a time server.

Sounds like you need to replace a few motherboard batteries. They
cause the time to go out while your machines are turned off.

my-name-with-a-dot-in-it
Tue Apr 29 08:46:01 PDT 2008

"Pegasus (MVP)" wrote:

> "Brian McCauley" <my-name-with-a-dot-in-it@uhb.nhs.uk.invalid> wrote in

> > .... rebooted their clocks are (for a short time) all over the place
> > (for 10-30 minutes) because, it would appear, that nothng is keeping the CMOS
> > (non-volatile harware) clock in sycronisation with the Windows clock.
>
> Sounds like you need to replace a few motherboard batteries. They
> cause the time to go out while your machines are turned off.

This happens on serveral servers that are only a few years old and the
effect is not dependant on the machines being turned off so I think I can
rule out the batteries as being the cause.

stephen
Tue Jun 10 04:49:26 PDT 2008

On Apr 29, 4:46=A0pm, Brian McCauley <my-name-with-a-dot-
in...@uhb.nhs.uk.invalid> wrote:
> "Pegasus (MVP)" wrote:
> > "Brian McCauley" <my-name-with-a-dot-in...@uhb.nhs.uk.invalid> wrote in
> > > =A0.... rebooted their clocks are (for a short time) all over the plac=
e
> > > (for 10-30 minutes) because, it would appear, that nothng is keeping t=
he CMOS
> > > (non-volatile harware) clock in sycronisation with the Windows clock.
>
> > Sounds like you need to replace a few motherboard batteries. They
> > cause the time to go out while your machines are turned off.
>
> This happens on serveral servers that are only a few years old and the
> effect is not dependant on the machines being turned off so I think I can
> rule out the batteries as being the cause.

I have a similar problem with Compaq/HP server hardware that is
working according to specification - The hardware clock drifts at
approximately 1 second a day if uncorrectrd. Using Windows 2000 I
synchronise to an NTP server, but have specified the maximum time
correction allowed as 60 seconds, When the server is rebooted the time
is corrected, until it has been running long enough for the time
correction required is greater than 60 seconds, at which time it
requires manual correction to allow it to synchronise. It appears
that the CMOS time is not adjusted to the corrected system time by the
normal operation of W32TIME.