Hi,

I have a tru64 machine which always boots with wrong time(time is
always in future ie. 2007 or above).   After booting, ntp starts and
syncs time properly with the server and time is back to normal. Once
i reboot problem appears again. Is this a problem with a TOY chip or do
we
need to explicity set the time to TOY chip.  I believe ntp calls
settimeofday
which syncs the time to TOY chip.   But i donot see any errors
in the log file with respect to TOY chip battery failure.

Thanks,
Balaji

Hi,

sbalaji79@gmail.com wrote:
> Hi,
> 
> I have a tru64 machine which always boots with wrong time(time is
> always in future ie. 2007 or above).   After booting, ntp starts and
> syncs time properly with the server and time is back to normal. Once
> i reboot problem appears again. Is this a problem with a TOY chip or do
> we
> need to explicity set the time to TOY chip.  I believe ntp calls
> settimeofday
> which syncs the time to TOY chip.   But i donot see any errors
> in the log file with respect to TOY chip battery failure.

I'm pretty sure the machine's hardware clock (RTC) has the wrong date. At
boot time the operating system time is initialized with that time.

I don't know tru64, but settimeofday must not necessarily set the machine's
RTC. For example under linux there's a special command which writes the
time of the software clock to the RTC. Sometimes this is called
periodically every few minutes, and im most cases it is run when the system
shuts down.

You must check which command tru64 provides to do so.

Martin
-- 
Martin Burnicki

Meinberg Funkuhren
Bad Pyrmont
Germany

Martin Burnicki wrote:
> Hi,
> 
> sbalaji79@gmail.com wrote:
>> Hi,
>>
>> I have a tru64 machine which always boots with wrong time(time is
>> always in future ie. 2007 or above).   After booting, ntp starts and
>> syncs time properly with the server and time is back to normal. Once
>> i reboot problem appears again. Is this a problem with a TOY chip or do
>> we
>> need to explicity set the time to TOY chip.  I believe ntp calls
>> settimeofday
>> which syncs the time to TOY chip.   But i donot see any errors
>> in the log file with respect to TOY chip battery failure.
> 
> I'm pretty sure the machine's hardware clock (RTC) has the wrong date. At
> boot time the operating system time is initialized with that time.
> 
> I don't know tru64, but settimeofday must not necessarily set the machine's
> RTC. For example under linux there's a special command which writes the
> time of the software clock to the RTC. Sometimes this is called
> periodically every few minutes, and im most cases it is run when the system
> shuts down.
> 
> You must check which command tru64 provides to do so.
> 
> Martin

Agreed. VMS had this problem every year in January where if it hadn't
been booted for a very long time the date would be wrong on the next
boot. The date was not being written back in any reliable way so the
date ended up way off. I don't know if the Alphas suffered the same
problem, this was on VAXen.

Danny
_______________________________________________
questions mailing list
questions@lists.ntp.isc.org
https://lists.ntp.isc.org/mailman/listinfo/questions

Danny Mayer wrote:

> Martin Burnicki wrote:
> 
>>Hi,
>>
>>sbalaji79@gmail.com wrote:
>>
>>>Hi,
>>>
>>>I have a tru64 machine which always boots with wrong time(time is
>>>always in future ie. 2007 or above).   After booting, ntp starts and
>>>syncs time properly with the server and time is back to normal. Once
>>>i reboot problem appears again. Is this a problem with a TOY chip or do
>>>we
>>>need to explicity set the time to TOY chip.  I believe ntp calls
>>>settimeofday
<snip>
>>I don't know tru64, but settimeofday must not necessarily set the machine's
>>RTC. For example under linux there's a special command which writes the
>>time of the software clock to the RTC. Sometimes this is called
>>periodically every few minutes, and im most cases it is run when the system
>>shuts down.
>>
>>You must check which command tru64 provides to do so.
>>
>>Martin
> 
> 
> Agreed. VMS had this problem every year in January where if it hadn't
> been booted for a very long time the date would be wrong on the next
> boot. The date was not being written back in any reliable way so the
> date ended up way off. I don't know if the Alphas suffered the same
> problem, this was on VAXen.
> 

The VAX and Alpha TOY clocks do not keep the year itself.  VMS stores 
the current year on disk.  I believe that it was actually a part of the 
executive image.  VMS updates this information every time a SET TIME is 
done.  Don't know what Tru64 does; I had the admin course for it about 
ten years ago and spent maybe a total of three days actually working 
with it.

Since a lot of VMS engineers worked on Tru64, it's possible that the 
solution is to set the time with the "date" command.

Richard B. Gilbert wrote:
> Danny Mayer wrote:
> 
>> Martin Burnicki wrote:
>>
>>> Hi,
>>>
>>> sbalaji79@gmail.com wrote:
>>>
>>>> Hi,
>>>>
>>>> I have a tru64 machine which always boots with wrong time(time is
>>>> always in future ie. 2007 or above).   After booting, ntp starts and
>>>> syncs time properly with the server and time is back to normal. Once
>>>> i reboot problem appears again. Is this a problem with a TOY chip or do
>>>> we
>>>> need to explicity set the time to TOY chip.  I believe ntp calls
>>>> settimeofday
> <snip>
>>> I don't know tru64, but settimeofday must not necessarily set the
>>> machine's
>>> RTC. For example under linux there's a special command which writes the
>>> time of the software clock to the RTC. Sometimes this is called
>>> periodically every few minutes, and im most cases it is run when the
>>> system
>>> shuts down.
>>>
>>> You must check which command tru64 provides to do so.
>>>
>>> Martin
>>
>>
>> Agreed. VMS had this problem every year in January where if it hadn't
>> been booted for a very long time the date would be wrong on the next
>> boot. The date was not being written back in any reliable way so the
>> date ended up way off. I don't know if the Alphas suffered the same
>> problem, this was on VAXen.
>>
> 
> The VAX and Alpha TOY clocks do not keep the year itself.  VMS stores
> the current year on disk.  I believe that it was actually a part of the
> executive image.  VMS updates this information every time a SET TIME is
> done. 

Right. The trouble was you rarely needed to reboot a VMS system so the
SET TIME was almost never done. It was a problem that happened every
January for reasons that I don't recall. I believe a SET TIME was done
in the SYS$SHUTDOWN code but it's been years since I looked at it and my
VAXstation is currently not running...

 Don't know what Tru64 does; I had the admin course for it about
> ten years ago and spent maybe a total of three days actually working
> with it.
> 

More than I got. I was handed a CD, the installation manual and an
AlphaStation and told to get it up and running for my new boss who was
coming the following week.

> Since a lot of VMS engineers worked on Tru64, it's possible that the
> solution is to set the time with the "date" command.

Not as many as you might think. VMS was going hot and heavy and
migrating the VAX base over to Alpha was the top priority. Most of the
OSF/1 folks (as it was then called) came from the MIPS Ultrix group and
additional people were hired.

Danny

_______________________________________________
questions mailing list
questions@lists.ntp.isc.org
https://lists.ntp.isc.org/mailman/listinfo/questions