[Cog] Funny thing about time

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

[Cog] Funny thing about time

Igor Stasenko
 
Yesterday on my comp two separate running images are hung
simultaneously, while i were doing something completely outside of
them.

And now during reading  the sqUnixHeartBeat.c:


        /* The native clock may go backwards, e.g. due to NTP adjustments, although
         * why it can't avoid small backward steps itself, I don't know.  Simply
         * ignore backward steps and wait until the clock catches up again.  Of
         * course this will cause problems if the clock is manually adjusted.  To
         * which the doctor says, "don't do that".
         */

i start recalling that yesterday i adjusted clock back by 1 hour,
because after installing a windoze in dual boot it was set it wrongly.

But of course it could be anything else :)

--
Best regards,
Igor Stasenko AKA sig.
Reply | Threaded
Open this post in threaded view
|

Re: [Cog] Funny thing about time

Chris Muller-3

Every time I jump back and forth from Windows to Linux on my primary
laptop, the clock is wrong...

On Tue, Jan 11, 2011 at 4:37 PM, Igor Stasenko <[hidden email]> wrote:

>
> Yesterday on my comp two separate running images are hung
> simultaneously, while i were doing something completely outside of
> them.
>
> And now during reading  the sqUnixHeartBeat.c:
>
>
>        /* The native clock may go backwards, e.g. due to NTP adjustments, although
>         * why it can't avoid small backward steps itself, I don't know.  Simply
>         * ignore backward steps and wait until the clock catches up again.  Of
>         * course this will cause problems if the clock is manually adjusted.  To
>         * which the doctor says, "don't do that".
>         */
>
> i start recalling that yesterday i adjusted clock back by 1 hour,
> because after installing a windoze in dual boot it was set it wrongly.
>
> But of course it could be anything else :)
>
> --
> Best regards,
> Igor Stasenko AKA sig.
>
Reply | Threaded
Open this post in threaded view
|

Re: [Cog] Funny thing about time

johnmci

Some images grind logic thru the locale plugin to set the current time/date based on UTC & locale's viewpoint of time zone.
You should peer deeper to understand exactly who/where the printed time is coming from...

On 2011-01-11, at 5:18 PM, Chris Muller wrote:

>
> Every time I jump back and forth from Windows to Linux on my primary
> laptop, the clock is wrong...
>
> On Tue, Jan 11, 2011 at 4:37 PM, Igor Stasenko <[hidden email]> wrote:
>>
>> Yesterday on my comp two separate running images are hung
>> simultaneously, while i were doing something completely outside of
>> them.
>>
>> And now during reading  the sqUnixHeartBeat.c:
>>
>>
>>        /* The native clock may go backwards, e.g. due to NTP adjustments, although
>>         * why it can't avoid small backward steps itself, I don't know.  Simply
>>         * ignore backward steps and wait until the clock catches up again.  Of
>>         * course this will cause problems if the clock is manually adjusted.  To
>>         * which the doctor says, "don't do that".
>>         */
>>
>> i start recalling that yesterday i adjusted clock back by 1 hour,
>> because after installing a windoze in dual boot it was set it wrongly.
>>
>> But of course it could be anything else :)
>>
>> --
>> Best regards,
>> Igor Stasenko AKA sig.
>>

--
===========================================================================
John M. McIntosh <[hidden email]>   Twitter:  squeaker68882
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
===========================================================================




Reply | Threaded
Open this post in threaded view
|

Re: [Cog] Funny thing about time

Frank Shearar
In reply to this post by Chris Muller-3
 
Yes, Windows like to set the CMOS clock to local time, and Linux likes
to use UTC.

My brother recently ran into a similar problem with a FreeBSD machine
running as a guest inside a Windows host. He had to touch
/etc/wall_cmos_clock to fix it.

 From the adjkerntz man page:

If the file /etc/wall_cmos_clock exists, it means that CMOS clock keeps
local time (MS-DOS and MS-Windows compatible mode).  If that file does
not exist, it means that the CMOS clock keeps UTC time.

Maybe there's a similar Linux/Mac fix?

frank

On 2011/01/12 01:18, Chris Muller wrote:

>
> Every time I jump back and forth from Windows to Linux on my primary
> laptop, the clock is wrong...
>
> On Tue, Jan 11, 2011 at 4:37 PM, Igor Stasenko<[hidden email]>  wrote:
>>
>> Yesterday on my comp two separate running images are hung
>> simultaneously, while i were doing something completely outside of
>> them.
>>
>> And now during reading  the sqUnixHeartBeat.c:
>>
>>
>>         /* The native clock may go backwards, e.g. due to NTP adjustments, although
>>          * why it can't avoid small backward steps itself, I don't know.  Simply
>>          * ignore backward steps and wait until the clock catches up again.  Of
>>          * course this will cause problems if the clock is manually adjusted.  To
>>          * which the doctor says, "don't do that".
>>          */
>>
>> i start recalling that yesterday i adjusted clock back by 1 hour,
>> because after installing a windoze in dual boot it was set it wrongly.
>>
>> But of course it could be anything else :)
>>
>> --
>> Best regards,
>> Igor Stasenko AKA sig.
>>
>
>

Reply | Threaded
Open this post in threaded view
|

Re: [Cog] Funny thing about time

Levente Uzonyi-2
 
On Wed, 12 Jan 2011, Frank Shearar wrote:

>
> Yes, Windows like to set the CMOS clock to local time, and Linux likes to use
> UTC.
>
> My brother recently ran into a similar problem with a FreeBSD machine running
> as a guest inside a Windows host. He had to touch /etc/wall_cmos_clock to fix
> it.
>
> From the adjkerntz man page:
>
> If the file /etc/wall_cmos_clock exists, it means that CMOS clock keeps local
> time (MS-DOS and MS-Windows compatible mode).  If that file does not exist,
> it means that the CMOS clock keeps UTC time.
>
> Maybe there's a similar Linux/Mac fix?

Yes, there is, but different linux distributions have different files
IIRC. Btw, if you tell linux to assume that the hardware clock stores the
local time, you'll still have problems after daylight saving related time
changes, because both OSs will adjust the clock.


Levente

>
> frank
>
> On 2011/01/12 01:18, Chris Muller wrote:
>>
>> Every time I jump back and forth from Windows to Linux on my primary
>> laptop, the clock is wrong...
>>
>> On Tue, Jan 11, 2011 at 4:37 PM, Igor Stasenko<[hidden email]>  wrote:
>>>
>>> Yesterday on my comp two separate running images are hung
>>> simultaneously, while i were doing something completely outside of
>>> them.
>>>
>>> And now during reading  the sqUnixHeartBeat.c:
>>>
>>>
>>>         /* The native clock may go backwards, e.g. due to NTP adjustments,
>>> although
>>>          * why it can't avoid small backward steps itself, I don't know.
>>> Simply
>>>          * ignore backward steps and wait until the clock catches up
>>> again.  Of
>>>          * course this will cause problems if the clock is manually
>>> adjusted.  To
>>>          * which the doctor says, "don't do that".
>>>          */
>>>
>>> i start recalling that yesterday i adjusted clock back by 1 hour,
>>> because after installing a windoze in dual boot it was set it wrongly.
>>>
>>> But of course it could be anything else :)
>>>
>>> --
>>> Best regards,
>>> Igor Stasenko AKA sig.
>>>
>>
>>
>
>
Reply | Threaded
Open this post in threaded view
|

Re: [Cog] Funny thing about time

Igor Stasenko

Well, putting the CMOS stuff aside,
i really don't like that VM could hang indefinitely if user will
adjust the clock settings.

On 12 January 2011 13:22, Levente Uzonyi <[hidden email]> wrote:

>
> On Wed, 12 Jan 2011, Frank Shearar wrote:
>
>>
>> Yes, Windows like to set the CMOS clock to local time, and Linux likes to
>> use UTC.
>>
>> My brother recently ran into a similar problem with a FreeBSD machine
>> running as a guest inside a Windows host. He had to touch
>> /etc/wall_cmos_clock to fix it.
>>
>> From the adjkerntz man page:
>>
>> If the file /etc/wall_cmos_clock exists, it means that CMOS clock keeps
>> local time (MS-DOS and MS-Windows compatible mode).  If that file does not
>> exist, it means that the CMOS clock keeps UTC time.
>>
>> Maybe there's a similar Linux/Mac fix?
>
> Yes, there is, but different linux distributions have different files IIRC.
> Btw, if you tell linux to assume that the hardware clock stores the local
> time, you'll still have problems after daylight saving related time changes,
> because both OSs will adjust the clock.
>
>
> Levente
>
>>
>> frank
>>
>> On 2011/01/12 01:18, Chris Muller wrote:
>>>
>>> Every time I jump back and forth from Windows to Linux on my primary
>>> laptop, the clock is wrong...
>>>
>>> On Tue, Jan 11, 2011 at 4:37 PM, Igor Stasenko<[hidden email]>
>>>  wrote:
>>>>
>>>> Yesterday on my comp two separate running images are hung
>>>> simultaneously, while i were doing something completely outside of
>>>> them.
>>>>
>>>> And now during reading  the sqUnixHeartBeat.c:
>>>>
>>>>
>>>>        /* The native clock may go backwards, e.g. due to NTP
>>>> adjustments, although
>>>>         * why it can't avoid small backward steps itself, I don't know.
>>>> Simply
>>>>         * ignore backward steps and wait until the clock catches up
>>>> again.  Of
>>>>         * course this will cause problems if the clock is manually
>>>> adjusted.  To
>>>>         * which the doctor says, "don't do that".
>>>>         */
>>>>
>>>> i start recalling that yesterday i adjusted clock back by 1 hour,
>>>> because after installing a windoze in dual boot it was set it wrongly.
>>>>
>>>> But of course it could be anything else :)
>>>>
>>>> --
>>>> Best regards,
>>>> Igor Stasenko AKA sig.
>>>>
>>>
>>>
>>
>>
>



--
Best regards,
Igor Stasenko AKA sig.
Reply | Threaded
Open this post in threaded view
|

Re: [Cog] Funny thing about time

Eliot Miranda-2
 
So what's the algorithm?  If the time changes by more than half an hour (between the heartbeat sampling the clock) it's a clock change but if its less than half an hour its clock drift.  What threshold value would you use?  10 minutes? Half an hour?  A minute? I'm interested in opinions here This is easy to add and a good idea, but getting that magic constant right is important.

On Wed, Jan 12, 2011 at 4:24 AM, Igor Stasenko <[hidden email]> wrote:

Well, putting the CMOS stuff aside,
i really don't like that VM could hang indefinitely if user will
adjust the clock settings.

On 12 January 2011 13:22, Levente Uzonyi <[hidden email]> wrote:
>
> On Wed, 12 Jan 2011, Frank Shearar wrote:
>
>>
>> Yes, Windows like to set the CMOS clock to local time, and Linux likes to
>> use UTC.
>>
>> My brother recently ran into a similar problem with a FreeBSD machine
>> running as a guest inside a Windows host. He had to touch
>> /etc/wall_cmos_clock to fix it.
>>
>> From the adjkerntz man page:
>>
>> If the file /etc/wall_cmos_clock exists, it means that CMOS clock keeps
>> local time (MS-DOS and MS-Windows compatible mode).  If that file does not
>> exist, it means that the CMOS clock keeps UTC time.
>>
>> Maybe there's a similar Linux/Mac fix?
>
> Yes, there is, but different linux distributions have different files IIRC.
> Btw, if you tell linux to assume that the hardware clock stores the local
> time, you'll still have problems after daylight saving related time changes,
> because both OSs will adjust the clock.
>
>
> Levente
>
>>
>> frank
>>
>> On 2011/01/12 01:18, Chris Muller wrote:
>>>
>>> Every time I jump back and forth from Windows to Linux on my primary
>>> laptop, the clock is wrong...
>>>
>>> On Tue, Jan 11, 2011 at 4:37 PM, Igor Stasenko<[hidden email]>
>>>  wrote:
>>>>
>>>> Yesterday on my comp two separate running images are hung
>>>> simultaneously, while i were doing something completely outside of
>>>> them.
>>>>
>>>> And now during reading  the sqUnixHeartBeat.c:
>>>>
>>>>
>>>>        /* The native clock may go backwards, e.g. due to NTP
>>>> adjustments, although
>>>>         * why it can't avoid small backward steps itself, I don't know.
>>>> Simply
>>>>         * ignore backward steps and wait until the clock catches up
>>>> again.  Of
>>>>         * course this will cause problems if the clock is manually
>>>> adjusted.  To
>>>>         * which the doctor says, "don't do that".
>>>>         */
>>>>
>>>> i start recalling that yesterday i adjusted clock back by 1 hour,
>>>> because after installing a windoze in dual boot it was set it wrongly.
>>>>
>>>> But of course it could be anything else :)
>>>>
>>>> --
>>>> Best regards,
>>>> Igor Stasenko AKA sig.
>>>>
>>>
>>>
>>
>>
>



--
Best regards,
Igor Stasenko AKA sig.

Reply | Threaded
Open this post in threaded view
|

Re: [Cog] Funny thing about time

Bert Freudenberg


On 12.01.2011, at 19:17, Eliot Miranda wrote:

> So what's the algorithm?  If the time changes by more than half an hour (between the heartbeat sampling the clock) it's a clock change but if its less than half an hour its clock drift.  What threshold value would you use?  10 minutes? Half an hour?  A minute? I'm interested in opinions here This is easy to add and a good idea, but getting that magic constant right is important.

The smallest legitimate difference due to timezone adjustment is 15 minutes (e.g. when you travel between Nepal and India). That should not get ignored, so it is a hard upper bound.

But you want to go as low as possible. When my computer's clock has drifted more than a minute I might want to adjust it, and I'd expect Squeak to catch up. So going below a minute would be good. Do we have evidence what deltas you actually get for a drifting clock?

- Bert -