[mdlug] Can the date stamp on a file be ahead of the clock

R. Kannan rkannan at peoplepc.com
Fri Sep 7 12:23:25 EDT 2007


Thanks Dr.Meier & Ray.

BTW, My issue was more complicated than I thought. The problem was only happening in my $HOME which was ina mounted filesystem from another machine. That machine had its clock out of sync with the machine on which I was running on.

-----Original Message-----
>From: Robert Meier <eaglecoach at wwnet.com>
>Sent: Sep 6, 2007 11:20 PM
>To: mdlug at mdlug.org
>Subject: Re: [mdlug] Can the date stamp on a file be ahead of the clock
>
>Ray, R.,
>
>R.>> I am seeing a puzzling behavior in a linux machine at work. 
>R.>> See the series of commands and the responses below
>R.>>
>R.>> kez at vision->uname -a
>R.>>   Linux vision 2.6.16.21-0.8-smp #1 SMP Mon Jul 3 18:25:39 UTC 2006 ...
>R.>> kez at vision->touch tt1
>R.>> kez at vision->ls -l tt1
>R.>>   -rw-r--r-- 1 kez hks 0 2007-09-06 14:49 tt1
>R.>> kez at vision->date
>R.>>   Thu Sep  6 14:42:16 EDT 2007
>
>Ray> Linux, and probably other unixes, [keep] a system clock separate
>Ray> from the hardware clock. ...
>Ray> pts/3 # hwclock;date
>Ray>   Thu Sep  6 15:05:33 2007  -0.585800 seconds
>Ray>   Thu Sep  6 15:05:32 EDT 2007
>Ray> pts/3 # date 09061515
>Ray>   Thu Sep  6 15:15:00 EDT 2007
>Ray> pts/3 # hwclock;date
>Ray>   Thu Sep  6 15:06:12 2007  -0.305832 seconds
>Ray>   Thu Sep  6 15:15:03 EDT 2007
>Ray> pts/3 # touch foo
>Ray> pts/3 # ls -l foo
>Ray>   -rw-r--r--  1 root root 0 Sep  6 15:15 foo
>Ray> It seems that, on my system at least,
>Ray>   the hardware clock is used for file system time stamps.
>
>I assume you meant to say "the system clock is used for file system
>time stamps", as the file timestamp matched the date(1) setting
>in your example above.
>     "date - print or set the system date and time"
>	   -- date(1)
>
>Ray> I've read the relevant man pages repeatedly and still don't
>Ray> really see the point of keeping two different clocks. I just
>Ray> keep mine synchronized. First I use the date command to set
>Ray> the system clock, then I run (as root of course):
>Ray>    ->hwclock --systohc
>
>The reason for the two clocks is convenience, performance and
>volatility.  The hardware clock keeps time even when the system
>is shutdown.   The system clock is available to the kernel (and
>its filesystem drivers) millions of times per second if necessary
>without the overhead of an i/o interface.
>
>You can operate a unix system without a hardware clock,
>if you simply set the system clock from another source (e.g.
>network time server) each time you boot.
>
>Accurate time is the defacto standard for coordinating network services
>like email, source file sharing and remote makes.
>
>A hardware clock, consulted at boot (and otherwise occasionally)
>is so cheap a motherboard acccessory, that its rare to find a
>motherboard without one.
>
>
>
>R.>> How could this be? I did not think this was possible...
>
>I'm not sure what "this" is.  Do you mean a timestamp in the future?
>
>An idiom for setting a (low-resolution) timeout in a script
>is to set a future timestamp and check for passage.
>
>    bash> touch --date '+1day' /tmp/timelimit
>    bash> while test /tmp/timelimit -nt log
>        > do something-repeatable > log
>        > done
>
>Anything (e.g. something repeatable) scriptable can use flexible
>time specifications without having to implement parsing (already
>supplied by date(1)).
>
>Hopefully helpful,
>-- 
>DrB
>
>_______________________________________________
>mdlug mailing list
>mdlug at mdlug.org
>http://mdlug.org/mailman/listinfo/mdlug


________________________________________
PeoplePC Online
A better way to Internet
http://www.peoplepc.com



More information about the mdlug mailing list