Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-15 Thread David Baron
On Monday 15 June 2009 08:20:23 debian-user-digest-requ...@lists.debian.org 
wrote:
  Even more. I get a bootup error message failing to set date (and time)
  to ... the following day + 3 hours!!

 You may have a bad system clock. That does happen on some older
 hardware or if a CMOS battery goes bad. I'd do the following:

 - Make sure you're calling hwlock --systohc --localtime at some
   point when your time has been correctly set.
This is ostensibly being done unless an error is being made in 
/etc/init.d/hwclock.sh.
If this is the error, howclockfirst.sh is trying to read a one day + three 
hours locatime date. This ostensiblty fails (it takes the time, not the date).

 - Make sure /etc/default/ntp contains NTPD_OPTS='-g' so that you
   can handle large clock offsets.
No such option available in my ntpdate installation.

 If that still doesn't really resolve your problem, then you might want
 to try chrony instead, which assumes that your clock has a stable drift
 rate that can be compensated for. NTP is certainly better for accuracy,
 but chrony is a little more tolerant of clock issues as far as I can
 tell.
It would seem that a bad battery would be a slow clock, drifting somewhat 
consistantly slower. Had that once. The consistancy of the problem plus the 
fact of three hours implies the first option. May I should back up the script 
and then hard code it to --localtime?


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-14 Thread David Baron
Has been happening quite a bit lately:

Problem with ntpdate. If and when it finally works, get a change of 10,800 
seconds ... that is 30 minutes.

The dovecot IMAPd server is unreadable. Restarting it yields a complaint
about 10,800 seconds of time, killing itself now! Restarting it again
works OK.

What's happening? Bug? Fix?

Of course, we are missing the point. No, it is not 30 minutes like my eyes 
fooled me into a cute title. This is three hours and that is a significant 
number. I live at GMT + 2. Add one for daylight savings time.

However, I have never used GMT (universal) time to save my system time.
So why is this happening now? Ntpdate has not been upgraded past few days.

... and why should dovecot care?

OK, time to fix this. Since my system comes up with the correct time zone 
but three hours later, it must be interpreting the stored (hw) time as GMT 
and adding the three hours (gmt+2 + daylight savings).

How to fix this?

Even more. I get a bootup error message failing to set date (and time) to ... 
the following day + 3 hours!!


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-14 Thread Todd A. Jacobs
On Sun, Jun 14, 2009 at 04:50:23PM +0300, David Baron wrote:

 Even more. I get a bootup error message failing to set date (and time)
 to ... the following day + 3 hours!!

You may have a bad system clock. That does happen on some older
hardware or if a CMOS battery goes bad. I'd do the following:

- Make sure you're calling hwlock --systohc --localtime at some
  point when your time has been correctly set.

- Make sure /etc/default/ntp contains NTPD_OPTS='-g' so that you
  can handle large clock offsets.

If that still doesn't really resolve your problem, then you might want
to try chrony instead, which assumes that your clock has a stable drift
rate that can be compensated for. NTP is certainly better for accuracy,
but chrony is a little more tolerant of clock issues as far as I can
tell.

-- 
Oh, look: rocks!
-- Doctor Who, Destiny of the Daleks


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-11 Thread David Baron
Has been happening quite a bit lately:

Problem with ntpdate. If and when it finally works, get a change of 10,800 
seconds ... that is 30 minutes.

The dovecot IMAPd server is unreadable. Restarting it yields a complaint
about 10,800 seconds of time, killing itself now! Restarting it again
works OK.

What's happening? Bug? Fix?

Of course, we are missing the point. No, it is not 30 minutes like my eyes 
fooled me into a cute title. This is three hours and that is a significant 
number. I live at GMT + 2. Add one for daylight savings time.

However, I have never used GMT (universal) time to save my system time.
So why is this happening now? Ntpdate has not been upgraded past few days.

... and why should dovecot care?

OK, time to fix this. Since my system comes up with the correct time zone 
but three hours later, it must be interpreting the stored (hw) time as GMT and 
adding the three hours (gmt+2 + daylight savings).

How to fix this?


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-11 Thread Andrei Popescu
On Thu,11.Jun.09, 16:08:54, David Baron wrote:
 
 OK, time to fix this. Since my system comes up with the correct time zone 
 but three hours later, it must be interpreting the stored (hw) time as GMT 
 and 
 adding the three hours (gmt+2 + daylight savings).
 
 How to fix this?

If you only run Linux on this machine I suggest setting the hardware 
clock to UTC (GMT). Otherwise change the line:

UTC=yes

in /etc/default/rcS

Regards,
Andrei
-- 
If you can't explain it simply, you don't understand it well enough.
(Albert Einstein)


signature.asc
Description: Digital signature


Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-11 Thread Andrew M.A. Cater
On Thu, Jun 11, 2009 at 04:08:54PM +0300, David Baron wrote:
 Has been happening quite a bit lately:
 
 Problem with ntpdate. If and when it finally works, get a change of 10,800 
 seconds ... that is 30 minutes.
 
 The dovecot IMAPd server is unreadable. Restarting it yields a complaint
 about 10,800 seconds of time, killing itself now! Restarting it again
 works OK.
 
 What's happening? Bug? Fix?
 
 Of course, we are missing the point. No, it is not 30 minutes like my eyes 
 fooled me into a cute title. This is three hours and that is a significant 
 number. I live at GMT + 2. Add one for daylight savings time.
 
 However, I have never used GMT (universal) time to save my system time.
 So why is this happening now? Ntpdate has not been upgraded past few days.
 
 ... and why should dovecot care?
 
 OK, time to fix this. Since my system comes up with the correct time zone 
 but three hours later, it must be interpreting the stored (hw) time as GMT 
 and 
 adding the three hours (gmt+2 + daylight savings).
 
 How to fix this?
 

Set the computer time to GMT.

date -u MMDDhhmmCCYY.ss

date -u 061122002009

hwclock --systohc

[Sets board hardware clock from the date you just gave.]

dpkg-reconfigure -plow tzdata

[And give the appropriate time zone offset. I live in UK and keep my 
time on GMT all year round, so I'd accept GMT here.]

HTH,

AndyC

 
 -- 
 To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
 with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-08 Thread David Baron
Has been happening quite a bit lately:

Problem with ntpdate. If and when it finally works, get a change of 10,800 
seconds ... that is 30 minutes.

The dovecot IMAPd server is unreadable. Restarting it yields a complaint
about 10,800 seconds of time, killing itself now! Restarting it again
works OK.

What's happening? Bug? Fix?

Of course, we are missing the point. No, it is not 30 minutes like my eyes 
fooled me into a cute title. This is three hours and that is a significant 
number. I live at GMT + 2. Add one for daylight savings time.

However, I have never used GMT (universal) time to save my system time. So why 
is this happening now? Ntpdate has not been upgraded past few days.

... and why should dovecot care?


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-07 Thread Daryl Styrk
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1

On Sun, Jun 07, 2009 at 05:07:08PM +0300, David Baron wrote:
 Has been happening quite a bit lately:
 
 Problem with ntpdate. If and when it finally works, get a change of 10,800 
 seconds ... that is 30 minutes.
 
 The dovecot IMAPd server is unreadable. Restarting it yields a complaint 
 about 
 10,800 seconds of time, killing itself now! Restarting it again works OK.
 
 What's happening? Bug? Fix?

I've had nothing but great results with chrony.  Aside from an issue of
chrony falling back to 127.127.1.1 when it cannot connect to a ntp server.. 
Which happens often with my laptop as I move around during the day..  I 
fixed that by restarting the service hourly with cron.  I'm never more 
than a half second off across boots and stupid accurate while connected. 

System time : 0.00 seconds fast of NTP time

http://www.linuxjournal.com/article/5657
- -- 
Daryl Styrk
Naples FL, USA

-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAkor1eYACgkQ6baBhW8CzrgmpQCfWe4RNDLP004TwGkOVmOVNzn7
esAAnAjmV5PTMVGdK/pDz5mWPpHIZtQm
=bbnl
-END PGP SIGNATURE-


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-07 Thread Celejar
On Sun, 7 Jun 2009 10:59:50 -0400
Daryl Styrk darylst...@gmail.com wrote:

...

 I've had nothing but great results with chrony.  Aside from an issue of
 chrony falling back to 127.127.1.1 when it cannot connect to a ntp server.. 
 Which happens often with my laptop as I move around during the day..  I 
 fixed that by restarting the service hourly with cron.  I'm never more 
 than a half second off across boots and stupid accurate while connected. 

I use chrony and like it, but I've found the fact that it drops ntp
servers whose names don't resolve really annoying, since on my laptop,
I'm often not connected, and if chrony is started when unconnected, it
rejects all my servers specified by DNS name.  I could use hard-coded
IP addresses, but I find that distasteful, so I've settled for a
trivial script that restarts chrony when I bring a network connection
up.

There's a bug open since 2004:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=268289

Celejar
--
mailmin.sourceforge.net - remote access via secure (OpenPGP) email
ssuds.sourceforge.net - A Simple Sudoku Solver and Generator


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org



Re: 30 Minutes (system time/ntpdate/dovecot)

2009-06-07 Thread John Hasler
The current chrony initscript looks for a defaultroute at startup.  If one
is found it puts chronyd online.  The ip-up script is only used with PPP.
-- 
John Hasler


-- 
To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org 
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org