Prev: Intel PCIe 1Gig NIC driver changes coming
Next: problem with my remote machine
From: Àngel Català on 8 Apr 2008 13:09
Dave Uhring escribió:
> On Tue, 08 Apr 2008 09:33:43 +0200, Àngel Català wrote:
>
>> It does not seems a configuration problem because, when a restart my ntp
>> service, after boot, in a minute or a couple of minutes I get the right
>> time. It seems a problem with ntp starting. At boot time, ntp starts (I
>> can see ntpd process with ps), but it seems it is not working as it
>> should be (when I type ntpq -q, it answers me theres is no association
>> IDs).
>
> Then perhaps you should hack your /etc/init.d/ntp script to run ntpdate
> prior to starting the daemon.
>
> But "ntpq -q", really? How about "ntpq -p"?

Hi again,

I have tryed calling ntpdate (with time servers I have configured in
ntp.conf) prior to start ntpd in /etc/init.d/ntp (and a "sleep 2"
command between both calls) and it still fails.

This is really a great mistery to me !!!!. I have never had problems
with NTP.

Currently I have Debian 4.0 'Etch' r3 i386 installed, but a month ago I
had Debian 4.0 'Etch' r0 x64 and it worked fine. It is driving me crazy.

Bye.
From: Dave Uhring on 8 Apr 2008 13:24
On Tue, 08 Apr 2008 19:09:23 +0200, Àngel Català wrote:
> Dave Uhring escribió:

>> Then perhaps you should hack your /etc/init.d/ntp script to run ntpdate
>> prior to starting the daemon.

> I have tryed calling ntpdate (with time servers I have configured in
> ntp.conf) prior to start ntpd in /etc/init.d/ntp (and a "sleep 2"
> command between both calls) and it still fails.

Stop ntpd, /etc/init.d/ntp stop

Run ntpdate your_NTP_server just once, then 'date'. Does the correct
time appear?

> Currently I have Debian 4.0 'Etch' r3 i386 installed, but a month ago I
> had Debian 4.0 'Etch' r0 x64 and it worked fine. It is driving me crazy.

I upgrade from Etch r1 to Lenny as soon as I do the installation. Never
once had a problem with NTP. I even run a local NTP server on my network
and *it* is running Lenny and synced to one closest stratum 2 server plus
two of the servers in the debian.pool.

The other systems on the net are Solaris -several versions- , OpenBSD,
FreeBSD, Slackware and Debian Lenny. All sync to the local NTP server
and all keep the correct time.

From: Àngel Català on 9 Apr 2008 03:10
Hi, Chris:

Chris Davies escribió:
> Àngel Català <ninguna(a)null.org> wrote:
>> I have installed Debian 4.0 Etch r3 i386 in my laptop and I have a
>> strange problem with clock setting with NTP that I can't solve by my self.
>
> What kind of laptop?
>

My laptop is a HP Compaq nw8440. A month ago, this laptop, was running
Debian 4.0 r0 x64 and time worked fine. Now is running Debian 4.0 r3
i386 and time is not working properly.


>> When I boot my laptop, my clock is always +2 hour over my current time.
>
> I had a similar problem with my DELL, which was nothing to do with ntp
> (other than it refusing to run because the time was so far off) and
> the solution turned out to be to add "HWCLOCKPARS=--directisa" to the
> /etc/default/rcS file.
>
> You can confirm whether you need this fix by trying this command
> (as root):
>
> hwclock --show
>
> If you get the message, "select() to /dev/rtc to wait for clock tick
> timed out" try it again:
>
> hwclock --show --directisa
>
> If this works for you (returning within a second with a time) then you
> need the fix.
>
> Chris
>

I have tryed what you said, and when I typed "hwclock --show" I have got
that message, and for "hwclock --show --directisa" it has worked, so I
hacked /etc/default/rcS, but it still is failing.

Thanks Chris.
From: Àngel Català on 9 Apr 2008 03:15
Dave Uhring escribió:
> On Tue, 08 Apr 2008 19:09:23 +0200, Àngel Català wrote:
>> Dave Uhring escribió:
>
>>> Then perhaps you should hack your /etc/init.d/ntp script to run ntpdate
>>> prior to starting the daemon.
>
>> I have tryed calling ntpdate (with time servers I have configured in
>> ntp.conf) prior to start ntpd in /etc/init.d/ntp (and a "sleep 2"
>> command between both calls) and it still fails.
>
> Stop ntpd, /etc/init.d/ntp stop
>
> Run ntpdate your_NTP_server just once, then 'date'. Does the correct
> time appear?

Yes, it works fine. I stop ntp daemon, then I call ntpdate with my NTP
server and "date" answers me with the right time (also clock is set
properly). So I think that configuration parameters are OK, but I can
not understand what is failing.

By the way, everytime time is set properly (with manual restarting ntp
daemon or by typing ntpdate), my screen becomes black with a big "X"
symbol inside. I don't know if this detail is important.

>
>> Currently I have Debian 4.0 'Etch' r3 i386 installed, but a month ago I
>> had Debian 4.0 'Etch' r0 x64 and it worked fine. It is driving me crazy.
>
> I upgrade from Etch r1 to Lenny as soon as I do the installation. Never
> once had a problem with NTP. I even run a local NTP server on my network
> and *it* is running Lenny and synced to one closest stratum 2 server plus
> two of the servers in the debian.pool.
>
> The other systems on the net are Solaris -several versions- , OpenBSD,
> FreeBSD, Slackware and Debian Lenny. All sync to the local NTP server
> and all keep the correct time.
>

Thanks again.
From: Dave Uhring on 9 Apr 2008 04:04
On Wed, 09 Apr 2008 09:15:16 +0200, �ngel Catal� wrote:
> Dave Uhring escribi�:

>> Stop ntpd, /etc/init.d/ntp stop
>>
>> Run ntpdate your_NTP_server just once, then 'date'. Does the correct
>> time appear?
>
> Yes, it works fine. I stop ntp daemon, then I call ntpdate with my NTP
> server and "date" answers me with the right time (also clock is set
> properly). So I think that configuration parameters are OK, but I can
> not understand what is failing.

Then hack that very ntpdate command into /etc/init.d/ntp. Be sure to put
the ntpdate command in the right place,

start)
log_daemon_msg "Starting NTP server" "ntpd"
if [ -z "$UGID" ]; then
log_failure_msg "user \"$RUNASUSER\".....
exit 1
fi
right here ===> /usr/sbin/ntpdate your_NTP_server
start-stop-daemon --start --quiet --oknodo....

It's silly to run 2 consecutive invocations of ntpdate since the intent
is to get your system time "close" to accurate before starting ntpd.

> By the way, everytime time is set properly (with manual restarting ntp
> daemon or by typing ntpdate), my screen becomes black with a big "X"
> symbol inside. I don't know if this detail is important.

That only shows that your screensaver works.

First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: Intel PCIe 1Gig NIC driver changes coming
Next: problem with my remote machine