FAQ Search Today's Posts Mark Forums Read
» Video Reviews

» Linux Archive

Linux-archive is a website aiming to archive linux email lists and to make them easily accessible for linux users/developers.


» Sponsor

» Partners

» Sponsor

Go Back   Linux Archive > Debian > Debian User

 
 
LinkBack Thread Tools
 
Old 09-15-2012, 02:47 PM
Camaleón
 
Default ntpd crashes.

On Sat, 15 Sep 2012 15:43:04 +0200, Mauro wrote:

> I think ntpd crashes are because my server lost time.

How can that be? If ntpd daemon is running, the server has to be synced
and showing the right time. And in the event the time is too much skewed,
ntpd shouldn't crash but left the time unsynced and registering the error
at the logs (check if adding "-x" argument to ntpd helps here).

> I have ntpd in two server, now I've seen that in one of these ntp
> crashes and the time of the server is 1 hour forward. That's why ntp
> crashes: server time goes 1 hour forward and ntp can't resynchronize so
> it crashes.

IIRC, you mentioned that after the crash, ntpd could be restarted again
without problems. If that's true, it means at the time ntpd daemon is
started, the time of the server is still close to a good enough for ntpd
can be launched without manual corrections.

> Now I don't know why my server time goes 1 hour forward.

Becasue ntpd crashed?

> Hwclock --debug says that the time is correct, it is set on UTC, so why
> sometimes it goes forward?

Virtual machines do suffer from time sync issues so maybe a clusterized
environment can also be affected somehow :-?

> Perhaps a problem in the cmos battery?

You can change the CMOS battery to see if that makes a difference but a
death or bad battery would also have caused additional side effects such
as motherboard settings restoring to defaults values which is not the
case.

Greetings,

--
Camaleón


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/k324d6$200$4@ger.gmane.org
 
Old 09-16-2012, 03:46 PM
Camaleón
 
Default ntpd crashes.

El 2012-09-16 a las 00:17 +0200, Mauro escribió:

(resending to the list)

> On 15 September 2012 16:47, Camaleón <noelamac@gmail.com> wrote:
> > On Sat, 15 Sep 2012 15:43:04 +0200, Mauro wrote:
> >
> >> I think ntpd crashes are because my server lost time.
> >
> > How can that be? If ntpd daemon is running, the server has to be synced
> > and showing the right time. And in the event the time is too much skewed,
> > ntpd shouldn't crash but left the time unsynced and registering the error
> > at the logs (check if adding "-x" argument to ntpd helps here).
> >
> >> I have ntpd in two server, now I've seen that in one of these ntp
> >> crashes and the time of the server is 1 hour forward. That's why ntp
> >> crashes: server time goes 1 hour forward and ntp can't resynchronize so
> >> it crashes.
> >
> > IIRC, you mentioned that after the crash, ntpd could be restarted again
> > without problems. If that's true, it means at the time ntpd daemon is
> > started, the time of the server is still close to a good enough for ntpd
> > can be launched without manual corrections.
> >
> >> Now I don't know why my server time goes 1 hour forward.
> >
> > Becasue ntpd crashed?
> >
>
> Then for me it is a great problem because there are no reasons for
> ntpd to crash.

Sure, there's no single reason for a daemon to crash (their normal
status should be "running" unless you tell otherwise :-P), a crash is
something that needs to be investigated in deep but with debug flag
turned off an no other insightful logs, that's hard to achieve.

> In one of my two cluster nodes ntp crashes only in the second node.
> The nodes have the same hardware and the same software so I don't know
> why ntpd crashes always in the second node.

Try by appending the "-x" argument at the "/etc/default/ntp" file from
the server that crashes. If your thoughs are correct, this could mitigate
the time difference.

Greetings,

--
Camaleón


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 20120916154640.GA5117@stt008.linux.site">http://lists.debian.org/20120916154640.GA5117@stt008.linux.site
 
Old 09-16-2012, 04:36 PM
Mauro
 
Default ntpd crashes.

On 16 September 2012 17:46, Camaleón <noelamac@gmail.com> wrote:

> Try by appending the "-x" argument at the "/etc/default/ntp" file from
> the server that crashes. If your thoughs are correct, this could mitigate
> the time difference.

I've uninstalled ntp and installed openntp from squeeze backports.
I hope this works.


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: CAE17a0W+s79RdjqATWRsYhou2FAysMBRL7Z-1kVN6E11Hsci5g@mail.gmail.com">http://lists.debian.org/CAE17a0W+s79RdjqATWRsYhou2FAysMBRL7Z-1kVN6E11Hsci5g@mail.gmail.com
 
Old 09-17-2012, 05:33 PM
Gerald Turner
 
Default ntpd crashes.

Mauro <mrsanna1@gmail.com> writes:
> I think ntpd crashes are because my server lost time.
> I have ntpd in two server, now I've seen that in one of these ntp
> crashes and the time of the server is 1 hour forward.
> That's why ntp crashes: server time goes 1 hour forward and ntp can't
> resynchronize so it crashes.
> Now I don't know why my server time goes 1 hour forward.
> Hwclock --debug says that the time is correct, it is set on UTC, so
> why sometimes it goes forward?
> Perhaps a problem in the cmos battery?

Hi Mauro, I find this interesting because I have seen the same behavior
with ntpd crashes on my server. Time jumps forward one hour every time
this has happened. However I'm not convinced it's the hardware causing
ntpd to crash. I don't understand why ntpd process exiting silently
would allow the jump either, I would expect it to drift a few seconds
over time, but one hour exactly every time?! Even with monit in place,
within the two minute window that polling takes place and ntpd has
crashed, before monit automatically starts ntpd, time is forward an
hour.

Do you happen to run a SuperMicro 5016T-MTFB¹ or a simlar board with an
IPMI/baseboard management controller?

¹ http://www.supermicro.com/products/system/1U/5016/SYS-5016T-MTF.cfm

--
Gerald Turner Email: gturner@unzane.com JID: gturner@unzane.com
GPG: 0xFA8CD6D5 21D9 B2E8 7FE7 F19E 5F7D 4D0C 3FA0 810F FA8C D6D5
 
Old 09-17-2012, 07:47 PM
Bob Proulx
 
Default ntpd crashes.

Mauro wrote:
> I think ntpd crashes are because my server lost time.
> I have ntpd in two server, now I've seen that in one of these ntp
> crashes and the time of the server is 1 hour forward.
> That's why ntp crashes: server time goes 1 hour forward and ntp can't
> resynchronize so it crashes.

Is it really crashing or is this intended behavior. The ntpd is
documented with:

-g Normally, ntpd exits with a message to the system log if the
offset exceeds the panic threshold, which is 1000 s by
default. This option allows the time to be set to any value
without restriction; however, this can happen only once. If
the threshold is exceeded after that, ntpd will exit with a
message to the system log. This option can be used with the
-q and -x options.

Therefore if the time is one hour forward that will be greater than
the threshold 1000 seconds and ntp will exit. As described this is so
that the admin may adjust the time manually to some specific value
(for whatever reason, testing, whatever) and ntpd will get out of the
way.

It also says that ntpd will log a message to the system log when this
condition occurs.

I think it is possible that you have some additional process that is
setting the clock and this is jumping the time forward one hour and
because the time is jumped forward one hour that exceeds the ntpd
panic threshold and therefore ntpd decides that it should get out of
the way and it intentionally exits. This intentional behavior is
quite a bit different from a crash.

> Now I don't know why my server time goes 1 hour forward.

I have in past experience run into a case where someone had set up a
date setting cron task that is running exactly on the hour. Due to
small race condition differences the cron task would sometimes set the
time incorrectly and would cause an hour jump as you describe. In my
case removing the date setting cron task solved the problem I
described. But I do not know if that is your problem or not.

> Hwclock --debug says that the time is correct, it is set on UTC, so
> why sometimes it goes forward?
> Perhaps a problem in the cmos battery?

The hardware clock is normally only used at two different places. One
is at boot when the operating system time is loaded from the hardware
clock time. The other is at system shutdown when the hardware clock
time is set from the current operating system time. The hardware
clock is not otherwised used by the operating system. The operating
system always uses the operating system time.

Bob
 
Old 09-18-2012, 07:50 AM
Mauro
 
Default ntpd crashes.

On 17 September 2012 19:33, Gerald Turner <gturner@unzane.com> wrote:
> Mauro <mrsanna1@gmail.com> writes:
>> I think ntpd crashes are because my server lost time.
>> I have ntpd in two server, now I've seen that in one of these ntp
>> crashes and the time of the server is 1 hour forward.
>> That's why ntp crashes: server time goes 1 hour forward and ntp can't
>> resynchronize so it crashes.
>> Now I don't know why my server time goes 1 hour forward.
>> Hwclock --debug says that the time is correct, it is set on UTC, so
>> why sometimes it goes forward?
>> Perhaps a problem in the cmos battery?
>
> Hi Mauro, I find this interesting because I have seen the same behavior
> with ntpd crashes on my server. Time jumps forward one hour every time
> this has happened. However I'm not convinced it's the hardware causing
> ntpd to crash. I don't understand why ntpd process exiting silently
> would allow the jump either, I would expect it to drift a few seconds
> over time, but one hour exactly every time?! Even with monit in place,
> within the two minute window that polling takes place and ntpd has
> crashed, before monit automatically starts ntpd, time is forward an
> hour.

I'm trying openntp now, I hope it works.


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: CAE17a0V5UgHBZGwxRAar4dSMKTrv9HAgt8DZAeWhr9iW+SS1g w@mail.gmail.com">http://lists.debian.org/CAE17a0V5UgHBZGwxRAar4dSMKTrv9HAgt8DZAeWhr9iW+SS1g w@mail.gmail.com
 
Old 09-18-2012, 07:52 AM
Mauro
 
Default ntpd crashes.

On 17 September 2012 21:47, Bob Proulx <bob@proulx.com> wrote:
> Mauro wrote:
>> I think ntpd crashes are because my server lost time.
>> I have ntpd in two server, now I've seen that in one of these ntp
>> crashes and the time of the server is 1 hour forward.
>> That's why ntp crashes: server time goes 1 hour forward and ntp can't
>> resynchronize so it crashes.
>
> Is it really crashing or is this intended behavior. The ntpd is
> documented with:
>
> -g Normally, ntpd exits with a message to the system log if the
> offset exceeds the panic threshold, which is 1000 s by
> default. This option allows the time to be set to any value
> without restriction; however, this can happen only once. If
> the threshold is exceeded after that, ntpd will exit with a
> message to the system log. This option can be used with the
> -q and -x options.
>
> Therefore if the time is one hour forward that will be greater than
> the threshold 1000 seconds and ntp will exit. As described this is so
> that the admin may adjust the time manually to some specific value
> (for whatever reason, testing, whatever) and ntpd will get out of the
> way.
>
> It also says that ntpd will log a message to the system log when this
> condition occurs.

No message logs.

>
> I think it is possible that you have some additional process that is
> setting the clock and this is jumping the time forward one hour and
> because the time is jumped forward one hour that exceeds the ntpd
> panic threshold and therefore ntpd decides that it should get out of
> the way and it intentionally exits. This intentional behavior is
> quite a bit different from a crash.

I have no additional processes.
Now I'm trying openntp.


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/CAE17a0WNw_a=vTPtNu3mh4_KD_zX6KAyS8QORsUMyivQFyTJU w@mail.gmail.com
 
Old 09-18-2012, 01:51 PM
Stefan Monnier
 
Default ntpd crashes.

> with ntpd crashes on my server. Time jumps forward one hour every time
> this has happened. However I'm not convinced it's the hardware causing

Sounds like something is causing the one-hour jump, and that in turns
causes ntpd to go bonkers.
As for what causes this jump, I don't know. Some cron job, maybe?


Stefan


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: jwvfw6f4f46.fsf-monnier+gmane.linux.debian.user@gnu.org">http://lists.debian.org/jwvfw6f4f46.fsf-monnier+gmane.linux.debian.user@gnu.org
 
Old 09-18-2012, 08:21 PM
Mauro
 
Default ntpd crashes.

On 18 September 2012 15:51, Stefan Monnier <monnier@iro.umontreal.ca> wrote:
>> with ntpd crashes on my server. Time jumps forward one hour every time
>> this has happened. However I'm not convinced it's the hardware causing
>
> Sounds like something is causing the one-hour jump, and that in turns
> causes ntpd to go bonkers.
> As for what causes this jump, I don't know. Some cron job, maybe?

No I have no cron jobs.


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: http://lists.debian.org/CAE17a0Vd0h2jMPSco2t9=m1GaA7ZqP0nNVrCogGabCcADUszd Q@mail.gmail.com
 
Old 09-20-2012, 02:03 PM
Stefan Monnier
 
Default ntpd crashes.

>>> with ntpd crashes on my server. Time jumps forward one hour every time
>>> this has happened. However I'm not convinced it's the hardware causing
>> Sounds like something is causing the one-hour jump, and that in turns
>> causes ntpd to go bonkers.
>> As for what causes this jump, I don't know. Some cron job, maybe?
> No I have no cron jobs.

Not sure what else it can be. The one-hour difference suggests it might
be linked to time-zone or DST issues, maybe some code that syncs
up the internal time with some "external" (could be an RTC) clock?


Stefan


--
To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: jwvd31gztdh.fsf-monnier+gmane.linux.debian.user@gnu.org">http://lists.debian.org/jwvd31gztdh.fsf-monnier+gmane.linux.debian.user@gnu.org
 

Thread Tools




All times are GMT. The time now is 08:22 AM.

VBulletin, Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO ©2007, Crawlability, Inc.
Copyright ©2007 - 2008, www.linux-archive.org