ntpd madness

ntpd madness

Post by Mike Scot » Thu, 09 Jun 2011 22:13:21



This is seriously in the realms of 'cannot happen'.....

I've a (or rather /had/ a) perfectly well running 6.2 system. In
particular, I've been using ntpdate and ntp to keep the time synced.
That's been running untouched, across boots, for 4 years or so.

Today, I took the system down to plug in a new disk to install 8.2.
* hiccup there (separate problem), so I I've gone back to the old
disk. (Always handy to do reinstalls to something completely separate
from the existing disk! Then you know nothing's changed there.... except
that...)

Now I get ntdate and ntpd both crashing out moaning about no IPv6
interfaces. There aren't any; and never have been.

The problem is, I've touched nothing (apart from the disk cables, that
is; and I've just checked the file dates to be doubly sure).

ntpdate I've now coaxed into operation by running with a '-4' flag --
but why should this be needed all at once when it wasn't last week? The
ntpd man page offers the same '-4' option, but ntpd won't take it.
Without, it moans about no IPv6 then coredumps.

I'm at a loss, and can't even think where to start looking! Can anyone
suggest why a perfectly well configured ntp setup should suddenly stop
working like this? Or how to get ntpd back up again? Help!!

--
Mike Scott (unet2 <at> [deletethis] scottsonline.org.uk)
Harlow Essex England

 
 
 

ntpd madness

Post by BobH » Fri, 10 Jun 2011 04:22:08



Quote:> This is seriously in the realms of 'cannot happen'.....

> I've a (or rather /had/ a) perfectly well running 6.2 system. In
> particular, I've been using ntpdate and ntp to keep the time synced.
> That's been running untouched, across boots, for 4 years or so.

> Today, I took the system down to plug in a new disk to install 8.2.
> * hiccup there (separate problem), so I I've gone back to the old
> disk. (Always handy to do reinstalls to something completely separate
> from the existing disk! Then you know nothing's changed there.... except
> that...)

> Now I get ntdate and ntpd both crashing out moaning about no IPv6
> interfaces. There aren't any; and never have been.

> The problem is, I've touched nothing (apart from the disk cables, that
> is; and I've just checked the file dates to be doubly sure).

> ntpdate I've now coaxed into operation by running with a '-4' flag --
> but why should this be needed all at once when it wasn't last week? The
> ntpd man page offers the same '-4' option, but ntpd won't take it.
> Without, it moans about no IPv6 then coredumps.

> I'm at a loss, and can't even think where to start looking! Can anyone
> suggest why a perfectly well configured ntp setup should suddenly stop
> working like this? Or how to get ntpd back up again? Help!!

This is a WAG, but is there any chance that the BIOS configuration got
changed? If the disk did not change, the BIOS config seems like all that
is left.

Good Luck,
BobH

 
 
 

ntpd madness

Post by Alan Bawde » Fri, 10 Jun 2011 13:29:23


I'll bet that one or more of the servers mentioned in your /etc/ntp.conf
file now has an IPv6 address that it didn't have the last time you booted
the machine.
 
 
 

ntpd madness

Post by Mike Scot » Fri, 10 Jun 2011 16:25:00



Quote:> I'll bet that one or more of the servers mentioned in your /etc/ntp.conf
> file now has an IPv6 address that it didn't have the last time you booted
> the machine.

Bingo!

I use 3 pool addresses - just one of which now has a set of IPv6
addresses. Guess which I hadn't checked :-(

OTOH that really shouldn't make ntpd bomb out. It was eventually exiting
with a sig 11. Not helpful.

But I've switched ntp server now, and ntpd is running for the time
being. Maybe I need to get the upgrade finished pdq :-)

Many thanks.

--
Mike Scott (unet2 <at> [deletethis] scottsonline.org.uk)
Harlow Essex England

 
 
 

ntpd madness

Post by Rob Warno » Sat, 11 Jun 2011 11:27:31



+---------------
| > I'll bet that one or more of the servers mentioned in your /etc/ntp.conf
| > file now has an IPv6 address that it didn't have the last time you booted
| > the machine.
|
| Bingo!
| I use 3 pool addresses - just one of which now has a set of IPv6
| addresses. Guess which I hadn't checked :-(
+---------------

Yesterday (and today?!?) were "World IPv6 Day", a day for turning
on IPv6 in a large number of places that don't ordinarily use it
and seeing what breaks.  ;-}  ;-}

    http://iscold.sans.edu/diary.html?storyid=11023
    IPv6 Day Started
    Published: 2011-06-08,
    Last Updated: 2011-06-08 00:30:56 UTC
    by Johannes Ullrich (Version: 1)

    IPv6 day officially started at midnight GMT. Over the next 24 hrs,
    a number of large web sites will be reachable via IPv6. For example
    Google, Yahoo and Facebook added AAAA records.
    ...
    The next 24 hrs bring a unique opportunity to test IPv6 and to
    experiment with it. I recommend that you setup at least a test
    system and attempt to connect to IPv6 via a tunnelbroker.
    ...
    More information about IPv6 day:
    http://ipv6day.org
    ...

The first one was June 6th, 2006, and apparently it's now an annual
event(?) -- at least until the Internet switches over for good!  ;-}
Some results from this year's test:

    http://iscold.sans.edu/diary/IPv6+Day+Summary/11035
    IPv6 Day Summary
    Published: 2011-06-09,
    Last Updated: 2011-06-09 23:02:06 UTC
    by Johannes Ullrich (Version: 1)

    http://www.eweek.com/c/a/IT-Infrastructure/World-IPv6-Day-Ends-Everyo...
    World IPv6 Day Ends, Everyone Goes Back to IPv4
    By: Fahmida Y. Rashid
    2011-06-09

    http://www.theregister.co.uk/2011/06/09/ipv6_fails_to_kill/
    World IPv6 Day fails to kill the internet
    Publicity stunt over, now the work begins
    By Kevin Murphy
    9th June 2011 15:47 GMT

    World IPv6 Day was the non-event everybody hoped it would be.

    For 24 hours yesterday, more than 400 major internet companies,
    including Google, Facebook and Microsoft, joined in with the
    largest-scale production test of the next generation Internet
    Protocol to date.

    Organised by the Internet Society, the project was intended to
    raise awareness about the need to start the global transition to
    IPv6 and to enable participants to gather data about potential
    glitches.
    ...

+---------------
| OTOH that really shouldn't make ntpd bomb out. It was eventually exiting
| with a sig 11. Not helpful.
+---------------

Hey, you found a glitch!! Good for you! File a bug report
with the NTP maintainers!! It will be helpful. (Really.)

+---------------
| But I've switched ntp server now, and ntpd is running for
| the time being. Maybe I need to get the upgrade finished pdq :-)
+---------------

Like maybe before June 2012?  ;-}

-Rob

-----

627 26th Avenue         <http://rpw3.org/>
San Mateo, CA 94403

 
 
 

ntpd madness

Post by Rob Warno » Sat, 11 Jun 2011 11:31:19


+---------------
| Yesterday (and today?!?) were "World IPv6 Day", a day for turning
| on IPv6 in a large number of places that don't ordinarily use it
| and seeing what breaks.  ;-}  ;-}
...
|     More information about IPv6 day:
|     http://ipv6day.org
+---------------

That link looks obsolete, not updated since 2006. This is better:

    http://www.worldipv6day.org/
    About World IPv6 Day
    On 8 June, 2011, top websites and Internet service providers around
    the world, including Google, Facebook, Yahoo!, Akamai and Limelight
    Networks joined together with more than 1000 other participating
    websites in World IPv6 Day for a successful global-scale trial of the
    new Internet Protocol, IPv6. By providing a coordinated 24-hour "test
    flight", the event helped demonstrate that major websites around the
    world are well-positioned for the move to a global IPv6-enabled
    Internet, enabling its continued exponential growth.
    ...

-Rob

-----

627 26th Avenue         <http://rpw3.org/>
San Mateo, CA 94403

 
 
 

ntpd madness

Post by Mike Scot » Sat, 11 Jun 2011 22:57:47



...

Quote:> +---------------
> | OTOH that really shouldn't make ntpd bomb out. It was eventually exiting
> | with a sig 11. Not helpful.
> +---------------

> Hey, you found a glitch!! Good for you! File a bug report
> with the NTP maintainers!! It will be helpful. (Really.)

But as an old version, it's probably already 'fixed in the next
release'. I'll see soon (the upgrade is on the way. Amazing the hassles
when you upgrade a 4?-year-old system and jump 2 major versions of the OS.)

And thanks for the note about IPv6 day. I'll check some of those links.
Not, I suspect, that my ISP will change any time soon :-{

--
Mike Scott (unet2 <at> [deletethis] scottsonline.org.uk)
Harlow Essex England

 
 
 

ntpd madness

Post by Rick Jone » Thu, 16 Jun 2011 07:51:13



> I'll bet that one or more of the servers mentioned in your
> /etc/ntp.conf file now has an IPv6 address that it didn't have the
> last time you booted the machine.

And the ntp/ntpdate code in question is making a getaddrinfo() call
not specifying a particular address family, but not setting
AI_ADDRCONFIG to tell the call "don't bother with addresses in a given
family unless I have a non-loopback interface configured with an
address in that family."

rick jones
--
firebug n, the idiot who tosses a lit cigarette out his car window
these opinions are mine, all mine; HP might not want them anyway... :)
feel free to post, OR email to rick.jones2 in hp.com but NOT BOTH...

 
 
 

1. Dell Poweredge 2600 SMP RH 7.3 ntpd and other problems

Hi people,

I'm having different problems with Dell Poweredge 2600 with
one Xeon 1.8GHz CPU installed.

The timeserver f. e. is having problems staying synchronized. I get time
resets every ~21 minutes on all machines (6 of them). I tryed the RH RPM
and a selfcompiled 4.1.2 as well - same thing.

People on the ntpd list did not have any explanation too. It seems that
the internal clock is somehow running mad.

Jul 31 02:02:35 otello ntpd[3737]: time reset 0.691616 s
Jul 31 02:02:35 otello ntpd[3737]: synchronisation lost
Jul 31 02:22:11 otello ntpd[3737]: time reset 1.128246 s
Jul 31 02:22:11 otello ntpd[3737]: synchronisation lost
Jul 31 02:44:08 otello ntpd[3737]: time reset 0.759849 s
Jul 31 02:44:08 otello ntpd[3737]: synchronisation lost

The other thing ist that sometimes one single process is taking all the
CPU time of one logical CPU. Once it was a crond the other time the rngd
from the rng-tools. Very mysterious...

Any hints on that?

The BIOS is latest (A09), kernel 2.4.21 - I tried other kernels as well
- no change. The other hardware is standard.

Regards,

Edin

2. squid and ipfw

3. Cannot get ntpd to work on redhat Linux 7.2

4. KDE install error

5. NTPD question / problem

6. alan@redhat.com: Linux 2.4.20-pre7-ac2 - How's the IDE thing coming?

7. ntpd - ports problem ?

8. Konqueror Javascript /w Suse 6.4

9. NTP daemon works for a bit, then "ntpd dead but subsys locked"

10. Why can't I run ntpd from /usr/local/bin

11. ntpd not adjusting time

12. help with NTPD

13. ntpd will not set my machines date/time