later kernels vs ntpd

later kernels vs ntpd

[Gene Heskett]

Greetings;

I'm observing that the last time ntpd logged that it was synchronized
with a time src, was:

6 Aug 17:44:36 ntpd[1886]: synchronized to 140.221.8.88, stratum=1

This was about a week before I left for month long trip, and I left it
running 2.6.13-rc6 which had been stable for several days prior to my
leaving.  In early September, I had to lead the missus thru a reboot
because the system time had jumped to some other time zone and the
heyu based lighting automation was afu.  That seemed to fix it till I
got back on the 14th.  At that time I noted that my watch appeared to be 
off
by several minutes.  The next day I did a 'service ntpd restart',
which backed the system time up about 4 minutes, bringing my watch a
lot closer.

Then I noticed that yesterday, before I built and installed 2.6.13.1,
that ntpd was apparently not synching.  A restart, which runs ntpdate
to crash set the clocks, does work, but ntpd is not.  I haven't
changed anything in the configs for ntpd in several months.

Historicly, this box has lost track of what time zone its on on
several occasions, but I believe this is a seperate problem.

Is anyone else having ntpd synch problems?  Take a look at
your /var/log/ntpd.log just for grins.  All I'm getting is the
restart messages, like these:

18 Sep 10:41:44 ntpd[17787]: ntpd exiting on signal 15
18 Sep 10:37:07 ntpd[29714]: running as uid(38)/gid(38)
euid(38)/egid(38).

>From the restart I just did, note the time reset of over 4 minutes.

>From the dates on my /boot/vmlinuz* files, it worked for 2.6.13-rc5,
but not for rc6 & (apparently) later.

Comments anybody?

I'm going to reboot to an older kernel (2.6.13-rc5) as a check, if
ntpd starts working, I'll advise here.

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.35% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.

Re: later kernels vs ntpd

[Gene Heskett]

On Sunday 18 September 2005 10:46, Gene Heskett wrote:
>Greetings;
>
>I'm observing that the last time ntpd logged that it was synchronized
>with a time src, was:
>
>6 Aug 17:44:36 ntpd[1886]: synchronized to 140.221.8.88, stratum=1
>
>This was about a week before I left for month long trip, and I left it
>running 2.6.13-rc6 which had been stable for several days prior to my
>leaving.  In early September, I had to lead the missus thru a reboot
>because the system time had jumped to some other time zone and the
>heyu based lighting automation was afu.  That seemed to fix it till I
>got back on the 14th.  At that time I noted that my watch appeared to be
>off
>by several minutes.  The next day I did a 'service ntpd restart',
>which backed the system time up about 4 minutes, bringing my watch a
>lot closer.
>
>Then I noticed that yesterday, before I built and installed 2.6.13.1,
>that ntpd was apparently not synching.  A restart, which runs ntpdate
>to crash set the clocks, does work, but ntpd is not.  I haven't
>changed anything in the configs for ntpd in several months.
>
>Historicly, this box has lost track of what time zone its on on
>several occasions, but I believe this is a seperate problem.
>
>Is anyone else having ntpd synch problems?  Take a look at
>your /var/log/ntpd.log just for grins.  All I'm getting is the
>restart messages, like these:
>
>18 Sep 10:41:44 ntpd[17787]: ntpd exiting on signal 15
>18 Sep 10:37:07 ntpd[29714]: running as uid(38)/gid(38)
>euid(38)/egid(38).
>
>>From the restart I just did, note the time reset of over 4 minutes.
>
>>From the dates on my /boot/vmlinuz* files, it worked for 2.6.13-rc5,
>but not for rc6 & (apparently) later.
>
>Comments anybody?
>
>I'm going to reboot to an older kernel (2.6.13-rc5) as a check, if
>ntpd starts working, I'll advise here.

Ok, on rebooting to 2.6.13-rc5, time synch was achieved in
approximately 5 minutes after the reboot.  So apparently -rc6 and
13.1 are broken for ntpd.

Sorry your coal mine canary was asleep on the job, to let this go on
for over a month without a chirp.  But I've been out of state, trying
to make a junkyard tv station work, in a market that cannot support a
real tv station.

Does anyone have a hint?

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.35% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.

Re: later kernels vs ntpd

[Gene Heskett]

On Sunday 18 September 2005 10:59, Gene Heskett wrote:
>On Sunday 18 September 2005 10:46, Gene Heskett wrote:
>>Greetings;
>>
>>I'm observing that the last time ntpd logged that it was synchronized
>>with a time src, was:
>>
>>6 Aug 17:44:36 ntpd[1886]: synchronized to 140.221.8.88, stratum=1
>>
>>This was about a week before I left for month long trip, and I left it
>>running 2.6.13-rc6 which had been stable for several days prior to my
>>leaving.  In early September, I had to lead the missus thru a reboot
>>because the system time had jumped to some other time zone and the
>>heyu based lighting automation was afu.  That seemed to fix it till I
>>got back on the 14th.  At that time I noted that my watch appeared to
>> be off
>>by several minutes.  The next day I did a 'service ntpd restart',
>>which backed the system time up about 4 minutes, bringing my watch a
>>lot closer.
>>
>>Then I noticed that yesterday, before I built and installed 2.6.13.1,
>>that ntpd was apparently not synching.  A restart, which runs ntpdate
>>to crash set the clocks, does work, but ntpd is not.  I haven't
>>changed anything in the configs for ntpd in several months.
>>
>>Historicly, this box has lost track of what time zone its on on
>>several occasions, but I believe this is a seperate problem.
>>
>>Is anyone else having ntpd synch problems?  Take a look at
>>your /var/log/ntpd.log just for grins.  All I'm getting is the
>>restart messages, like these:
>>
>>18 Sep 10:41:44 ntpd[17787]: ntpd exiting on signal 15
>>18 Sep 10:37:07 ntpd[29714]: running as uid(38)/gid(38)
>>euid(38)/egid(38).
>>
>>>>From the restart I just did, note the time reset of over 4 minutes.
>>>
>>>>From the dates on my /boot/vmlinuz* files, it worked for 2.6.13-rc5,
>>
>>but not for rc6 & (apparently) later.
>>
>>Comments anybody?
>>
>>I'm going to reboot to an older kernel (2.6.13-rc5) as a check, if
>>ntpd starts working, I'll advise here.
>
>Ok, on rebooting to 2.6.13-rc5, time synch was achieved in
>approximately 5 minutes after the reboot.  So apparently -rc6 and
>13.1 are broken for ntpd.
>
>Sorry your coal mine canary was asleep on the job, to let this go on
>for over a month without a chirp.  But I've been out of state, trying
>to make a junkyard tv station work, in a market that cannot support a
>real tv station.
>
>Does anyone have a hint?

Someone suggested that I needed the ipv6 stuff, so I built a 2.6.13.1
with all that turned on, but an overnight run of it also failed to
synch, so I'm back on 2.6.13-rc5, which works well.

But this is preventing me from playing the coal mine canary.  Does
anyone else have a suggestion?

-- 
Cheers, Gene
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
99.35% setiathome rank, not too shabby for a WV hillbilly
Yahoo.com and AOL/TW attorneys please note, additions to the above
message by Gene Heskett are:
Copyright 2005 by Maurice Eugene Heskett, all rights reserved.

Re: later kernels vs ntpd

[Sid Boyce]

Gene Heskett wrote:
 > Someone suggested that I needed the ipv6 stuff, so I built a 2.6.13.1
 > with all that turned on, but an overnight run of it also failed to
 > synch, so I'm back on 2.6.13-rc5, which works well.
 >
 > But this is preventing me from playing the coal mine canary. Does
 > anyone else have a suggestion?

Right up to 2.6.13-rc6-git12 I've used 
linux-2.6.13-rc6_timeofday-all.patch with no drift at all. With 
unpatched kernels (won't apply) up to 2.6.14-rc1-git5 on SuSE 9.3, I'm 
seeing -3 secs drift since reboot 1 hour ago, I think that's the max 
I've seen over several hours, I have a heat/hardware problem that causes 
a solid lockup at random times.
On the other box with the same hardware and Mandriva LE2005, no drift.
bumble:/root # ptktime&
[1] 27567
bumble:/root # localhost                       0 Mon Sep 19 22:35:58 2005
128.118.25.3                    0 Mon Sep 19 22:35:58 2005

bumble:/root # uptime
  22:36:15 up 22:35,  3 users,  load average: 0.00, 0.00, 0.00
bumble:/root # uname -r
2.6.14-rc1-git4

Regards
Sid.
-- 
Sid Boyce ... Hamradio License G3VBV, licensed Private Pilot
Retired IBM/Amdahl Mainframes and Sun/Fujitsu Servers Tech Support 
Specialist
Microsoft Windows Free Zone - Linux used for all Computing Tasks