Weird system clock problem...
This is a resend... I think my first e-mail didn't go through.
Since I upgraded this server to the 2.6 kernel branch, I have this,
weird, and HEAVY system clock problem. This machine was running
perfectly using a 2.4 kernel. In less than 24 hours, the system clock
seems to get stuck for a totally unkown reason. I synchronised the
system clock yesterday morning at 9h10 after a reboot and everything
went fine until afternoon/night.
Now it's 10h00, nov 8, and if I repeatly type the date command, I obtain
this:
emedia:~# date
Tue Nov 7 17:16:05 EST 2006
emedia:~# date
Tue Nov 7 17:16:02 EST 2006
emedia:~# date
Tue Nov 7 17:16:02 EST 2006
emedia:~# date
Tue Nov 7 17:16:03 EST 2006
emedia:~# date
Tue Nov 7 17:16:05 EST 2006
Reading the hardware clock, it was about half an hour late:
emedia:~# hwclock --show
Wed 08 Nov 2006 09:28:31 AM EST -0.494898 seconds
The kernel I'm running:
Linux 2.6.17-2-686 #1 SMP Wed Sep 13 16:34:10 UTC 2006 i686 GNU/Linux
And the server is running debian Etch.
I definetly need this to get fixed.... I can't rollback to 2.4 kernel
since Apache 2.2 won't run on it (using mpm-prefork).
I don't know if this can help to diagnose my problem, but here's the
content of adjtime I had:
-679.643128 1162907374 0.000000
1162907374
UTC
I removed this file, and it didn't help anything. I also installed the
package ntp, and it's still doing that after perhaps 20 hours of uptime.
Thanks to you all!
--
Louis-David Perron
ldperron@ldperron.com
«Sans la musique, la vie serait une erreur» - Nietzsche
Reply to: