I have recently built a new server using CentOS4. I've found one major oddity: time resolution is completely whacked.
I'm not using an external time source nor am I relying on NTP for time resolution. For some reason, time on this server moves forward at a far more rapid rate.
Right now, all my other systems show Sunday May 8, 3:19 pm. The CentOS server thinks its Monday May 9, 3:54 pm. I just reset it about four hours ago to the correct time.
I'm using an inexpensive eMachines T6212 AMD Athon 64 processor and 1 GB RAM.
Can anyone suggest a reason why this server thinks it's 25 hours in the future and growing more distant with each minute?
Thanks, Bill
On Sun, 2005-05-08 at 15:21 -0400, Bill Diamond wrote:
I have recently built a new server using CentOS4. I've found one major oddity: time resolution is completely whacked.
I'm not using an external time source nor am I relying on NTP for time resolution. For some reason, time on this server moves forward at a far more rapid rate.
Right now, all my other systems show Sunday May 8, 3:19 pm. The CentOS server thinks its Monday May 9, 3:54 pm. I just reset it about four hours ago to the correct time.
I'm using an inexpensive eMachines T6212 AMD Athon 64 processor and 1 GB RAM.
Can anyone suggest a reason why this server thinks it's 25 hours in the future and growing more distant with each minute?
There have been reports of AMD64 machines revving their clock faster than normal, so you're not alone. Check the Fedora mailing lists as well.
Bill Diamond wrote:
I have recently built a new server using CentOS4. I've found one major oddity: time resolution is completely whacked.
I'm not using an external time source nor am I relying on NTP for time resolution. For some reason, time on this server moves forward at a far more rapid rate.
Right now, all my other systems show Sunday May 8, 3:19 pm. The CentOS server thinks its Monday May 9, 3:54 pm. I just reset it about four hours ago to the correct time.
I'm using an inexpensive eMachines T6212 AMD Athon 64 processor and 1 GB RAM.
Can anyone suggest a reason why this server thinks it's 25 hours in the future and growing more distant with each minute?
Sounds like it's got a defective hardware clock chip (not a software issue at all). If it annoys you, you could always just get a new one (computer or mboard). Or just set up a cron job to run once an hour and use ntp to set the hardware clock back to the correct time.
Cheers,
C
On May 8, 2005, at 3:21 PM, Bill Diamond wrote:
I'm not using an external time source nor am I relying on NTP for time resolution. For some reason, time on this server moves forward at a far more rapid rate.
if you don't mind me asking, why are you not using NTP? it seems an inoffensive enough mechanism, and if there are good reasons not to use it, i'd be interested in hearing them.
-steve
p.s. i suspect that the reason why you are experiencing time problems with this particular system is because Madness is Taking Its Toll.
--- If this were played upon a stage now, I could condemn it as an improbable fiction. - Fabian, Twelfth Night, III,v
Bill Diamond wrote:
I have recently built a new server using CentOS4. I've found one major oddity: time resolution is completely whacked.
I'm not using an external time source nor am I relying on NTP for time resolution. For some reason, time on this server moves forward at a far more rapid rate.
Right now, all my other systems show Sunday May 8, 3:19 pm. The CentOS server thinks its Monday May 9, 3:54 pm. I just reset it about four hours ago to the correct time.
I'm using an inexpensive eMachines T6212 AMD Athon 64 processor and 1 GB RAM.
Can anyone suggest a reason why this server thinks it's 25 hours in the future and growing more distant with each minute?
Thanks, Bill
I don't know about your 64 bit processor but I had a similar problem with a 32 bit Athlon XP 2600+ and Fedora Core 3. Turned out to be the SMP kernel that was installed as GRUB's default. It was so bad that NTP never settled enough to create a drift file. When I changed the default to non-SMP, the clock resumed keeping time properly. Yeah, I've read several times that it's O.K. to run SMP kernels with single processors....
-
Bill Diamond -
Chris Mauritz -
Ignacio Vazquez-Abrams -
Robert -
Steve Huff