nigel5
asked on
Fedora Server clock keeps stopping.
Hi there.
I recently had Redhat 7.3 running on my IBM Netvista without a problem. I installed Fedora Core 3 (non graphical standard install) on a new hard drive in the machine.
All went well, except the clock seemed out by an hour, Ah, I thought, DST... nope, the clock had stopped... or would oscilate within a couple of seconds... or jump wildly to a year ago...
I have installed NTP in an attempt to kee pthe clock in sync, but hat is relying on the timer which is stuffed.
What is going on, and how do I fix this.
I recently had Redhat 7.3 running on my IBM Netvista without a problem. I installed Fedora Core 3 (non graphical standard install) on a new hard drive in the machine.
All went well, except the clock seemed out by an hour, Ah, I thought, DST... nope, the clock had stopped... or would oscilate within a couple of seconds... or jump wildly to a year ago...
I have installed NTP in an attempt to kee pthe clock in sync, but hat is relying on the timer which is stuffed.
What is going on, and how do I fix this.
Change the cmos battery of your PC (should be 3V).
As root, run
ntpdate clock.redhat.com (sync your system clock with Redhat NTP server)
hwclock -w ( sync your HW clock with system clock)
If your clock still off or drift very often, then
1. Change the motherboard battery as rindi's suggestion.
2. As root, do
crontab -e
1/15 * * * * /usr/sbin/ntpdate clock.redhat.com; /sbin/hwclock -w > /dev/null 2>&1
To sync the clock every 15 minutes.
ntpdate clock.redhat.com (sync your system clock with Redhat NTP server)
hwclock -w ( sync your HW clock with system clock)
If your clock still off or drift very often, then
1. Change the motherboard battery as rindi's suggestion.
2. As root, do
crontab -e
1/15 * * * * /usr/sbin/ntpdate clock.redhat.com; /sbin/hwclock -w > /dev/null 2>&1
To sync the clock every 15 minutes.
ASKER
I cannot run ntpdate - it just hangs.
NTP is also running with a frequency of 64 seconds, so the crontab entry is not required.
I'll get another battery and give the other bits a go.
Thanks
Nigel.
NTP is also running with a frequency of 64 seconds, so the crontab entry is not required.
I'll get another battery and give the other bits a go.
Thanks
Nigel.
ASKER
I am not sure it is the battery...
The 'real' time as per my working machine is 14:15:03..
[root@rdspace ~]# hwclock
Wed 11 May 2005 13:15:03 BST -0.000301 seconds
[root@rdspace ~]# date
Wed May 11 13:10:56 BST 2005
[root@rdspace ~]# ntpdate -u clock.redhat.com
11 May 15:09:25 ntpdate[3581]: step time server 66.187.224.4 offset 7109.793246 sec
[root@rdspace ~]# hwclock
Wed 11 May 2005 14:09:23 BST -0.000466 seconds
[root@rdspace ~]# date
Wed May 11 15:09:25 BST 2005
Then trying the time again at 15:20:41
[root@rdspace ~]# hwclock
Wed 11 May 2005 14:20:41 BST -0.000173 seconds
[root@rdspace ~]# date
Wed May 11 15:09:25 BST 2005
I'm just wondering whether the issue is because the hardware clock is showing GMT, but reporting it as BST??? whether that would screw up things with NTP? and the internal clock drift???
The 'real' time as per my working machine is 14:15:03..
[root@rdspace ~]# hwclock
Wed 11 May 2005 13:15:03 BST -0.000301 seconds
[root@rdspace ~]# date
Wed May 11 13:10:56 BST 2005
[root@rdspace ~]# ntpdate -u clock.redhat.com
11 May 15:09:25 ntpdate[3581]: step time server 66.187.224.4 offset 7109.793246 sec
[root@rdspace ~]# hwclock
Wed 11 May 2005 14:09:23 BST -0.000466 seconds
[root@rdspace ~]# date
Wed May 11 15:09:25 BST 2005
Then trying the time again at 15:20:41
[root@rdspace ~]# hwclock
Wed 11 May 2005 14:20:41 BST -0.000173 seconds
[root@rdspace ~]# date
Wed May 11 15:09:25 BST 2005
I'm just wondering whether the issue is because the hardware clock is showing GMT, but reporting it as BST??? whether that would screw up things with NTP? and the internal clock drift???
> I cannot run ntpdate - it just hangs.
when the ntp is running, ntpdate won't work since they use the same port.
Please run
hwclock -w
to force your HW clock sync with your system clock FIRST.
when the ntp is running, ntpdate won't work since they use the same port.
Please run
hwclock -w
to force your HW clock sync with your system clock FIRST.
ASKER
I finally solved it, nothing to do with the battery, or NTP..
I reinstalled RH7.3 all was fine.
I reinstalled FC3 - went pear shaped.
I fondeld with lots of things, another reinstall, and then switched kernels.
While my motherboard supports the multi-processor architecture, if I use the SMP kernel, the CPU clocks seem to get muddled up... since I am ony using the one CPU I switched t othe UP (uniprocessor) kernel, and so far, all has been well for 24 hours so far.
still monitoring, although 24 hours is longer than ever left before.
I reinstalled RH7.3 all was fine.
I reinstalled FC3 - went pear shaped.
I fondeld with lots of things, another reinstall, and then switched kernels.
While my motherboard supports the multi-processor architecture, if I use the SMP kernel, the CPU clocks seem to get muddled up... since I am ony using the one CPU I switched t othe UP (uniprocessor) kernel, and so far, all has been well for 24 hours so far.
still monitoring, although 24 hours is longer than ever left before.
> my motherboard supports the multi-processor architecture, if I use the SMP kernel, the CPU clocks seem to get muddled up
Does your CPU has hyperthread feature?
In this case, I consider it is Fedora's bug which SMP shouldn't mess up the system clock.
Does your CPU has hyperthread feature?
In this case, I consider it is Fedora's bug which SMP shouldn't mess up the system clock.
ASKER
The CPU does not have HT technology, it is about 2 years old.
It does seem odd that Redhat would introduce this bug between RH7.3 and FC3.
6 days uptime, and all is fine.
It does seem odd that Redhat would introduce this bug between RH7.3 and FC3.
6 days uptime, and all is fine.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.