clock runs very fast on HP DL380 G4 with Redhat AS 4 64bit

murrayphobbs · 11-28-2006, 09:15 PM

Hi

my name is murray

we are running 4 HP ProLiant DL380 G4 Storage Servers - two running Redhat Enterprise AS version 4 64 bit and two running Windows 2003 Server

the linux boxes both exhibit the same problem - the clock runs up to 6 HOURS fast per day

at the moment i have a cronjob running on them both to sync with a local ntp server every 3 minutes

i've had a look around but have not found any reference to a problem of this extreme nature

any suggestions?

thanks

macemoneta · 11-29-2006, 02:47 PM

This is usually solved by adding the kernel boot parameter "clock=pit". For example, in /boot/grub/grub.conf:

Code:

title xxxx (2.6.xx-x.xx.xxx)
        root (hd0,0)
        kernel /vmlinuz-2.6.xx-x.xxxx.xx ro root=LABEL=/ clock=pit quiet
        initrd /initrd-2.6.xx-x.xxxx.xxx.img

This sets the clock to use the programmable interrupt timer, and usually solves the problem (not always).

murrayphobbs · 11-30-2006, 12:10 AM

re the below suggestion, no, thanks for the try, but it makes no apparent difference

murray

Quote:

Originally Posted by macemoneta

This is usually solved by adding the kernel boot parameter "clock=pit". For example, in /boot/grub/grub.conf:

Code:

title xxxx (2.6.xx-x.xx.xxx)
        root (hd0,0)
        kernel /vmlinuz-2.6.xx-x.xxxx.xx ro root=LABEL=/ clock=pit quiet
        initrd /initrd-2.6.xx-x.xxxx.xxx.img

This sets the clock to use the programmable interrupt timer, and usually solves the problem (not always).

macemoneta · 11-30-2006, 01:00 AM

OK, the lesser seen issues are:

- if you are also seeing 50% CPU utilization when the machine is idle, use kernel parameter no_timer_check

- You can try the noapic kernel parameter

- You can try the notsc kernel parameter

In any case, you should also check for a BIOS update.