Clock does not advance

Hi,

we have two Linux boxes running Redhat 7.1 kernel version
2.4.18-18.7.xsmp running on HP Netserver L 3000. They are in use as
apache web servers and also have some mail and dns tasks. Both servers
ran very stable for the last year.
In spring we noticed a strange behaviour in one box. The clock wasn't
advancing. This means the clock was for exmple at 10:03:05 after three
seconds it moved to 10:03:08 as it should. But after other three
seconds it stepped back to 10:03:05. We normally use ntp to
synchronize our clocks and disabled ntp to check if there is a
difference, but there wasn't. I turned off nearly all processes in the
server, but didn't localize the error.
After rebooting the machine this behaviour had been gone, but
reappeared after a month or so and now it is a monthly (irregular)
problem.
Now after four month it appeared for the first time on the second box.
Therfore I am asking you:
- anybody knows this bug
- has a remedy
- knows if it is a hardware, a OS problem or a faulty daemon

Thanks in advance
Stefan

In comp.os.linux.misc Stefan Wenzel <swenzel@aspherio.com> wrote:
> _after_ this behaviour arise. What do you mean with the syslogd
> daemon. I thought it is only for writing in the log files?

Yes, but it does also write a "mark" when the machine is doing
nothing, just to be sure that is alive.

Davide