Hiya,
I'd like to recommend setting the hardware clock in the NTP scripts. A
simple call to "hwclock --systohc" (add --utc if applicable) should
suffice. A good place would be the kill script (reasoning that xntpd has
had sufficient time to synchronize). In case step-tickers are available, it
can be set in the startup scripts.
The reasoning is that if the hardware clock and the real time differ a lot,
there is a period during booting where the time is vastly wrong, but this
time difference is not noted during normal operations (because ntp has
kicked in). A number of daemons and programs that process time behave in a
very weird way of they are started in "the future". To give just two
examples, cron will stop executing cron scripts until the time it was
`really' started has been reached again, and the process accounting tools
will print completely wrong results.
Regards,
---Ingo Luetkebohle / 21st Century Digital Boy
Confident, cocky, lazy, dead.
[J. Dread, Otherland]
--
To unsubscribe:
mail -s unsubscribe [EMAIL PROTECTED] < /dev/null
