On Thu, Oct 12, 2006 at 12:13:13AM +0200, Rob van der Heij wrote:
> On 10/11/06, Marcy Cortes <[EMAIL PROTECTED]> wrote:
>
> >Sure, it wastes a little, but it doesn't look that bad here (we have to
> >run NTP on every server to sync security tickets and stuff). Velocity
> >reports the idle ones at 0.01% of a CPU, and that's with their agent
> >presumbaly doing a little stuff too).
>
> IIRC the NTP mechanism was to review adjusting the change of the drift
> every 2 seconds or so. Even though this is very little work, it does
> make VM think the guest is busy and keeps it in queue. Asking the snmp
> agent every minute for some MIBs (when the guest did real work) is an
> order of maginute less annoying for VM.
No. Thats not true. When starting NTP looks quite often (dont remember how
often).
The more the reference clock and the local clock are in sync the less NTP
checks the state. This goes down to a check every few hours.
So after a while NTP should not be visible at all from a VM point of view.
As an example an excerpt from our server:
23 Sep 13:46:39 ntpd[5626]: kernel time sync disabled 0041
23 Sep 13:47:43 ntpd[5626]: kernel time sync enabled 0001
23 Sep 13:47:50 ntpd[5626]: synchronized to xxx.yyy.160.50, stratum 2
23 Sep 15:50:13 ntpd[5626]: synchronized to xxx.yyy.160.1, stratum 2
23 Sep 16:59:01 ntpd[5626]: synchronized to xxx.yyy.160.50, stratum 2
24 Sep 10:36:43 ntpd[5626]: synchronized to xxx.yyy.160.54, stratum 2
26 Sep 11:34:20 ntpd[5626]: synchronized to xxx.yyy.160.50, stratum 2
26 Sep 20:06:12 ntpd[5626]: synchronized to xxx.yyy.160.54, stratum 2
>
> I think the Kerberos stuff requires same time within 30 seconds.
> That's not a very strong requirement and could allow for some drift.
> If your applications can afford stepping back time a little, then an
> ntpd -q via cron might be good enough.
This is a hack and a bad idea too. Time is a value that should constantly grow
and have _no_ jumps. If your cronjobs executes at 4:00 clock and sets the time
to 3:55
you will have 4:00 two times. This may confuse application and other programms.
NTP is increasing or decreasing the speed of the clock and thus garanties that
you will not have two times the same timestamp.
The use of NTP is twofolded:
- correct a drift of the kernelclock.
This should be no problem on s390. If there is a drift, please
file a bug.
- synchronize multiple computers to have the same time
This is a real usecase for NTP on linux on s390.
If you want to compare timestamps on zLinux with timestamps
on other computers (none zSeries).
Think of a zLinux acting as a fileserver and PCs as clients.
All should have the same time, when you rely on timestamps
(Think of make).
Best regards/Mit freundlichen Grüßen
Ihno Krumreich
"Never trust a computer you can lift."
--
Ihno Krumreich [EMAIL PROTECTED]
SUSE LINUX Products GmbH Projectmanager S390 & zSeries
Maxfeldstr. 5 +49-911-74053-439
D-90409 Nürnberg http://www.suse.de
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
