Greetings;
First, go to
http://www.eecis.udel.edu/~mills/
and you will find several links concerning time near the
lower right corner of Prof. Mills' page. This includes a
list of publicly accessible stratum 2 time servers.
Second, do a google search on time server and you
will, as usual, find way more info than you wanted to
know.
The usual way of synching time is as you suggested.
Set up one system, be it linux/390 or other, that synchs
with three or four stratum2 time servers. Then all other
systems at your site can synch to that. This is not a
high bandwidth activity! Both the request and response
packets are very small, in comparison to other activities,
and it is an infrequent event, by design. Unfortunately
VM, and the S/390 in general I think, will not
play with others in this regard.
I have one linux guest that synchs with three nearby
stratum 2 time servers. Then the other linux guests as
well as my desktop and the other sysprog PCs synch to
it.
I think most of the software you need for linux is included
in the distribution. For windows get Dimension 4 at
http://www.thinkman.com/dimension4/index.html
I have been using it for actually years and years without
problem. And it is free!
Good Luck!
Dennis
Rob van der
Heij To: [EMAIL PROTECTED]
<[EMAIL PROTECTED] cc:
> Subject: Re: NTP Clients in a Linux390
server farm?
Sent by: Linux
on 390 Port
<[EMAIL PROTECTED]
ARIST.EDU>
09/11/02 02:12
AM
Please respond
to Linux on 390
Port
At 00:41 11-09-02, Len Thomson wrote:
>1. There seems to be no nice way to get VM to maintain it's time synch
>2. There seems to be no nice way to get a Linux guest to synch with the VM
>time
>3. I really don't want NTP clients all popping up at the same time to do a
>network synch on every virtual server...
I admit I only briefly looked at the latest code, but afaik
especially with the 'demand timer patch' Linux uses the TOD
to tell time, and I don't recall seeing the drift corrections
applied. This would mean an ntpd cannot adjust time of Linux
running on z/VM, but one might be able to change that. I can
even imagine hwclock to issue SCK instruction to update the
virtual machine's private offset of the VM TOD clock (to
live through a reboot, but not logoff).
My understanding is that the S/390 TOD does not drift much,
but we do have a problem of the operator setting time on VM
by looking on his $10 Mickey Mouse watch.
You're very right that you dont't want them all to measure
time separately by talking to some remote ntpd. Howerver, I
did see there is a lot of smartness in ntpd to increase the
time between probes when it finds there is little drift. You
could imagine one local ntpd to talk to a few good clocks,
and have your local systems use that single server.
It's on my list of things to do, so I would be interested in
what others do (that is the fastest way these days for things
from my list to get done ;-)
Rob
