HP3000-L Archives

May 2000, Week 1

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Jeff Kell <[log in to unmask]>
Reply To:
Jeff Kell <[log in to unmask]>
Date:
Mon, 1 May 2000 22:10:23 -0400
Content-Type:
text/plain
Parts/Attachments:
text/plain (30 lines)
Ted Ashton wrote:
>
> Thus it was written in the epistle of Jeff Kell,
> >
> > It isn't good netiquette to sync multiple servers, particularly
> > the typical desktop, to a public, high-stratum server.
>
> While I happily agree with the main message Jeff's presenting, I beg > to differ just a little on one detail.  The Network Time Protocol is
> designed to do its best to find what the time really is from
> multiple sources, so in our design here, we have 4 servers which
> each sync off of about 4 public servers and each other.  That gives
> us a solid base which we then use to sync the rest of the
> campus.

OK, to clarify, our 'primary' server syncs to 3 stratum 1 and 3 stratum
2 'public' servers.  The master is almost always one of the
stratum 1's but we do have backup just in case.

We then have 3 local servers which sync to each other plus the one
local 'primary'.  We could add strat 2's to their peer list but I'm
trying not to bother too many of the public servers too much.

Similarly, for ntpdate, we sync to the local primary and the three
peers.  For Dimension4, you can only sync to one, so they generally
use our local primary.  It's a cisco 7505 and it has a fairly stable
clock (we tried using our border router, a 3640, but the higher-end
routers would never sync to it due to clock variations).

Jeff Kell <[log in to unmask]>

ATOM RSS1 RSS2