[Info-vax] SYS$TIMEZONE.DAT

[lists at openmailbox.org]

Thanks for answering this. That's very interesting. I can't remember when
IBM didn't have an external time reference so I looked it up now and the
announcement was only in 1990, much later than I remembered.

Anyway it's surprising VMS clustering can tolerate clocks that are out of
whack. Today that's a good feature for security given how attacks on time
sources can wreak havoc with crypto and much else. But it also calls into
question how well some of those systems could work across VMS cluster nodes
since they all seem to rely on synchronized clocks.

On Mon, 16 Mar 2015 11:15:06 +0100 Paul Sture via Info-vax
<info-vax at rbnsn.com> wrote:

> On 2015-03-16, <lists at openmailbox.org> <lists at openmailbox.org> wrote:
> > On Sun, 15 Mar 2015 13:37:21 -0400
> > David Froble via Info-vax <info-vax at rbnsn.com> wrote:
> >
> >> Yes, but maybe the pertinent question is, "how many users need
> >> accurate time?"
> >
> > VMS clustering can work without accurate time across the cluster?
> 
> Yes.  The place where problems become evident is the batch system, but
> VMS clustering can work quite happily when times across members are not
> in sync.  The hardware clocks on different systems usually go out of
> sync in a gentle fashion, which isn't a big deal for clustering itself.
> 
> It's important to remember here that clustering came long before
> connectivity to an external time source was common.  A former colleague
> did rig up a radio source to implement a home grown time synchronisation
> service, but that was later rendered ineffective by relocating the
> systems in a deep cellar :-)

-- 
Please DO NOT COPY ME on mailing list replies. I read the mailing list.
RSA 4096 fingerprint 7940 3F02 16D3 AFEE F2F8  ACAA 557C 4B36 98E4 4D49