HP3000-L Archives

December 1998, Week 3

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:
Ron Burnett <[log in to unmask]>
Reply To:
Ron Burnett <[log in to unmask]>
Date:
Fri, 18 Dec 1998 08:29:45 +1100
Content-Type:
text/plain
Parts/Attachments:
text/plain (55 lines)
At 11:19 17/12/1998 -0500, Chris Enderle (IBM Global Services)  wrote:

>We have converted our data and applications to accomodate 8 digit dates.
>We tested with Time Machine and Time Shift and everything is performing
>normally.   Our customer has asked us to perform a 'real' simulation by
>booting the system with a date in the year 2000 and then testing system and
>application software.  Has anyone done this before?

 We have.  As people will understand, in a health care environment (tremendous
pressure for a clean 7*24 shop), we chose not to disturb our production
environment any more than necessary, so we built up a crash & burn 928
and loaded most core medical applications on it.  We turned off user logging
on the 928 copies, and then got stuck into Y2K renovations.  We're about
three-quarters the way through (roughly 16,000 programs, many TI DBs).

A couple of weeks ago, I validated the previous nights full backup, then
created a TEST account, and copied half a dozen files into it of various
types.
Then I rebooted the system and changed the clock to 12 December 1999,
23:50 hours.  We watched as the clock ticked over into 1 January 2000.

We made a few changes to the test files, and then looked at the directory
entries.  Just as you would expect.  We tried all our third-party software
(COGNOS, MPEX, QEDIT, ADAGER).  Adager choked on a license
expiry date in 1999, but the other three were fully functional and happy with
the sysdate.  This is NOT a problem, since we don't plan to do any database
therapy or transformation while the clock is rolled forward, and we're
absolutely
confident that Alfredo will get us updated well before we need it (or
anytime we
ask for it).

Our applications performed exactly as we expected.  There is no drama or magic
in handling a century-included date as it moves across the century boundary.

We don't have to analyse our system log files on the 928.  So we just rebooted
the machine and set the clock back to correct time.  MPE/iX doesn't mind if
a file has a date/timestamp into the future.  But it will happily change
them back to
current time even if the current time is before the file's existing
date/timestamps,
when it needs to.  That's what it should do.

As I said to our CIO, it was all a bit boring.  No blood dripped from the
ceiling,
no heads spun through 360 degrees, no giant bugs leapt out of the monitors and
dragged us into the nether regions.  Just boring, predictable, accurate
computing.

FWIW,
Ron Burnett
Manager, HP3000 Systems
Women's & Children's Healthcare Network
[log in to unmask]

ATOM RSS1 RSS2