HP3000-L Archives

November 1999, Week 2

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:
Bruce Senn <[log in to unmask]>
Reply To:
Date:
Thu, 11 Nov 1999 11:34:54 -0500
Content-Type:
Text/Plain
Parts/Attachments:
Text/Plain (58 lines)
We are planning our Y2K test and recovery for next week.  I am interested
in hearing ideas, experiences, and opinions about the preferred way to
recover after testing.

Our system is a 918LX with 4 2.0 Gb drives, all in the system volume set.
This is our only HP3000 so it is the production system.

The tentative testing plan is to use SETCLOCK to set the clock ahead to
the following dates and times, e.g., SETCLOCK DATE=12/31/1999;TIME=23:50.

12/31/1999  23:50  to observe date rollover to 01/01/2000
01/01/2000  09:00  for user testing
02/28/2000  23:50  to observe date rollover to 02/29/2000
02/29/2000  10:00  for user testing
02/29/2000  23:50  to observe date rollover to 03/01/2000
03/01/2000  11:00  for user testing

For recovery there seem to be three alternatives.

1. INSTALL from the current SLT
   Change the date at the prompt.
   Use VOLUTIL to scratch and redefine the system volume set.
   Restore (with KEEP option) all files from our Road Runner backup tapes.

2. Normal START NORECOVERY
   Change the date at the prompt.
   Use MPEX to identify files created, modified, or accessed after
   12/31/1999.  Delete or restore these files.
   Restore user databases from our Road Runner backup tapes.

3. Normal START NORECOVERY
   Change the date at the prompt.
   Use FIXDATE from Allegro to set any future dates to today.
   Restore user databases from our Road Runner backup tapes.

#1 seems to be the cleanest, but the risk of a bad tape has severe
   consequences.

#2 seems to be the most straightforward, but runs the risk of some system
   or log files still having future dates.

#3 seems to depend on FIXDATE, but might provide an audit trail for those
   files changed during the test, i.e., they would all have a timestamp
   shortly after the restart.

All ideas, experiences, opinions (even rants, I suppose) welcome.

TIA.

Bruce.

----------------------------------------------------------------------
   Bruce J. Senn                             Phone:  (518) 388-6664
   Senior System Manager                     FAX:    (518) 388-6458
   Union College                   Internet:  [log in to unmask]
   Schenectady, NY 12308
----------------------------------------------------------------------

ATOM RSS1 RSS2