Summary: y2k test / restore question

From: Martin Meadows (PZXKYS@atdva3.atd.gmeds.com)
Date: Thu Feb 04 1999 - 14:34:26 CST


Many thanks for all the great responses to my question about Y2K
testing on Sun Sparcstations.

Here's a summary of the responses I received from the following
folks: Steve Elliott, Elan Finch (Sun tech rep), David Harrington,
Colin Melville, Brion Leary, Michelle Clark, Alan Hill, Greg Obremski,
Jay Morgan, Timothy Symes and Scott Iseminger. My apologies to
anyone I may have missed!

Summary
-------
Responses & feelings seemed to vary quite a lot on this subject.

Several recommendations were made to restore from tape. This
response is consistent with the /www.sun.com/y2000/devguide.html
and the /www.sun.com/y2000/testguide.html that Elan Finch pointed
me to. Note, however, that the Sun devguide.html page does not
state that this step is mandatory. It only suggests it.

Several responses were made in which the Sun managers declared
that they had performed the clock/date tests without restoring
the system from tape afterward. They cautioned, however, that
the tests should be run in single user mode & all 3rd party
applications should be shut down. They also strongly recommended
that all network connections be stopped during the tests.
Furthermore, they pointed out that cron should be turned off
before the tests begin.

One comment that I received suggested that it was not necessary
to run these tests at all, on the grounds that, "... Sun has
obviously done extensive tests themselves and has produced a set
of patches ... What's wrong with assuming that Sun will have made
all the necessary fixes? It's not like your proposed tests do
anything obscure that Sun might not have thought of."

While I tend to agree with the above statement, I don't really
have much choice in the matter. Management wants the "warm
fuzzy".

Also, there was one interesting caution that,
"It may not be as simple as just restoring from tape should something
go wrong. If the date gets burnt into the system board then you will
be dealing with a hardware problem where as you can't get the system
to boot even to the point of changing the date back."

I wonder if anyone out there can actually elaborate on the chance
that this could happen? I don't quite see how the date could get
burned into the system board.

Finally, I received the following caution:
"Sun (and others) are still comming out with Y2K patches although
the latest ones may not be as critical. You need to get the very
latest and greatest before testing, and it may be wise to
hold off a bit longer to give Sun a chance to throughly update the
Y2K patch list."

I have to proceed. We have deadlines. If Sun adds more y2k patches
at this point, I guess we'll have to go back & re-test...
I sure hope they're really on top of things!

Here's a copy of the original question:

>Hi. I'm looking for some sound advice regarding Y2K testing of a
>sparcstation 2 running SunOS 4.1.3_u1 vB. I have several checks
>that I intend to run and am concerned that after the tests are
>complete I may have to restore the entire system from tape. Can
>anyone offer suggestions or helpful comments? It's not clear to
>me what kinds of problems I might have if I run the date tests
>and simply reset the clock back to the correct time when I'm
>finished. To keep this simple, let's say that there are no third
>party applications on this system.
>
>The tests I intend to run are listed below for anyone interested
>in the details. Comments about the tests are certainly welcome.
>
>Thanks very much,
>Martin Meadows
>Indianapolis, IN.
>
>
>ROLLOVER, REBOOT, DAY OF WEEK TESTS
>-----------------------------------
>
>1. Rollover - 1999 to 2000 - Power On
>
> Set the date to 31-dec-1999
> Set the time to 23:59
> Observe the system date after 00:00AM
>
>
>2. Day of week
>
> Verify the date is 1-Jan-2000 from previous test
> Observe the system day of week display
>
>
>3. Reboot - Date Retention
>
> Verify the date is 1-Jan-2000 from previous test
> Power down the system
> Power up the system
> Observe the system date
>
>
>4. Rollover - 1999 to 2000 - Power off
>
> Set the date to 31-Dec-1999
> Set the time to 23:50
> Power down the system before it can roll over to year 2000
> Wait until after "midnight" with the power off
> Power up the system
> Observe the system date
>
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
>MANUAL DATE SET TEST
>
>5. Date set - 1 Jan 2000
>
> Set the date to 1-Jan-2000
> Observe the system date.
>
>
>6. Date set - Date retention
>
> With the date still in the year 2000, power down the system
> Power up the system
> Observe the system date
>
>
>7. Date set - 29 Feb 2000
>
> Set the date to 29-Feb-2000
> Observe the system date
>
>
>8. Leap year - Rollover 2/28 - Power on
>
> Set the date to Monday, 28-Feb-2000
> Set the time to 23:59
> Observe the system date after midnight
>
>
>9. Leap year - Reboot 2/29
>
> Verify the date is 29-Feb-2000 from the previous test
> Power down the system
> Power up the system
> Observe the system date
>
>
>10. Leap year - Rollover 2/29 - Power on
>
> Verify the date is 29-Feb-2000 from the previous test
> Set the time to 23:59
> Observe the system date after 00:00 AM
>
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
>11. Set the date back to the correct time.
>
>
>12. Restore the entire system from tape?

Thanks again to everyone for responding to my post!!

Martin Meadows
Indianapolis, IN.



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:15 CDT