STL Time Machine Report #1 - 1997-12-30

At the STL Y2K Holiday Happy Hour I learned that at least one other shop
in St. Louis is working on Time Machine testing right now.  They run IBM
OS/390 and they're having problems integrating all the releases of
system software.  That seems to be a bigger problem that fixing the
applications themselves.  We run MVS/ESA 5.2.2 and are having similar
problems.

Monday I returned to work after 6 days of vacation & holidays.  In my
absence my project leader had volunteered me for the daily death march
meeting while he's on vacation.  It's only supposed to be 15 minutes, 
but dragged on for 35 minutes today because there hadn't been a 
meeting since 12/24.  We're still shaking the setup problems out.

Before I went on vacation I managed to run a test of COBOL II in the
time machine.  With an IPL date of 01/01/2000, my small test program
compiled fine, and was executed separately with COBOL II and LE/MVS
runtime libraries.  Date functions worked as expected with no problems. 
I want to repeat this test with IPL dates of 02/29/2000 and 03/01/2000,
but I don't see any obvious problems running the IBM COBOL II compiler
or runtime after the year 2000.

We already know CICS 3.3 won't work after 2000.  Our third-party
database doesn't work with LE/MVS (waiting on a new release from the
vendor), so we have to test CICS 4.1 with in two different
configurations:  LE/MVS runtime libraries and COBOL II runtime
libraries.  Although our Tech Services wanted all CICS 4.1 regions to
use LE/MVS, it looks like anything that uses this third-party database
will have to use COBOL II runtime libraries in the Time Machine until 
we get a new release from the vendor.  

It took us about a week to convert our CICS 3.3 regions to 4.1, and we
needed MVS and ACF2 patches to get security working.  Our business rules
do not permit IPL's or LLA refreshes from mid-November to Early January,
so the patches are only available in the time machine (since it can be
IPL'ed separately from production).  We've gotten past most of the
security and storage tuning issues of CICS 4.1, but we don't have 4.1 
in production yet.

Right now the biggest delays are being caused by staff shortages due to
end of year vacation time, and problems building the job schedules.  The
job schedules are complicated by a conversion to a different vendor's job
schedule.

I spent most of the day chasing down a problem with the Emulator which 
is not able to send messages into the Time Machine test network.  It 
turned out to be a CICS setup problem (missing terminal definition) 
rather than a system software or Y2K problem.

We have problems moving VSAM files into the Time Machine.  The users
complain it takes too long, so they are looking at some alternatives. 
We've been told to expect the Time machine to be down for up to 24 hours
every time we need to IPL backwards in time, due to the need to rebuild
the whole system.  IPL'ing forwards is not a problem. 

I also had a call from a guy who got beeped on Christmas eve for a date
problem.  Our old non-compliant date routine blew up because its holiday
table expires in 1997.  He's adding 1998 holidays to it, but this system
is scheduled to be replaced next year.  Six months ago we published the 
fact that this routine would not be made Y2K compliant and that all users should replace it with one of the available compliant routines.

All this fun and I get to attend tomorrow's death march meeting.

--
Arnold Trembley     
http://home.att.net/~arnold.trembley/
"Y2K?  Because Centuries Happen!"