STL Time Machine Report #5 - 1998-01-28 We completed one cycle of testing, using IPL dates from 05 JUL 1999 thru 01 JAN 2001. Not all test cases were successfully tested, due to continuing problems with test data and setups. We did learn some things. We refined our test plan. Added some additional reports for analysis, dropped one test case that turned out to be untestable (and low-risk, contingencies are available). I have been having daily meetings with the business analyst to review test results. It might actually take us longer to review the test results than to run the tests. This ties up the subject matter experts, which includes me. Since my application is a real-time system, there are some particularly difficult parts of the test plan. Certain transactions must occur within a very narrow time window just before rollover. This tends to put the testers under a little stress, since failing to complete that test means it cannot be retested for two weeks. We have two more test cycles to run, and then we're kicked out of the time machine in mid February for the next group of raw recruits. The COBOL II compiler and runtime library (Release 4.0) work fine through 01 JAN 2001. If we need it, it could be important. The next group has to deal with the database that won't support Language Environment for MVS. The vendor and IBM have been cooperating on the upgrade. We get the beta version on 16 FEB 1998. We thought we found a nifty JES2 problem when 01 MAR 2000 resulted in a JES separator header date of 96/99/00 instead of the expected 03/01/00, but it turned out to be a known problem. We're stilling running JES2 Release 4.3 from our old MVS instead of JES2 release 5.2. Tech Services was already aware of it, and they plan to install the upgrade "real soon now". SAR (Sysout Archive and Retrieval) still has some problems locating reports by date when you move back and forth in time (you can still search by jobname and generation). The powers-that-be have decided to wipe the SAR database when IPL'ing backwards in time, so that should improve. We've had a lot of problems printing test results with a single Laserjet 5si. Loooong waits (8 hours or more) to get reports out. Another application area is testing the same time as we are and there just isn't enough print capacity. They're adding two more 5si's next week. A buddy of mine at a different company is working on guerrilla remediation, since his PHM forbids any mainframe development. They have VM/ESA and VSE (compliant versions, according to IBM), so they set up a VSE guest and started booting it up. No test plan, just fire it up and see what breaks. The first thing that happened was it wouldn't IPL with a forward date. Oops, configuration error, add ENABLETOD=YES. Fix that, try again. The next thing that happened was a bunch of licenses and passwords expired. Call the vendor, get a workaround. This same friend of mine was at a Superbowl party last Sunday and met a whole bunch of geeks I haven't seen in a long time. Several of them are now making BIG bucks working on SAP and PeopleSoft for places like Anheuser-Busch and Boeing (formerly McDonnell-Douglas). Guess what? All their SAP and PeopleSoft work is on legacy interfaces! They're not replacing their legacy systems, they're attaching additional systems to them. I thought Cory was kidding with his story about putting a TN3270 emulator on a PC and saying, "See we're off the mainframe now!" Here's a time machine tip, maybe it's worthwhile, maybe it isn't. The plan we chose called for fixing the applications for Y2K, running them thru Y2K unit testing, then running normal regression test (with current dates) and installing into production. We know our Y2K fixes work with current dates. Only production code goes into the Time Machine. Only date problems are tested in the time machine. Y2K date fixes, three so far, are tested in the time machine and rolled back into production as soon as possible. I don't know if that's the best way, but it seems to be working so far. -- Arnold Trembley http://home.att.net/~arnold.trembley/ "Y2K? Because Centuries Happen!"