Now, the mainframe is known for being very robust and reliable. But the human beings operating them are not always.
This is the first installment of Mainframe Mishaps, a new feature on Search390.com. If you have a funny, illuminating or educational story about mainframe mishaps, please send it here.
Steven A. Kimball of Aetna submitted this story:
Years ago as a CICS system programmer, I was testing (just beta-released) terminal autoinstall under CICS 1.7, on one of my play test LPAR CICS regions. About that time, I then got a beeper alert that one of our larger volume production regions was having a problem. I dropped what I was doing in test, toggled up another host VTAM session using RUMBA and logged onto the production CICS region to research problem.
A while later, after the prod problem was resolved, I returned to my test work and decided to shut down my play region to pick up the flavor of the autoinstall program. I issued CEMT PERFORM SHUT and found out several minutes later (you guessed it) that I had just accidentally shut down the production (2 million transactions per day) CICS at 10 o'clock in the morning. My supervisor, soon thereafter, assigned me the task of writing XZCATT/XZCIN global User exits that would interrogate tranid for 'CICS' and if it was, look for 'perform shut'. If so, change tranid to 'SHUT' instead. SHUT tranid would throw up a BMS map that would ask user for confirmation that region shutdown was, in fact, intended to be shut down.
Pretty challenging stuff for a novice ASM programmer new to CICS; particularly because as we all know, the edits of the TIOA area needed to test for all combos of:
* P SHUT
* Perf Shut
* perf shu
* Perfor Shut
* P Shut
* perform shu
* You get the idea!
My exits stayed in place until CICS 4.1 was traded in for a new version ... after that, we did away with all user mods, included my well earned exits.