Friday, May 23, 2014

IBM thinks that MCH3601 and the creation of dumps is normal behavior during a restore operation

The readers of my blog know, that i am a big fan of the "IBM i" operating system, because it is the most stable and reliable system ever built. We have been using "IBM i" and its predecessor "OS/400" since 1993 and we had only one crash (V3R0M5) of the operating system since then. One other big advantage of the system is that backup and restore is an integrated part of the base operating system so that backup and restore is rock solid too. 

But two weeks ago my believe in IBM i got a big hit. We want to restore a library from our backup and made the mistake that one user was still in the application which uses this library. Normally this should result in a CPF3762 error and the file wich is in use should be skipped. But this time we got a unhandled MCH3601 in QSCJSDMP and many dumps and problem records. I do not think that a MCH3601 error and the problem records should be generated in such a situation so we canceled the restore operation and got a second problem that the cancel operation leaves an object in a damaged state in the library. I also think that a cancel of a restore should not leave objects in a damaged state. So i have opened a PMR 22505,060,618 and IBM responded after one and a half week of analyzing the problem with:

MCH3601 and the Creation of Problem records when an restore fails caused by a lock is working as designed. The problem that objects are damaged when a restore is canceled is also working as designed.

This answers are really very disappointing and i can not believe that IBM do not want to solve this problems.

1 comment:

  1. First they killed their Notes team and since this is finished, on to the AS/400.