Re: Physical Files & Logical Files

rob@xxxxxxxxx wrote:

The QAOK* fiasco does happen once in awhile. I used to notice
it on OS upgrades. In fact I have, in my standard OS upgrade
instructions to:

WRKF QUSRSYS/QAOKL*
Search for any that
1) Do not end with an A
2) Instead end with an increment, such as 001
3) And have a description that begins with Old name
If you see any that meet all three of these conditions then
delete these files.

WRKF QUSRSYS/QAOKP*
Search for any that
4) Do not end with an A
5) Instead end with an increment, such as 001
6) And have a description that begins with Old name
If you see any that meet all three of these conditions then
delete these files.

If renamed QUSRSYS/QAO* files [or any other QA* files in a system library] "does happen" as part of an install\upgrade then there is a defect in the install or the install was impacted by remnants of a past usage error. Such files having been renamed should only be a side effect of the usage error whereby an "all object differences" restore was done into the QUSRSYS [or similar issue for other "system files" in the other pseudo-system libraries; e.g. QGPL & QSYS2].

At a minimum, for the fact that the problem may instead or also exist at the member level [i.e. members renamed versus or in addition to database files renamed], the above instructions are incomplete; IMO very suspect, such that the original data would be lost. The problem should be avoided by *never* issuing a RSTxxx ALWOBJDIF(*ALL) into a pseudo-user\pseudo-system library. If some differences must be allowed for the restore, specify each individual special value explicitly\separately.

I would recommend that nobody follow the quote instructions as written. The process of recovering from the past usage issue is much more complicated if the feature(s) which use the database files are expected to function properly and data recovery is desirable to avoid having to re-enter the user data.

Regards, Chuck