On 22-Jun-2017 07:32 -0600, Rich Loeber wrote:
While moving the security environment to a new Power 8 server
running 7.2 (from 6.1), the certificates do not show up on the new
box. In the joblog, it indicates that system file QAYDOBACK was not
restored because of a differing file level.
  With regard only to what was [presumably, per unstated,] the error 
msg CPF3283 "Saved file or member level ID not same as file &1.":
   • Restoring the so-called "user data" for a quasi-system database 
file [member], from a previous release is not a valid/supported means to 
"migrate" the data; the proper method is to scratch-restore the 
down-level file, and then *install* the feature [e.g. install QUSRSYS].
   • Given the [member] data can be transported as-is to effect proper 
operation, then provide a special value specification that is inclusive 
of the *FILELVL capability on the Allow Object Differences (ALWOBJDIF) 
parameter of the Restore Object (RSTOBJ) request,
I have manually checked the file level for both the source and
target files and they appear to match.
  Per Display File Description (DSPFD) output?  Was that by review of 
the Record Format Level Identifier, or the File Level Identifier?  A 
mismatch in the former would fail the restore irrespective the parameter 
specifications on the restore request, whereas a mismatch in the latter 
can be overridden with the specification of [, inclusive of; i.e. *ALL 
encompasses] the *FILELVL capability on the Allow Object Differences 
(ALWOBJDIF) parameter of the Restore Object (RSTOBJ) or Restore Library 
(RSTLIB) command.
I could just manually copy the three members in this file over to
the new box, but I don't want to shoot myself in the foot.
  I am unaware if there was any conversion program(s) between the 
levels of IBM i 6.1 and IBM i 7.2 [could have included a conversion to 
IBM i 7.1] for which the data in a member of that file had required 
modification(s) and thus for which a direct "copy" of the data could end 
up being invalid. Note that the symptoms for an improper migration could 
be as conspicuous as an error when using the feature that depends on the 
data, or the symptoms could be subtle, with effects such as confusing 
[and difficult to diagnose] errors or incorrect output that could go 
unnoticed initially or indefinitely.
Has anyone run into this before?
  Had the v6r1 system been properly migrated to the v7r2 system, having 
followed the documented migration path, then such features [with user 
data in a quasi-user system library] should have been operational as a 
side effect of the install process; i.e. as an effect of a supported 
upgrade, from either a release N-2 or release N-1 level of the OS.
  Rather than taking a chance that the file data can be copied or 
restored, as-is, re-importing the data [via the interface provided for 
the feature] could be the most opportune.
As an Amazon Associate we earn from qualifying purchases.