On 27 Sep 2013 04:23, Vimalraj Nallasamy wrote:
>
> we have two systems to be upgraded
> 1.V5R1 to V7R1
> 2.V5R3 to V7R1
>
> what is the correct way to do that ?
>
> Is direct upgrade possible? or
> V5R1/V5R3 -> V6R1 and then -> V7R1
>
> Please suggest the possible way
Officially the supported paths are [as best I can recall,] with the
minimal number of transitions [per N+2 maximal support for upgrades,
given successive releases v5r1:v5r2:v5r3:v5r4:v6r1:v7r1]:
1a> v5r1 -> v5r3 -> v6r1 -> v7r1
1b> v5r1 -> v5r2 -> v5r4 -> v7r1
2a> v5r3 -> v6r1 -> v7r1
2b> v5r3 -> v5r4 -> v7r1
The sticky point is getting to v6r1, because program creation
templates need to exist to enable required /retranslation/ of executable
/program/ objects on\since IBM i 6.1. There were PTFs on v5r3 and v5r4
to enable ANZOBJCVN to prepare for the transition; i.e. identification
of objects that would be unable to /convert/ [translate] onto the new
system [because the templates were removed].
Personally I probably would find such a migration easier effected by
installing the new system at v7r1 and then creating objects and
configuring that new system [from quasi-system\quasi-user data being
reviewed from, but specifically not _migrated_ from, the old system;
e.g. mostly data in QUSRSYS, QGPL, and QSYS2 or similar libraries for
3rd party products], and then moving [restoring] just the purely
user-data to that new system. For the OS and most LPPs that would
probably not be overly difficult; the bigger problem is knowing what is
required about any 3rd party applications to effect the same /jump/
across multiple releases.
--
Regards, Chuck