Re: Library Authority

Such a response ignores the true potential for stress that might be involved. The noted processing may be unacceptable to businesses for DR with large files, since doing so could add several tens of hours for access path recovery time. Without proper planning, applications may be unavailable for days while critical activity competes with rebuilds for CPU time, or while rebuilds are given dedicated time thus preventing activation of other processing that is critical to the business. While it may be workable for your situation, that does not make it generally applicable to others. Many will never even test DR to find out that they will be SOL when the time comes for a real DR, so implying it is acceptable without emphasizing the potential caveat will leave them disgusted with the OS and possibly with the /advice/ they were given.

The IBM i Operating System 6.1 offers some new techniques which enable spanning libraries without access path rebuilds, but only with explicit modifications to the DR processing.

Regards, Chuck

rob@xxxxxxxxx wrote:

Personally I don't find the separate library thing such a big hassle.
1 - Name the LF library higher in the alphabet than the pf library.
So if your PF library was MYLIBF then name your lf library MYLIBFLF.
Then in a complete unload/reload the PF's will all be there.
2 - If someone violates this, or if they do stuff like have a LF in
LIBA pointing to a PF in LIBB and a LF in LIBB pointing to a PF in
LIBA then I still wouldn't stress out. When restoring I simply do a second RSTLIB *NONSYS with the OPTION(*NEW). There's even some obscure reference to this in the Backup and Recovery Guide. Having done an unload/reload within the last month it was no problem.

I've got real issues to get stressed over instead of being concerned
with cross library logicals.

Hey, if my boss found that easy solution, and he hasn't written a
program in way over a decade...