Re: Another heads up on a restore

On 18-Aug-2014 07:19 -0500, rob@xxxxxxxxx wrote:

I really thought about running RCLSTG.
Some people who seem to be opposed to RCLSTG being ran routinely
seem to think that after such a restore is probably not a bad idea.

Given the effect [the recovery of the condition with missing *AUTL public authority] can be achieved _without_ a Reclaim Storage (RCLSTG) request, by composing and running a [simple] user program to process every external object in QSYS [perhaps even limited to a likely subset by object types], seems to suggest that performing all of the work to process *every permanent object* with an entry in the permanent storage directory so soon after every object has just been created as part of a full system restore [with few opportunities for /problems/ arisen for which a reclaim might assist] would be daft given any value placed on system availability. And instead of performing exactly just the one required Grant Object Authority (GRTOBJAUT) request in the user program to effect full correction for each object, the reclaim processing would perform numerous other validations and\or operations against every object, all while the system remains unavailable in restricted-state; just one operation against the affected objects within a small set of objects found within a somewhat larger [but again, a small set] of objects for which just one validation was performed, all while the system remains available [no restricted state required] seems conspicuously to be the better choice.

So while I am in the group of /some people/ who are opposed to using RCLSTG on a scheduled\routine basis, I am *not* one who would claim that RCLSTG is "probably not a bad idea" for the noted scenario; excepting if the time spent performing RCLSTG and thus the lack of system availability are no-cost to the requester. I noted only that one specific [documented] phase of processing by the reclaim feature, is the _authority recovery_ phase, and presumably that would effect the same corrective as what the [simple] user program could. I have been clear in my past statements about how some of the functions provided by the RCLSTG can be effected without actually invoking the RCLSTG command, and thus running that reclaim function to correct some specific errors is very costly in contrast with just /fixing/ the specific issues with some other action; e.g. using the GRTOBJAUT to assign the authority of the user *PUBLIC to redirect to the Authorization List (*AUTL) object, for the specific issue described in the OP. So repeating what I said in my prior comments, "I am in no way recommending this (the RCLSTG)" as a means to correct the authority; the reason being, exactly as described above, about how using RCLSTG would be ridiculously excessive.