Re: Why i??? NEED YOUR INPUT

[CRPence <CRPbottle@xxxxxxxxx>]

  Apparently the possible defect questioned-of in the quoted 
message was addressed by the HIPer APAR MA39127 per:
http://archive.midrange.com/midrange-l/201003/msg01026.html

  So as I had alluded, that the OS informed of possible damage and 
separately that an IPL was required to recover from the error, is 
IMO probably preferable to the system having allowed the unexpected 
failure to compound a problematic encounter by enabling a new SWA 
request against the object that was already "logically damaged". 
That is, logically, the object which is already active to an SWA can 
not be saved by a different [no longer active] SWA.

Regards, Chuck

CRPence wrote:
> rob@xxxxxxxxx wrote:
> > <<SNIP>>
> >
> > Oh, if your save abends during a save while active you can
> > forget saving that library active until after the next IPL.
> > "changes" are recorded off to the side and it gets really
> > confused when the save abends like that. Claims the objects
> > being processed during the save 'may' be damaged but they're
> > really not. The IPL clears them up.
>
> Presumably "abends" means an "abnormal end" from the perspective
> of the OS [code] itself <<SNIP>>
>
> <<SNIP>> if the message about "object may be damaged" due to an 
> internal OS sfw error having led to the SWA failure was issued,
> then the message is probably accurate, such that the IPL activity
> does not necessarily always just clear them up; i.e. YMMV.?  The
> system storage management will have started tracking changed
> pages to the objects [IIRC, by redirecting the save activity to a
> backup copy of the pre-changed segment], so if the abnormal end
> itself or something between then and the IPL might also mess up
> that tracking [maybe only loss of main memory\storage], then the
> object or its data may not survive well.?  I would have to think
> too hard to remember how things worked to understand the if\how a
> scenario might end with damaged object(s).