Subject: Re: CICS error
From: Tony Thigpen (tony@vse2pdf.com)
Date: Mon Apr 07 2008 - 17:19:34 EDT
I think you were seeing a 'storage compression'. It was very common in CICS/VS. If CICS ever gets low on storage, it would throw out all programs loaded in 24 bit area and not reload them until they were again referenced. Tony Thigpen -----Original Message ----- From: Frank Swarbrick Sent: 04/07/2008 02:41 PM >>>> On 4/4/2008 at 1:47 PM, in message > <003701c8968c$df396e90$9dac4bb0$@com>, > Dennis McLoud<dmcloud@systronicsinc.com> wrote: >> Are you running CICS/TS? Are the programs all running below the line? We >> have not seen this problem since going to CICS/TS and moving most of our >> programs above the line. You should only see this error after program >> space >> compression. CICS doesn't need the pointer to the library unless the >> program >> is flushed from memory. You could at least make the problem less >> frequent by >> giving CICS more DSA. > > I think there is one other case when this can happen. It appears, though I > can't prove it, that if a program is not used for quite a long time it also > gets 'removed'. > > We have a CICS that is only used during business hours. For a while we were > not cycling it at night, we'd just leave it up. We generally implement CICS > programs while the primary production CICS region is down; thus no PHASEIN. > But the secondary region (the 9-5 region) would still get this error the > first time an attempt was made to use this program in the morning. My > assumption is that at some time during non business hours, which lasts over > 12 hours, all of the 'inactive' programs are 'flushed', so that an attempt > is made to reload it; bit it fails since the pointer is no longer pointing > to the correct area (since the phase has moved elsewhere). > > If anyone has more information on this I'd be interested to know about it. > > What I don't understand is why CICS/TS cannot be modified so that, if it > hits the situation where it would currently issue the DFHLD0203 error, it > would instead do an implicit phasein. If I submitted this requirement to > WAVV would you all vote yes on it? > > Frank > >
This archive was generated by hypermail 2b25 : Fri May 16 2008 - 23:50:07 EDT