Hello,
Kees, sorry for the delay in answering:
>You say: "Note that SMFLIMxx REGIONBELOW/REGIONABOVE are *not* related
to the IEFUSI's "region size" parameter"
>Am I correct to state that SMFLIMxx REGIONBELOW/REGIONABOVE are equal
to IEFUSI setting a Regionsize=Regionlimit both below and above?
sbj: As the function exists now, VSM sets the job step's limit and size
values using a myriad of fields passed into it:
- JCL Region= or REGIONX= values
- SMFLIMxx REGIONxxxxx values (which will override the above JCL REGION
values)
- IEFUSI RegionLimit and RegionSize values
- SMFLIMxx SYSRESVxxxxx values, which will essentially be used in place
of the IEFUSI REGIONLIMIT value.
VSM will factor these values together and come up with a regionlimit and
a regionsize.
Since there is no regionsize keyword, when SYSRESVxxxx is used, the
regionsize should match the regionlimit.
If the JCL REGION or the SMFLIMxx REGIONxxxxx values are non-zero and
lower than the determined regionlimit, that means the user (or SMFLIMxx
coder) has a specific amount in mind and the regionsize will be set to
that amount. And as we know, if you request more region via JCL than is
available, an abend will result. IIRC, if SYSRESVxxxxx is not used and
IEFUSI doesn't set any values, VSM will add a default value to the
regionsize to set the regionlimit.
This is why people are simply coding REGIONBELOW(NOLIMIT)
REGIONABOVE(NOLIMIT) to simply override whatever the user requested and
giving them REGION=0M. I'm not sure why they aren't using SYSRESVxxxxx
to protect private for system use, unless they still have their IEFUSI
in place.
>IEFUSI doc explains the value of difference between regionsize and
regionlimit with regard to VL Getmains. Is this no important anymore,
because SMFLIMxx does not provide control of a different regionlimit
anymore?
SBJ: I won't say it's not important, just that it was not in the
function provided on day 1. And that the nuance about VL Getmains is
often missed.
>Region=0M implies Memlimit=0M. Does the SMFLIMxx Memlimit parameter
overrule Memlimit=0M for Region=0M tasks?
sbj: REGION=0M implies Memlimit=0M, but only when coded on the JCL.
Using REGIONxxxxx(NOLIMIT) does not specifically alter the MEMLIMIT. The
SMFLIMxx memlimit parameter will overrule the MEMLIMIT coded and the
implied memlimit from region=0M. Since IEFUSI and SMFLIMxx have the
ability to set any MEMLIMIT, there is no reason to imply a memlimit.
>If so, a filter on REGION= would be helpful or I need IEFUSI to ignore
SMFLIMxx for Region=0M steps.
sbj: Please open an RFE in DeveloperWorks. I think you'll get a lot of
votes for the idea.
>Does NOHONORIEFUSIREGION in SCHEDxx also apply to SMFLIMxx, i.o.w. is
SMFLIMxx also ignored like IEFUSI is? If so, will SCHEDxx docs be updated?
sbj: NOHONORIEFUSIREGION does override SMFLIMxx. I'll make a note to
get the doc updated.
Thanks,
Kees.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
