Disk daily housekeeping jobs randomly abend S878 so how to estimate the REGION size in order to possibly avoid this abend.
It could be possible that the S878 is because of extended private (above the 16MB line) exhaustion.
By specifying REGION=7680K for example, the job will limit the the below-the-line storage to 7680K and it will get the default value for above-the-line allocation of 32MB, even if the remaining storage above-the-line is over 1 GB.
If this value is changed to REGION=700M, or REGIONX=(7680K,700M), the below-the-line storage will be still limited to 7680K, but the above-the-line storage will be higher than the default of 32MB.
In order to estimate the correct REGION value, the parm could be set to REGIONX=(7680k,zzz) and using incrementally-increasing zzz values, starting with 33M, it is possible to try the daily job and find the correct zzz value to avoid the abend S878. Then it could be increased by 20%-30% as a buffer amount for long-term usage.
Anyway, since this is virtual storage in a batch job, specifying 700M shouldn't cause any adverse affects and Disk Labs encourage to over-allocate this space and not try to be too precise with it so that the tuning doesn't require much maintenance going forward. Since the resource whose space being occupied is private virtual storage (as opposed to something like DASD space), an over-allocation shouldn't cause any adverse affects.