We had a Db2 load job run that used a load control dataset with 1,764 “INTO TABLE” statements run against a database that had one tablespace with 1,366 tables having 1,306 indexes. The DBD length = 2628728.
When the job was run, the SYSDBDXA tablespace went from a couple of datasets to 79 datasets of 5,824 cylinders in size.
We opened IBM CASE #: TS002854455, but from the details it appears that Db2 is working as designed.
We were lucky that we did not see this in production and run out of datasets.
This could have been very painful. Our storage folks were adding volumes as fast as they could, but we could have run out of spare volumes to add and lead to an outage. We had to cancel the job and rerun it in smaller pieces.
Why is it useful?
|Who would benefit from this IDEA?||Anyone using multiple load statements in the load utility|
How should it work?
Since we cannot control how a user sets up the load statement, change the design to allow reuse of empty space in the tablespace or some other method to not fill up the tablespace.
|Priority Justification||This could result in an outage for users.|
|Customer Name||Jeff Jerry|