Archives of the TeradataForum
Message Posted: Thu, 01 Aug 2002 @ 18:00:43 GMT
| Subj: | | Re: Rollback on long-running update |
| |
| From: | | Christopher Platt |
One thing we noticed during a rollback is that the transient journal will grow and decrease in size.... and can take up more space
then what you have allocated for the DBC database. Several years ago I put in product enhancement request to have NCR's CCL product
page someone when DBC was out of space. It was laughed at. Years ago we were taught if DBC ran out of space your system stopped.
Imagine my suprise when it did not stop and continued to consume hundreds of GB of space. We were told in this situation it would
have taken over a thousand hours to rollback.....We opted to abort the transaction which trashes that table and then you restore it.
Keep in mind if you change the rollback priority.......I think rollback locks this table.... you prolong the amount of time this
table is unavailable. If memory serves me correctly, and I am getting old so it fails periodically, checkpoint processing (which may
or may not be an option depending on your queryman deployment of this transaction) would help to alleviate some of this problem....
but if I am wrong about that I am sure someone will correct me....
Basically there is not alot you can do with rollback.... short of let roll baby roll.... all night long....!
| 
