Collapse huge request leads to log is full (DB6)

Question

Dear Experts,
In one process chain, we have to collapse in one cube the initial request (10 millions records).
Each time we try, after a long time, we have the message
SQL0964C The transaction log for the database is full. SQLSTATE=57011 row=1
we use DB6 and have already increased the number and size of logs to
LOGPRIMARY = 140
LOGFILSIZ = 20476 4KB
Unfortunately this is not enough.
To reduce the log space, we have tried to collapse this request with removing indexes before, but at the end we always have the same issue.
How do you collapse your huge requests? do you only adapt size of DB6 logs or do you use an other way to do it?
Thank you for your help.
Best Regards;
Thibaud.

Answer 1

Hello,
you can save some space in the transaction log by dropping the indexes on the E-fact table before compression and rebuild them afterwards.  Compressioin is also faster then for large requests.  You have to do this manually in RSA1 as the DROP INDEX process for process chains only drops the F-table indexes. Don't delete all indexes on DB level, as some of them are required for compression.
Furthermore you can check note 597053. It is about request deletion, but it also works for the deletion step of the compression.
And if othing helps you could use infinite log space. I think the DB  parameter is LOGSECOND = -1.
Best regards,
Ralf