Hello All,
We are observing a strange problem for a couple of days. One of our batch
process which usually takes about 30 minutes to complete sometimes takes upto
8 hours to complete and this varies even though the batch load is almost the
same everynight.
Some details on the job
The job is a simple Oracle replication process of about 20-25
tables.
Put a trace on the session during the time it spent a lot of time and
observed that the inserts into our table took a very long time. No waits were
observed. The tkprof report shows the CPU time = Elapsed Time.
One other thing observed was the RBS usage jumped up to around 8 Gb.The
table size is only 700M, add about 5 indexes to it, still does not explain why
RBS usage should be so high. On normal days i.e. when it gets processed in 30
mts., the RBS usage is only about 100M.
After the job is done, the database size increase is only about 100M, so
wonder what goes into the RBS.
No transactions in the database at the time the batch runs.
Behaviour is not consistent, some days are good , others bad. Did
complete network testing, no problems, server tests, disk transfer tests ,
everythings seems fine. The DB is an 8.17 database.
Any idea folks how to proceed.
Thanks
Satish
