DSXchange

I have a parallel ETL job which has a ODBC stage for extrating data from Db2 database and loads the data in to a Sequential file stage .The volume of the data is 4 to 5 millions and i am getting the following fatal error every time i run that job .

APT_CombinedOperatorController,0: Operator terminated abnormally: Terminating with exception:APT_BadAlloc: Heap allocation failed.

I have a similar job but extracts different database in DB2 but has more volume( 16 million ) and Column lengths than the above job but it never failed . I have the same settings in the ODBC stage that i used in above job.

Any Suggestions

We never even bother to try to diagnose errors thrown by APT_CombinedOperatorController. Disable operator combination so you can learn which actual stage (operator) is throwing the error.

I tried to disable the operator combination ( combinality mode to Dont combine) and ran the job but got the following Fatal error in the ODBC stage.

DB2S_PWRSTCH_DATA,0: Operator terminated abnormally: Terminating with exception:APT_BadAlloc: Heap allocation failed.

DB2S_PWRSTCH_DATA is a ODBC stage


I kept the following parameters in the ODBC stage

Transaction Isolation = Uncommited read
Buffering_mode= Default

It's the same error you previously reported, now you know for sure that it's the ODBC stage which is generating the error. You now have to search for the exact error message in DSXchange forums, and work out what you are doing in your ODBC stage that might consume more heap memory than is available.

Few months back we also has the same problem and we change then no. of nodes for that job from 2 to 4 to solve the problem.

It has been working fine so far.