Integration

Hi

I have an ESB process that loads a flat file (csv) and inserts the records into a database table. All very easy, except that the file can be very large.

I have 1 question and 1 issue:

1. Question: I use batching of records on the inbound file adapter at 100 records. What does this actually do? I expected to see multiple ESB processes kicked off to manage 100 records at a time, or have I missed the point? Anyway, suffice it to say that if I have 250000 records, one ESB process attempts to manage them all and the jvm blows it's heap very quickly.

Any way to manage this better?

2. I have done some testing on smaller files (15000 records). The ESB process does it's job and the records are inserted into the database successfully. However, when I look at the ESB Manager, the process states that is has rolled back. This is odd, as actually it hasn't and the data is in the database.

In the server log file i do see a WARNING as follows, just before the rollback occurs:

JCA: esb:///ESB_Projects/blahblah.wsdl [ blah: insert(table_name)]] - JNDI lookup of 'jdbc/cb1' failed due to oracle.oc4j.sql.ManagedDataSource

This could be why the ESB is rolling back, but why, as it's actually done it's job?

Any ideas or thoughts anyone?

Cheers

Chris