I had an odd issue during some testing I was doing on a custom source component I’m writing for EBCDIC data conversion. I first assumed that surely my code was at fault but the test just seemed to simple to be “my code”. So here’s the scenario, I have a file with 100,000 bytes of the number ‘1′ in EBCDIC. I created a simple package to import the file as 1 byte rows into a Dataset Destination. I was surprised when I executed the package and found that only 65,536 rows were processed. Now if you have been around software for more than a week you probably recognize this magic number and unfortunately it triggers all sorts of theories about what could possibly be going wrong. Well, I’m still trying to investigate the “why” and I’ll update this post and welcome comments if anyone has input but for now the solution to this oddity is to increase the DefaultBufferMaxRows to greater than your number of records. This seems like a horrible workaround for a situation that should happen quite often so the hunt continues…
July 10, 2007
2 Comments »
RSS feed for comments on this post. TrackBack URI
Same problem here,
created query with sqlplus that’s generates 344743 records. When executing the same query within SSIS only 60000 records are imported. I’ve still not found the problem. I’m quering on a oracle 9 database with a oracle 9i client(+ latest patch). Also I’ve installed the latest SP for visual studio en SQL 2005.
Currently i’m trying out the max buffer sizes that causes some fluctuation in the amount of imported records.
Help is needed.
Thankz,
Ron Killaars
IT Engineer
Comment by Killaars — August 13, 2007 @ 3:07 am
did you make sure you were calling SetEndOfRowSet on all buffers?
Comment by benjamin harrell — August 23, 2007 @ 11:39 am