Hi there,
I have a pro*c program that fetches data from table using host arrays.
Everything seemed to work fine, until I tried to run the program on
our production server. I followed 'top' command output and found that
my program size was increasing in 1Mb chunks for every minute.
Starting with 16mb, in fifteen minutes it had shot upto 90Mb.
Ofcourse, I had to kill the process.
I traced the memory leaks and found that they were coming from Oracle
libraries. Every fetch statement alloc's memory which is not freed up.
If I were to fetch millions of rows from a table, I could easily use
up all the available memory. I am going to raise a bug in metalink,
but just wanted to check if anyone else had seen this before. I am
using 8170 version.
Any information would be extremely helpful.
Thanks
