Thanks for the reply. To put it in perspective, the same code/data works
fine on a machine with only 512M of RAM when running CF 4.5.
I have tried setting the MaxPermSize and the overall Heap as high as
they can go (512/1024MB respectively) to no avail. When I try setting
these any higher CF does not even start (although I have only 2G RAM on
the machine). Does MX really need that much more overhead the 4.5 and 5?
The answer is "it depends". Compiled Java applications can require substantially more memory than something like an interpreter written in C++ (which pre-MX versions were). It sounds like your particularly problems center around the database drivers and the memory usage associated with large record sets - I expect the memory footprint of such things is much larger in the Java implementation than in the original C++ implementation. Java can be pretty memory hungry (and a lot of people in the early days after the initial release of CFMX 6.0 complained about the increased memory usage).
You shouldn't need the MaxPermSize that high but the bigger you can make your overall heap the better. It's just not always easy to figure out exactly how to tune the JVM for a specific application. Try removing the MaxPermSize setting (so it uses the 64Mb default - since it mostly affects the number of classes that can be loaded, I believe you'll be OK) but pushing your overall heap as high as it will go.
If that doesn't work, you may have to consider either adding more RAM or changing the app so it can use smaller record sets. How many records are you trying to process at once? What are you doing with so many records? Could you do more of that processing inside the database?
Sean A Corfield -- http://www.corfield.org/blog/
Got Mach II? -- http://www.mach-ii.com/
--- You are currently subscribed to cfaussie as: [EMAIL PROTECTED] To unsubscribe send a blank email to [EMAIL PROTECTED]
MXDU2004 + Macromedia DevCon AsiaPac + Sydney, Australia http://www.mxdu.com/ + 24-25 February, 2004
