Hi All, I found the answer to my memory leak.
My application includes a CF process (or two) that controls an interface - the page is designed to never stop running - just looping over a tcpip port, getting data and updating the database. This process works fine on some machines, and on others chewed memory, then sent the CPU to 100%. I tried changing db drivers, tuned the JVM, found out more about Java garbage collection than I really want to know, and then looked at rewriting the process. During the rewrite, someone asks "what happens to the debug output from this page". Then it dawned on me - we had debug turned on for the pre-production machines - the output was loaded into memory, and had nowhere to go. Turning debug off certainly helped, but there was still a much slower leak. Encompassing the pages in <cfsilent> then fixed everything. So I have no idea why it works OK on some machines (without the <cfsilent>), but the problem was not with CF. Thanks to all who made suggestions. Phil. ----- Original Message ----- From: "Phil Evans" <[EMAIL PROTECTED]> To: "CFAussie Mailing List" <[email protected]> Sent: Friday, January 14, 2005 8:27 AM Subject: [cfaussie] Re: CFMX Memory Leak > Thanks Paul. > > I have my application running on a number of servers, 2 of which exhibit > this memory leak / CPU spike, and 3 don't. Following on from your post, I am > doing some comparisons and testing based around CF version (which patches > have been applied), O/S version, and odbc vs jdbc. If I make any > breakthrough I'll post the results back here. > > In the meantime, any other thoughts are still appreciated! > > Phil. > > > ----- Original Message ----- > From: <[EMAIL PROTECTED]> > To: "CFAussie Mailing List" <[email protected]> > Sent: Thursday, January 13, 2005 5:31 PM > Subject: [cfaussie] Re: CFMX Memory Leak > > > > On Thu, Jan 13, 2005 at 05:01:51PM +1100, Phil Evans wrote: > > > > > A while back Mike Kear posted the following on a list somewhere, but no > resolution > > > is posted. > > > > > > We have the same problem, whereby JRUN chews memory, and hits it's > limit. The > > > CPU then hits 100%, and the only way out is to restart the CF > Application Service. > > > > > > Has anyone (particularly Mike) found a resolution to this issue? > > > > I missed the earlier posts. Has the issue of database driver been > > raised? And possibly dismissed? > > > > We were plagued by this one for ages, until we moved our DB to > > one that has a native Java JDBC driver. > > > > Our previous DB used a Java<->ODBC bridge, and it's there that I suspect > > we had our problems. Tinkering with the codebase did absolutely nothing > > to stop the intermittent problems. > > > > Simply moving to a JDBC environment with no other code changes has changed > > uptimes from hours/days to > it-only-goes-down-during-scheduled-maintainence. > > > > It could be a totaly unrelated problem, but your symptoms are identical > > to ours. > > > > Cheers > > > > Paul Haddon > > Technical Services Manager > > Formstar Print Technologies > > > > > > --- > > You are currently subscribed to cfaussie as: [EMAIL PROTECTED] > > To unsubscribe send a blank email to > [EMAIL PROTECTED] > > Aussie Macromedia Developers: http://lists.daemon.com.au/ > > > > > --- > You are currently subscribed to cfaussie as: [EMAIL PROTECTED] > To unsubscribe send a blank email to [EMAIL PROTECTED] > Aussie Macromedia Developers: http://lists.daemon.com.au/ > --- You are currently subscribed to cfaussie as: [email protected] To unsubscribe send a blank email to [EMAIL PROTECTED] Aussie Macromedia Developers: http://lists.daemon.com.au/
