Ryan Bloom wrote: > > > -----Original Message----- > > From: Justin Erenkrantz [mailto:[EMAIL PROTECTED] > > Sent: Wednesday, February 06, 2002 8:59 AM > > To: Sander Striker > > Cc: [email protected] > > Subject: Re: [PATCH] Free memory over a certain threshold back to the > > system > > > > On Wed, Feb 06, 2002 at 01:24:36PM +0100, Sander Striker wrote: > > > Hi, > > > > > > The 'high free' patch. I'm not sure we want this. > > > It may hide pools abuse problems. > > > > Yeah, I agree we don't really want this. If we run out of memory > > with the normal pools, it means that the lifetimes are most likely > > incorrect. > > > > But, perhaps this could be a #define with a debug option? -- justin > > We definitely don't want this IMHO. If the pools are filling up, then > either pools are the incorrect model for your app, or they aren't being > cleared often enough. I would much rather not have this as an option at > all, because that just encourages people to use it. :-)
I was under the impression this patch simply kept the free lists from growing and staying excessively large, not 'filling up pools'. There are some apps which can and should use pools but which, during the course of their lifetime, have periods of heavy usage and thus create an abnormally large number of pools; since these pools are never truly freed and hang around in the free_list forever, the process clings to this memory even if it's never needed again. For instance, a process which accepts requests into an internal queue and processes them in seperate threads at a later time. If the arrival rate (temporarily) exceeds the processing rate, the queues (and the associated pools) will grow excessively large. Without this patch, even when all the pools are destroyed, the process will still hold onto that memory. As someone who's got a process that is suffering from this behavior, I was looking forward to this patch... perhaps it could be added as a compile time option? Is the pool model fundamentally flawed for this type of program? I'd hate to have to strip all of APR out of my code; other than this, it's worked perfectly for me. Thanks, Ron -- Ronald K. Park, Jr. [EMAIL PROTECTED] CNET Networks / CNET.com / ZDNet / mySimon 908.541.3738 The source for computers and technology.
