Hi all,

I've read through similar postings to this list before, but none of them
describe what I am seeing, and its not about that `apache restart` is leaking

Here's the case:
FreeBSD `myhost` 5.4-STABLE.
Apache 1.3.37, perl 5.8.8, mod_perl 1.29 - all built via the ports
(by default mod_perl builds dynamic into the server)

I start the server in single mode and make lets say 30 requests. The process
size at this point seems normal, a bit more than the original startup size,
but this is since I do not preload anything inside the httpd.conf file.

After these 30 (sometimes 40) requests, the process starts to grow by an avg.
1.4K per request , which I think is rather huge, since there are no other
modules than the default loaded by the Apache::* / mod_perl stuff .

I've tried to re-compile Perl (and mod_perl) outside the ports tree - with and
w/o the settings: -Dusemymalloc , and -Dbinmode5005, but none of the combos
seem to fix the problem.

The same thing described above applies to FreeBSD 6.0-STABLE, perl 5.8.7 -
same apache/mod_perl versions, same build procedures.

I guess it could be an external library leaking, but has anybody experienced
this on recent configurations like mine ?

This is the dummy handler package
package My::X;
sub handler {
my $r = shift;
$r->content_type( 'text/html' );
$r->print( $$ );

====== httpd.conf =======
PerlModule My:ummy

SetHandler perl-script
PerlHandler My:ummy

Here is the most recent build I've tried (again, via the bsd ports):
Embedded Perl version v5.8.8 for Apache/1.3.37 (Unix) mod_perl/1.29 process
osname=freebsd, osvers=5.4-stable, archname=i386-freebsd-64int
[ .. ]
-Duseshrplib -Doptimize=-O -Ui_gdbm -Dusethreads=n -Dusemymalloc=y
[ .. ]
usethreads=undef use5005threads=undef useithreads=undef usemultiplicity=undef
[ .. ]
usemymalloc=y, bincompat5005=undef



Assen Tchorbadjiev
8D6D 1FCF 96DB F37E E421 1BD2 D04E A856 20A4 AA20