On Sun, Sep 09, 2007 at 10:20:37AM -0700, Randal L. Schwartz wrote:
> >>>>> "Peter" == Peter Hartzler <[EMAIL PROTECTED]> writes:
> 
> Peter> Our site currently has around 200 templates.  What I *think* I'm seeing
> Peter> is memory consumption where each apache instance (child w/ mod_perl)
> Peter> keeps its own cache.  The next step from there is the idea of
> Peter> centralized caching.
> 
RS> Have you tried preloading before forking?  As long as the
RS> templates aren't changing, I've found that gives some relief.

Interesting.  The only detailed docs I can find that discuss this are
for Apache 1.x.  It describes the server-context shared memory as
copy-on-write [1], So I'm not too confident that loading before forking
would really yield much benefit...  Unless we were to slurp in all the
templates at startup, which is an interesting possibility.

I suppose we could explicitly use the Apache server-context memory pool,
but in either case, the real question is still what to cache.

Pete.

  ------
1. http://perl.apache.org/docs/1.0/guide/performance.html#toc_Sharing_Memory

_______________________________________________
templates mailing list
[email protected]
http://lists.template-toolkit.org/mailman/listinfo/templates

Reply via email to