httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Craig Tussey <ct8...@yahoo.com>
Subject [users@httpd] possible interaction?
Date Sun, 02 Dec 2012 16:27:38 GMT
This is a shot in the dark question.   Something unexpected happened with 1 part of the application
presented from my server.   I cannot explain it and have not been able to resolve it.  
It happen after I went through an exercise in website performance improvement

I'm an application person working in a Apache/2.0.64 (Win32) mod_perl/2.0.3 Perl/v5.8.3 environment
currently on a Window 7 platform using IE 8.  I'm up under mod_perl with no errors.  
Recntly I tried to improve the loading time of the website that presents the Perl/CGI/C application
I've developed.   I added compression, external caching with expiration dates, some meta
statements related to compression, and reduced the footprints of static images presented in
the main page.

The compression and caching statements I put into the httpd file are below.   The GTMetrix
site confirmed that both were working.  My site loading score was 98%.   I ran some application
tests to make sure that dynamic png and jpg images generated by the application still worked
and they did.  So far so good.

<Location />    
   SetOutputFilter DEFLATE
   AddOutputFilterByType DEFLATE text/html text/plain text/xml 
</Location>

Header unset ETag
FileETag None

<Directory "/usr/www/steepusa">
    Options FollowSymLinks MultiViews
    AllowOverride All
    Order allow,deny
    Allow from all
    <FilesMatch "\.(gif|jpe?g|png|ico)$">
        Expires A2592000

        Header unset Last-Modified
        Header set Expires A259200  
    </FilesMatch>
</Directory>

I added the following 2 meta statements related to compression in and out.

<meta HTTP-EQUIV="Accept-Encoding" content="gzip, deflate"/>
<meta HTTP-EQUIV="Content-Encoding" content="gzip, deflate"/>  

I also uncommented the lines below supplied with the original httpd file for the header and
expires date processing I think.

    LoadModule expires_module modules/mod_expires.so
    LoadModule headers_module modules/mod_headers.so

I want to say I don't throughly understand what I did above.  All of the above comes from
reading documentation and posts of others doing similiar things found in searches.

If I execute my webstite (main page) from the above directory and that website has static
images and provides 1) links to secondary pages kept in a subdirectory and 2) links to my
CGI application (in a subdirectory) which manages static images and generates and manages
dynamic images, will images in the main website page, any of the seccondary pages, and the
CGI application be cached out in the WWW?   Also, I'm thinking that only html/text pages
are compressed off this server?   Application modules (CGI) are loaded out of the mod_perl
repository already interpreted and should not be compressed.

Anyway, I went away for a couple of days for other work.   Of course, other, seemingly
unrelated things are happening on my web connected box, like OS updates, while I'm away. 
During that time I decided to back caching out because of the nature of my application and
commented out of the httpd everthing below the </Location> statement above.    I
reran the application tests to make sure that the dynamic png and jpg images generated by the
application still worked.  This time a Perl/CGI simulation of dynamically generated color
maps (using GD stuff) presenting the last five years of color maps one after another, which
had been working and stable for 6 months, did not present the map sequence.  I verified that
the process actually runs when executed and generates the color maps, but does not present
them. 

Its easy enough to take site compression out just by not specifying the file type when invoking
the html website.   Finally, my question is if any of the above httpd changes might have
impacted the simulation presentation of the color map sequence?    For example, might images
cached when I first put the caching stuff in the httpd and ran the application tests be cached
out in the web and impact this simulation in that way.   I think this website loading stuff
is the only work I've done since the last successful simulation tests.   That is why I'm
asking this shot in the dark question.  
Mime
View raw message