httpd-bugs mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From bugzi...@apache.org
Subject DO NOT REPLY [Bug 32101] New: - Failed to pass the generated dynamic content completely
Date Sun, 07 Nov 2004 10:05:52 GMT
DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG 
RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT
<http://issues.apache.org/bugzilla/show_bug.cgi?id=32101>.
ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND 
INSERTED IN THE BUG DATABASE.

http://issues.apache.org/bugzilla/show_bug.cgi?id=32101

Failed to pass the generated dynamic content completely

           Summary: Failed to pass the generated dynamic content completely
           Product: Apache httpd-2.0
           Version: 2.0.52
          Platform: PC
               URL: http://www.whizzideas.com/whizznotes/xhtml/2441.html
        OS/Version: Windows XP
            Status: NEW
          Severity: Normal
          Priority: Other
         Component: Core
        AssignedTo: bugs@httpd.apache.org
        ReportedBy: jozz@whizzideas.com


When integerating Apache with other container or script, such as Tomcat, or
PERL, the apache will sometime fail to output the content compeletely. That is
to say large amount of HTML content which is generated by the Tomcat/PERL, will
be incomplete after passing the Apache as part of the content is lost!

I met with such problem on my PC server: x86, AMD 1Ghz CPU, Windows XP, Apache
2.0.52, with Tomcat 5.0.28 and ActivePerl 5.8.

For brief, Apache 2.0.52/50/49 failed in some circumstances but Apache 1.3.33/31
succeeded.

I will just try to describe what happened on my Server. It will see boring.

For PERL-script, It happened like this:
After installing the AWStats, I opened the
http://www.whizzideas.com/awstats/awstats.pl
with IE6, the IE kept refreshing the right frame (The main content frame that is
generated by AWStats[PERL-script]. It took 3-5 seconds for my computer to
generatet the about 80K page) for some seconds (It seemed that the content was
not completely loaded and IE tried to loaded it again and again or such things,
I guess) and finally shown me a "404 page not found" message.

When I opened the same link with Mozilla 1.7.3, things went better. The mozilla
loaded the whole page without refreshing. But when I visited the same link
remotely using Mozilla 1.6, I only part of the page. I viewed the page source
and found that the page source was incomplete.

And later I changed to Apache 2.0.50 and Apache 2.0.49. IE6 (on my local
machine)  still could not load the AWStats summary page completely (Actually
sometimes after IE6 refreshing for some seconds it did get a complete page. But
it happened scarely.).

And finally I change to Apache 1.3.31 and Apache 1.3.33, no such problem
happened again.

I reported this problem as a bug of AWStats in :
https://sourceforge.net/tracker/?func=detail&atid=113764&aid=1061060&group_id=13764
But finally I realized this was a bug of Apache 2.0.*

For Tomcat, similar problems existed. Example: I visited
http://www.whizzideas.com/whizznotes/xhtml/441.html
Using Mozilla 1.6 remotely, I can see imcomplete HTML codes!
This page was about 27K. When visiting some some small-size page (4K-8K), the
page could be loaded completely. Ridiculous.

But for the most important things, if I visited the page of Tomcat directly, e.g:
http://www.whizzideas.com:8080/whizznotes/xhtml/441.html
No matter locally or remotely and no matter Mozilla or IE, complete pages were
always loaded.

(I used Mod_JK2.so in connecting the Tomcat and Apache 2.0.52, but when finally
I used Apache 1.3.33, I used Mod_jk_1.2.6_1.3.31.dll instead. But in AWStats, no
connector was used, and Apache 2.0.52/50/49 all failed to pass complete page,
and Apache 1.3.33 succeed to pass complete page. So I guess the factor of
connector can be ignored.)

It seemed that 30K+ was a little large for the Apache to flush the whole content
from Tomcat to client in one time, so Apache read twice or three times or more
times from Tomcat and flushed to client application each time. But as some
problems(or bugs) hidden here, the content was incorrectly dealt by Apache when
passing and client application discard the latter incorrect part of the content
(IE tried to refresh to get a new page but failed again and again and finally IE
gave up and displayed 404).

The keys of repeat such bug on Apache 2.0.52:
1. The pages should be generated by other container such as Tomcat or other scripts;
2. The server is somewhat slow: the Tomcat / PERL-script can not generated the
page quickly;
3. Visiting the pages remotely using IE (Maybe IE is slower than Mozilla in
loading pages);

For my experience:
Using Resin 3.0.6 with Apache 5.0.28 in generating HTML contents(less than 30K),
I always get imcomplete page using Mozilla 1.6 remotely (But things goes right
using IE locally)
Using Tomcat 5.0.28 with Apache 5.0.28 with less than 30K HTML pages, I always
get complete page using Mozilla 1.6 remotely (But sometimes mozilla did give me
imcomplete page. This was scare, and after refreshing it gave em complete page)
.But with page greater than 70K, mozilla will fail to load complete page remotely.
Using proxy in Browser will make the bug obvious even on those small-size pages.

Sugguestion for repeatting the bug: Using Resin 3.0.6 and integerate it with
Apache 2.0.52, and write a servlet to generated HTML (greater than 100K)  with
other *.js or *.css or images included page. And using Mozillia 1.6 to visit
such page through rough connection (such as 512K bandwith connection or using
low bandwidth proxy);

(Attention: the related links given above are already under Apache 1.3.33, so no
such bug as described exist)

I hope I am stating things clearily and making comments not enough boring. Thanks.

---------------------------------------------------------------------
To unsubscribe, e-mail: bugs-unsubscribe@httpd.apache.org
For additional commands, e-mail: bugs-help@httpd.apache.org


Mime
View raw message