[hunchentoot-devel] Fwd: Intermittent problem with codequarterly website (chunked coding glitch?)

Fri Jun 10 03:09:28 UTC 2011

I have no particular reason to believe Hunchentoot is at fault here
but I figured I'd ask in case anyone had seen anything like this. The
codequarterly.com website is being served by Apache proxying to
Hunchentoot (via ProxyPass and ProxyPassReverse directives in the
Apache config) and this user reports occasionally getting back some
hinky chunked-coded responses.

Any ideas of the top of anyone's head?

-Peter

---------- Forwarded message ----------
From: Thomas (AWS), Chris <ccthomas at amazon.com>
Date: Thu, Jun 9, 2011 at 6:14 PM
Subject: Intermittent problem with codequarterly website (chunked
coding glitch?)
To: "editor at codequarterly.com" <editor at codequarterly.com>

Greetings,

This may not be the appropriate place to report this issue, but I
couldn't find any technical contact information on your website.  Feel
free to forward as necessary.

Intermittently (maybe 10-20% of the time), when trying to view pages
on the codequarterly website, my browser is displaying what appears to
be the text of the raw HTML of the page, instead of the rendered page.
 When I look closer, I see that the text begins with a hexadecimal
number (e.g, '1FED'), followed by the usual HTML <DOCTYPE....  Header.

I reproduced the problem with curl  & took a packet trace (output at
end of email), to better see what is going on:

% curl -v http://www.codequarterly.com/2011/rich-hickey/

It shows that the response is coming back with chunked coding, but
there are two hexadecimal numbers preceding  the actual HTML content,
instead of just the single number that one would expect from a
chunked-coded response.  The remainder of the body shows additional
"extra" chunk length prefixes.  It is almost as if the HTTP server was
sending the response body through two passes of chunked encoding.

Anyway, you might want to look into how your website software is set
up that might be causing this, or perhaps whether there's a proxy that
is inappropriately adding its own chunked coding to the response as it
passes through.

Best Regards,
Chris Thomas

-- request --
GET /2011/rich-hickey/ HTTP/1.1
User-Agent: curl/7.18.1 (i686-pc-linux-gnu) libcurl/7.18.1
OpenSSL/0.9.8b zlib/1.2.3
Host: www.codequarterly.com
Accept: */*

-- response --
HTTP/1.1 200 OK
Date: Fri, 10 Jun 2011 00:51:12 GMT
Transfer-Encoding: chunked
Content-Type: text/plain

1f3e
1FE4<!DOCTYPE html>
<HTML lang='en'>
   <HEAD>
       <META http-equiv='content-type' content='text/html; charset=UTF-8'>
       <TITLE>Code Quarterly</TITLE>
....

-- 
Peter Seibel
http://www.codequarterly.com/