>Number: 2817 >Category: general >Synopsis: Some URLs w/o trailing slashes work, some don't >Confidential: no >Severity: critical >Priority: medium >Responsible: apache >State: open >Class: sw-bug >Submitter-Id: apache >Arrival-Date: Sun Aug 9 21:50:01 PDT 1998 >Last-Modified: >Originator: [EMAIL PROTECTED] >Organization: apache >Release: 1.2.6, 1.3.0 >Environment: Linux 1.2.8+, gcc 2.6.3 >Description: I've been running a production server on Linux with NCSA httpd for about three years now. I have upgraded the NCSA server with no problems. (I was a Beta tester and ported to several platforms.)
Now I'm trying Apache (I've been using it everywhere else for some time, so I'm familiar with it - I maintain lots of servers on lots of systems). One problem is giving me fits, and keeping me from switching to Apache on this one system. We have over 3,000 HTML pages [1] on this system, and run two virtual servers. There are pages in both DocumentRoots, and about a dozen users have their own public_html directories. There are probably 100 or so scripts. Everything works beautifully when I switch to the Apache server EXCEPT... One user's main index page contains several links to subdirectories of the form: <A HREF="/~eluser/hobbit">...</A> Which correctly redirect to /~eluser/hobbit/ , and this directory has an index.html file . Of the five links like this, three work fine (returning Content-type: text/html) . The other two, however, pop up a dialog box (on Navigator 3.x and 4.x and lynx), because they are returning Content-type: text/x-server-parsed-html . (I determined this by running telnet on the server port, of course). I have configured the server to parse *all* .html files (this site predates the .shtml standard, and almost every page on it uses SSI), but this has never been a problem with any version of NCSA from 1.2 through 1.5.x . This happens with Apache 1.2.6 and 1.3 . I started with fresh conf files, and tweaked them minimally for my site. I have no non-standard modules. I have tested almost the entire site at this point, and only these two links are problematical. I created a similar directory structure in my own public_html/ and it works *fine* there. I later cloned the offending tree and saw the same behavior, even when I reduced the index files slowly down to nothing. Three things that may be either clues or red herrings: - The two directories giving me fits have names 2 and 6 characters long, the three which work fine have names 3 and 5 characters long. - I modified the Apache 1.3 source tree by changing te only occurance of text/x-server-parsed-html to text/html and it made no difference. I have verified that there is no such string as text/x-server-parsed-html in the binary! - Only one of the offending index.html files actually uses SSI. The offending directories are drwxr-sr-x, but I have explicitly set XBitHack to off, just to be safe. There is no discernable difference between these directories or their index files and any others. >How-To-Repeat: http://www.netads.com:81/~psyber/ The "White Wabbit" and "Pat Hester - Cyberfreak" links will fail, the others will work fine. >Fix: I wish! >Audit-Trail: >Unformatted: [In order for any reply to be added to the PR database, ] [you need to include <[EMAIL PROTECTED]> in the Cc line ] [and leave the subject line UNCHANGED. This is not done] [automatically because of the potential for mail loops. ] [If you do not include this Cc, your reply may be ig- ] [nored unless you are responding to an explicit request ] [from a developer. ] [Reply only with text; DO NOT SEND ATTACHMENTS! ]
