On Wed, Mar 20, 2002 at 08:46:13AM -0500, akuchlin@mems-exchange.org wrote:
> On Wed, Mar 20, 2002 at 05:00:09AM -0500, Neal M. Holtz wrote:
> >The s.getpeername() call raises an exception, which causes
> >_isFCGI to be set to 0.  Why does this happen?
>
> This comes from the FastCGI specification, and is how a script tells
> whether it's being run under FastCGI or not.  See section 2.2
> http://www.fastcgi.com/devkit/doc/fcgi-spec.html .

Sorry, I guess I really meant why does s.getpeername() raise the exception,
because mod_fastcgi seems to be doing many of the right things?

>
> >The (err,errmsg) is 88, Socket operation on non-socket.
>
> Are you on Windows?  Perhaps the Windows socket API returns different
> errors, and the test should be changed to assume FastCGI if the error
> is 88.

No, its Linux (Caldera 2.4, but with kernel 2.4.9 and most other things
updated: python 2.1, apache 1.3.20, mod_fastcgi 2.2.12, gcc 2.95.3).

Note that I have not supplied any config options for mod_fastcgi, other than

  AddHandler fastcgi-script fcgi

So all the default config options apply.

Using 'ps axf' I can see the 'fcgi-' process under httpd, and for a
few seconds I can see a python process spawned by that, but of course it
exits afer handling one request.  So mod_fastcgi is spawning a
dynamic application, but ... ???

--
Neal Holtz                             http://www.docuweb.ca/~nholtz
Dept. of Civil and Environmental Engineering,   Carleton University,
Ottawa, Ontario, Canada K1S 5B6.                   nholtz@docuweb.ca