[mod_python] OS X / Apache 2 / mod_python bug

Jamie Kirkpatrick jkp at kirkconsulting.co.uk
Tue Mar 15 14:12:37 EST 2005


I just throught I'd drop a line to say that I'm not the only one who 
has had this issue.  A fellow OS X user who is actually the author of 
the python package i wanted to install (trac) has the same issue - it's 
a shame he didnt understand fully what I was seeing as he could have 
saved me days of hassle!

This should get nailed if possible.

Jamie

On 14 Mar 2005, at 11:13, Graham Dumpleton wrote:

>
> On Monday, March 14, 2005, at 08:07  PM, Jamie Kirkpatrick wrote:
>> That said, have you setup use of Apache in the main configuration 
>> file
>>
>> or in a .htaccess file? Can you post what configuration you are using
>>
>> so we can see if it is pretty normal or something strange? Post it if
>>
>> you can.
>>
>> I can post but its rather long - I am pretty darn confident that 
>> there is nothing wrong with the config tho.  I havent changed barely 
>> anything from the stock install, in-fact, I can reproduce the error 
>> by leaving everything as is from when the install is performed and 
>> only adding a python handler to the default htdocs directory.  The 
>> only thing I have consistently changed is the port the server is 
>> listening on to port 8080.  I have tested with the minimum of 
>> changes: added a handler and dropped in a small test script.
>
> The whole Apache config file isn't required, but just the small 
> section where
> your define SetHandler/AddHandler and PythonHandler. Ie., the 
> appropriate
> <Directory> construct covering your htdocs directory. This is just to 
> get the
> whole picture and clarify that it is being done within a Directory 
> directive
> as opposed to being at global scope in the config.
>
>> Do you get these 500 errors if the only mod_python application 
>> configured
>>
>> is a really basic "mptest" type test handler, or does it only happen
>>
>> when you go to set it up for this more complicated application?
>>
>>
>> I used the hello.py example from the site and consistently got the 
>> same results.  This was how I finally pinned it down to mod_python.  
>> See the last answer for further details.
>
> Sending a copy of the hello.py wouldn't hurt as well.
>
> Based on your other answers, is very strange and nothing obvious 
> stands out.
> The next thing I would try is to localise where the 500 error 
> actually gets
> raised. There are four possible places it could be occurring. These 
> are:
>
> 1. In Python code executed within the handler itself.
>
> 2. In the Python code part of mod_python. Ie., somewhere within the 
> file
> lib/python/mod_python/apache.py.
>
> 3. In the C code part of mod_python. Ie., somewhere within the file
> src/mod_python.c.
>
> 4. In Apache itself somewhere prior to, but possibly after mod_python 
> is
> invoked. Lets hope not. :-)
>
> Elimination of 1 is pretty straight forward by using a test handler 
> which
> contains:
>
>     from mod_python import apache
>     apache.log_error("start import")
>
>     def handler(req):
>         apache.log_error("start handler")
>         req.content_type = "text/plain"
>         req.send_http_header()
>         req.write("Hello World!")
>         apache.log_error("end handler")
>         return apache.OK
>
>     apache.log_error("end import")
>
> If the 500 error is not generated from within the import of the 
> handler
> module or the handler itself when called, you should see in the Apache
> error log all of the messages above.
>
> BTW, as an extra precaution, if Apache is able to write to your 
> document
> directories, you may want to manually remove any ".pyc" files that may
> have been generated as the result of a prior test, before running the 
> test.
>
> To test out 2, one would need to modify 
> lib/python/mod_python/apache.py
> and reinstall it. You should add calls to apache.log_error() into 
> apache.py
> to track where it may get to within its execution.
>
> I think the appropriate place to place log statements in that file is 
> at
> the start of the class method:
>
>   def HandlerDispatch(self, req):
>
>     log_error("start dispatch")
>
> Prior to the final return of the method, add:
>
>     log_error("end dispatch %s"%str(result))
>
> There is a return part way through the method which returns a 500 
> error,
> but it is preceded by logging of a message anyway.
>
> If it is found that 500 is being returned from the method, this would
> localise the cause of the problem to this method. At this point we 
> would
> have to start analysing further the method and the potential sources 
> of
> the 500 error by adding more logging and checking where result is 
> being
> set to 500.
>
> If this method always returns a 200 status response, yet the browser
> still sees 500, we would have to consider number 3 above.
>
> At this point things get a bit trickier as one is dealing with the C 
> code.
> It is worth noting though that most instances of where a 500 response 
> are
> generated in the mod_python.c file are preceded by message logging. 
> The
> only ones that aren't are all of the form:
>
>     /* get/create interpreter */
>     idata = get_interpreter(interp_name, req->server);
>
>     if (!idata)
>         return HTTP_INTERNAL_SERVER_ERROR;
>
> Not being able to get/create the interpreter is pretty severe and if 
> that
> was happening then no Python code at all would be getting executed. I 
> am
> not sure why this situation couldn't be logged as well. Thus instead 
> of
> that above, have:
>
>     if (!idata) {
>         ap_log_rerror(APLOG_MARK, APLOG_NOERRNO|APLOG_ERR, 0, req,
>                       "python_handler: Can't get/create 
> interpreter.");
>
>         return HTTP_INTERNAL_SERVER_ERROR;
>     }
>
> Anyway, that is the steps I would personally pursue to try and narrow 
> down
> where the 500 error is being generated from. Once one can reliably 
> pinpoint
> it down to a specific area of code in this way, have a better chance 
> of
> determining why it is happening.
>
> Graham
>
>
>
>



More information about the Mod_python mailing list