On Mon, Jun 29, 2009 at 4:45 PM, Graham
Dumpleton<[email protected]> wrote:
>
> 2009/6/30 mmcgrath <[email protected]>:
>>
>> We're using mod_wsgi to deploy a few apps, mostly TurboGears.  We are
>> getting some mysterious 500's from time to time.  The show up in our
>> access log, but not in the error logs.  The 500 itself seems to be
>> coming from apache and not cherrypy.  Also we cannot reproduce the
>> error manually with any success.  It just randomly happens and we
>> aren't sure what is going on.  Here is our config:
>>
>> WSGISocketPrefix run/wsgi
>>
>> # TG implements its own signal handler.
>> WSGIRestrictSignal Off
>>
>> # These are the real tunables
>> WSGIDaemonProcess fas processes=8 threads=2 maximum-requests=50000
>> user=fas group=fas display-name=fas inactivity-timeout=300 shutdown-
>> timeout=10
>> WSGIPythonOptimize 1
>>
>> WSGIScriptAlias /accounts /usr/lib/python2.4/site-packages/fas/
>> fas.wsgi/accounts
>
> You also have:
>
>  WSGIProcessGroup fas
>
> don't you? If not, you aren't using daemon mode as you think you are.
>

Sorry, part of my paste got cut off, this is at the end of that file:

<Directory /usr/lib/python2.4/site-packages/fas/>
    WSGIProcessGroup fas
    Order deny,allow
    Allow from all
</Directory>

>> Does anything jump out as obviously wrong there?
>
> What version of mod_wsgi are you using?
>

mod_wsgi-2.1-4.el5 - OS is RHEL5

> Have you set:
>
>  LogLevel info
>
> to enable greater level of mod_wsgi messages in error logs?
>

We had it up to debug for a while, we'll set it back to info again and
include this information.

> Do the 500 error correspond within when that additional logging says
> that daemon processes are being restarted?
>

Not that I could see.

> Do you handle any requests that can take more than 10 seconds?
>

Not that I've been able to see.

> Is there any sort of error in error logs at about time 500 error
> occurs, even ones not from mod_wsgi or Python?
>

Interestingly we didn't even realize where the 500's were coming from
for a very long time because we never saw them in the error log.  It
wasn't until we checked the access log that we saw them.  At least at
a normal loglevel.  I'll send more information when we catch one with
loglevel info.

      -Mike

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"modwsgi" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/modwsgi?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to