Haha. The crash was because of a separate problem in mod_wsgi 4.0.
Fixed that now and get same problem as 3.3.

Graham

On 10 January 2012 16:02, Graham Dumpleton <[email protected]> wrote:
> Interestingly I can totally reproduce this. Only occurring when I
> force preloading of WSGI script by specifying process-group and
> application-group to WSGIScriptAlias.
>
> That is with mod_wsgi 3.3. Under mod_wsgi 4.0 it will actual Apache
> child worker processes.
>
> Graham
>
> On 10 January 2012 15:33, Graham Dumpleton <[email protected]> wrote:
>> FWIW, a while back after 2.2.21 had come out I was doing some testing
>> with it I did see some strange Python thread errors at one point. I
>> didn't really go back and investigate and went back to using bundled
>> Apple Apache, promptly forgetting about it until now.
>>
>> I am again trying to use 2.2.21 with some stuff I am doing and getting
>> thread errors again, albeit completely different ones to before.
>>
>> [Tue Jan 10 15:24:14 2012] [info] mod_wsgi (pid=44499): Attach interpreter 
>> ''.
>> [Tue Jan 10 15:24:14 2012] [info] mod_wsgi (pid=44499, process='',
>> application=''): Loading WSGI script '...'.
>> pthread_cond_signal: Undefined error: 0
>> [Tue Jan 10 15:24:44 2012] [info] mod_wsgi (pid=44507): Initializing Python.
>> [Tue Jan 10 15:24:44 2012] [info] mod_wsgi (pid=44507): Attach interpreter 
>> ''.
>> [Tue Jan 10 15:24:44 2012] [info] mod_wsgi (pid=44507, process='',
>> application=''): Loading WSGI script '...'.
>> pthread_cond_wait: Invalid argument
>> pthread_cond_wait: Invalid argument
>> pthread_cond_wait: Invalid argument
>> pthread_cond_wait: Invalid argument
>>
>> The processes then proceeded to max out the CPU in some sort of tight loop.
>>
>> So maybe there is some underlying issue here related specifically to
>> Apache 2.2.21, or perhaps Python 2.7 as also using that when normally
>> would use Python 2.6.
>>
>> Graham
>>
>> On 8 January 2012 11:01, Christian Boos <[email protected]> wrote:
>>> On 1/7/2012 11:42 PM, Graham Dumpleton wrote:
>>>>
>>>> Thanks for tracking that down. I missed it in my mail until this
>>>> morning as had gone and marked the thread unread to remind me to go
>>>> back and look.
>>>>
>>>> I am starting to think about how this could even occur. Can you
>>>> perhaps give me a list of all the Apache modules which are loaded? I
>>>> don't know easy way of doing this on Windows. On Unix you would run:
>>>>
>>>>   /usr/sbin/httpd -M
>>>
>>>
>>> Also works on Windows:
>>>
>>> C:\Dev\Apache2.2.21>bin\httpd.exe -M
>>> Loaded Modules:
>>>  core_module (static)
>>>  win32_module (static)
>>>  mpm_winnt_module (static)
>>>  http_module (static)
>>>  so_module (static)
>>>  actions_module (shared)
>>>  alias_module (shared)
>>>  asis_module (shared)
>>>  auth_basic_module (shared)
>>>  auth_digest_module (shared)
>>>  authn_default_module (shared)
>>>  authn_file_module (shared)
>>>  authz_default_module (shared)
>>>  authz_groupfile_module (shared)
>>>  authz_host_module (shared)
>>>  authz_user_module (shared)
>>>  autoindex_module (shared)
>>>  cgi_module (shared)
>>>  dir_module (shared)
>>>  env_module (shared)
>>>  include_module (shared)
>>>  isapi_module (shared)
>>>  log_config_module (shared)
>>>  mime_module (shared)
>>>  negotiation_module (shared)
>>>  setenvif_module (shared)
>>>  wsgi_module (shared)
>>>
>>> I can probably reduce that list, ... right:
>>>
>>> Loaded Modules:
>>>  core_module (static)
>>>  win32_module (static)
>>>  mpm_winnt_module (static)
>>>  http_module (static)
>>>  so_module (static)
>>>  setenvif_module (shared)
>>>  wsgi_module (shared)
>>>
>>> And the problem persists.
>>>
>>>
>>>
>>>>
>>>> For a problem to arise at that point, would have to somehow involve
>>>> the output filter chain, be that Apache default output filters or
>>>> optional ones added by other Apache modules. That or I have screwed up
>>>> output bucket brigade management and not following API usage properly.
>>>>
>>>
>>> I've created http://code.google.com/p/modwsgi/issues/detail?id=255 btw,
>>> and there I noted that the problem could be traced back to b8f578d32018
>>> so in the 2.0c3 days! Quite strange.
>>>
>>>
>>> -- Christian
>>>
>>> --
>>> 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.
>>>

-- 
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